diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1.log
new file mode 100644
index 0000000000000000000000000000000000000000..7d25dda225276bc0195a5c5553d5f7cc3982ec41
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1.log
@@ -0,0 +1,57 @@
+Found metadata.json file, assuming customized QQTT dataset! [28/10 18:23:28]
+Init points from pcd... [28/10 18:23:28]
+Init points from mesh... [28/10 18:23:28]
+Loading Training Cameras [28/10 18:23:28]
+Loading Test Cameras [28/10 18:23:28]
+Number of points at initialisation :  130019 [28/10 18:23:28]
+Sampling 100000 points from mesh [28/10 18:23:28]
+{
+  "cameras": [
+    {
+      "id": 0,
+      "width": 1920,
+      "height": 1080,
+      "object_px": 11837,
+      "occlusion_px": 160413,
+      "valid_depth_px": 1953381,
+      "boundary_px": 3192,
+      "projection_max_error_px": 2.076818088326489e-05
+    },
+    {
+      "id": 1,
+      "width": 1920,
+      "height": 1080,
+      "object_px": 9880,
+      "occlusion_px": 147454,
+      "valid_depth_px": 2026613,
+      "boundary_px": 3005,
+      "projection_max_error_px": 6.672083653499315e-07
+    },
+    {
+      "id": 2,
+      "width": 1920,
+      "height": 1080,
+      "object_px": 10239,
+      "occlusion_px": 232647,
+      "valid_depth_px": 1974051,
+      "boundary_px": 3285,
+      "projection_max_error_px": 4.7477295301253664e-06
+    },
+    {
+      "id": 3,
+      "width": 1920,
+      "height": 1080,
+      "object_px": 7381,
+      "occlusion_px": 188884,
+      "valid_depth_px": 1905718,
+      "boundary_px": 2616,
+      "projection_max_error_px": 2.9697677452058956e-06
+    }
+  ],
+  "hybrid": {
+    "vertex_color_branch": true,
+    "mesh": 103680,
+    "observed": 26339,
+    "total": 130019
+  }
+} [28/10 18:23:28]
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..e3712a230bc683808c79d520e0d96e17f556909e
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_argv.json
@@ -0,0 +1,18 @@
+{
+  "argv": [
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_gs_train_env/.venv/bin/python",
+    "scripts/real_rgbd_appearance_train.py",
+    "check",
+    "--prior",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "--accepted-prior-manifest-sha256",
+    "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/appearance_check_v1"
+  ],
+  "cwd": "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027",
+  "environment": {
+    "PYTHONDONTWRITEBYTECODE": "1",
+    "MPLBACKEND": "Agg"
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..180bd4267b394d8715a29a4eb09f9bc0d214ec23
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_check_v1_result.json
@@ -0,0 +1,4 @@
+{
+  "returncode": 0,
+  "elapsed_s": 52.67811157100368
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1.log
new file mode 100644
index 0000000000000000000000000000000000000000..68a011e285d4880a78ef14b55097a5d835b209a7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1.log
@@ -0,0 +1,13 @@
+Optimizing /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/appearance_train_v1/model
+Output folder: /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/appearance_train_v1/model [28/10 18:24:49]
+Found metadata.json file, assuming customized QQTT dataset! [28/10 18:24:49]
+Init points from pcd... [28/10 18:24:50]
+Init points from mesh... [28/10 18:24:50]
+Loading Training Cameras [28/10 18:24:50]
+Loading Test Cameras [28/10 18:24:50]
+Number of points at initialisation :  130019 [28/10 18:24:50]
+Sampling 100000 points from mesh [28/10 18:24:50]
+
Training progress:   0%|          | 0/10000 [00:00<?, ?it/s]/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_gs_train_env/.venv/lib/python3.10/site-packages/torch/utils/cpp_extension.py:1965: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
Training progress:   0%|          | 0/10000 [00:00<?, ?it/s, Loss=0.00479, L1 Loss=0.00069, Depth Loss=0.00392, Normal Loss=0.00000, Seg Loss=0.00299, Anisotropic Loss=0.00000]
Training progress:   0%|          | 10/10000 [00:00<08:49, 18.87it/s, Loss=0.00479, L1 Loss=0.00069, Depth Loss=0.00392, Normal Loss=0.00000, Seg Loss=0.00299, Anisotropic Loss=0.00000]
Training progress:   0%|          | 10/10000 [00:00<08:49, 18.87it/s, Loss=0.00368, L1 Loss=0.00047, Depth Loss=0.00343, Normal Loss=0.00000, Seg Loss=0.00227, Anisotropic Loss=0.00000]
Training progress:   0%|          | 20/10000 [00:00<04:57, 33.60it/s, Loss=0.00368, L1 Loss=0.00047, Depth Loss=0.00343, Normal Loss=0.00000, Seg Loss=0.00227, Anisotropic Loss=0.00000]
Training progress:   0%|          | 20/10000 [00:00<04:57, 33.60it/s, Loss=0.00282, L1 Loss=0.00035, Depth Loss=0.00157, Normal Loss=0.00000, Seg Loss=0.00135, Anisotropic Loss=0.00000]
Training progress:   0%|          | 30/10000 [00:00<03:41, 44.97it/s, Loss=0.00282, L1 Loss=0.00035, Depth Loss=0.00157, Normal Loss=0.00000, Seg Loss=0.00135, Anisotropic Loss=0.00000]
Training progress:   0%|          | 30/10000 [00:00<03:41, 44.97it/s, Loss=0.00270, L1 Loss=0.00035, Depth Loss=0.00231, Normal Loss=0.00000, Seg Loss=0.00145, Anisotropic Loss=0.00000]
Training progress:   0%|          | 40/10000 [00:00<03:06, 53.47it/s, Loss=0.00270, L1 Loss=0.00035, Depth Loss=0.00231, Normal Loss=0.00000, Seg Loss=0.00145, Anisotropic Loss=0.00000]
Training progress:   0%|          | 40/10000 [00:01<03:06, 53.47it/s, Loss=0.00247, L1 Loss=0.00041, Depth Loss=0.00222, Normal Loss=0.00000, Seg Loss=0.00145, Anisotropic Loss=0.00000]
Training progress:   0%|          | 50/10000 [00:01<02:46, 59.69it/s, Loss=0.00247, L1 Loss=0.00041, Depth Loss=0.00222, Normal Loss=0.00000, Seg Loss=0.00145, Anisotropic Loss=0.00000]
Training progress:   0%|          | 50/10000 [00:01<02:46, 59.69it/s, Loss=0.00228, L1 Loss=0.00054, Depth Loss=0.00159, Normal Loss=0.00000, Seg Loss=0.00142, Anisotropic Loss=0.00000]
Training progress:   1%|          | 60/10000 [00:01<02:34, 64.20it/s, Loss=0.00228, L1 Loss=0.00054, Depth Loss=0.00159, Normal Loss=0.00000, Seg Loss=0.00142, Anisotropic Loss=0.00000]
Training progress:   1%|          | 60/10000 [00:01<02:34, 64.20it/s, Loss=0.00199, L1 Loss=0.00050, Depth Loss=0.00136, Normal Loss=0.00000, Seg Loss=0.00116, Anisotropic Loss=0.00000]
Training progress:   1%|          | 70/10000 [00:01<02:26, 67.76it/s, Loss=0.00199, L1 Loss=0.00050, Depth Loss=0.00136, Normal Loss=0.00000, Seg Loss=0.00116, Anisotropic Loss=0.00000]
Training progress:   1%|          | 70/10000 [00:01<02:26, 67.76it/s, Loss=0.00160, L1 Loss=0.00025, Depth Loss=0.00107, Normal Loss=0.00000, Seg Loss=0.00077, Anisotropic Loss=0.00000]
Training progress:   1%|          | 80/10000 [00:01<02:21, 69.92it/s, Loss=0.00160, L1 Loss=0.00025, Depth Loss=0.00107, Normal Loss=0.00000, Seg Loss=0.00077, Anisotropic Loss=0.00000]
Training progress:   1%|          | 80/10000 [00:01<02:21, 69.92it/s, Loss=0.00155, L1 Loss=0.00024, Depth Loss=0.00102, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   1%|          | 90/10000 [00:01<02:18, 71.38it/s, Loss=0.00155, L1 Loss=0.00024, Depth Loss=0.00102, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   1%|          | 90/10000 [00:01<02:18, 71.38it/s, Loss=0.00156, L1 Loss=0.00032, Depth Loss=0.00154, Normal Loss=0.00000, Seg Loss=0.00095, Anisotropic Loss=0.00000]
Training progress:   1%|          | 100/10000 [00:01<02:16, 72.55it/s, Loss=0.00156, L1 Loss=0.00032, Depth Loss=0.00154, Normal Loss=0.00000, Seg Loss=0.00095, Anisotropic Loss=0.00000]
Training progress:   1%|          | 100/10000 [00:01<02:16, 72.55it/s, Loss=0.00149, L1 Loss=0.00024, Depth Loss=0.00113, Normal Loss=0.00000, Seg Loss=0.00070, Anisotropic Loss=0.00000]
Training progress:   1%|          | 110/10000 [00:01<02:14, 73.60it/s, Loss=0.00149, L1 Loss=0.00024, Depth Loss=0.00113, Normal Loss=0.00000, Seg Loss=0.00070, Anisotropic Loss=0.00000]
Training progress:   1%|          | 110/10000 [00:01<02:14, 73.60it/s, Loss=0.00134, L1 Loss=0.00024, Depth Loss=0.00117, Normal Loss=0.00000, Seg Loss=0.00070, Anisotropic Loss=0.00000]
Training progress:   1%|          | 120/10000 [00:01<02:13, 74.02it/s, Loss=0.00134, L1 Loss=0.00024, Depth Loss=0.00117, Normal Loss=0.00000, Seg Loss=0.00070, Anisotropic Loss=0.00000]
Training progress:   1%|          | 120/10000 [00:02<02:13, 74.02it/s, Loss=0.00150, L1 Loss=0.00044, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00086, Anisotropic Loss=0.00000]
Training progress:   1%|▏         | 130/10000 [00:02<02:12, 74.52it/s, Loss=0.00150, L1 Loss=0.00044, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00086, Anisotropic Loss=0.00000]
Training progress:   1%|▏         | 130/10000 [00:02<02:12, 74.52it/s, Loss=0.00124, L1 Loss=0.00020, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   1%|▏         | 140/10000 [00:02<02:11, 74.74it/s, Loss=0.00124, L1 Loss=0.00020, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   1%|▏         | 140/10000 [00:02<02:11, 74.74it/s, Loss=0.00143, L1 Loss=0.00043, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00082, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 150/10000 [00:02<02:11, 75.06it/s, Loss=0.00143, L1 Loss=0.00043, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00082, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 150/10000 [00:02<02:11, 75.06it/s, Loss=0.00135, L1 Loss=0.00044, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00083, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 160/10000 [00:02<02:10, 75.13it/s, Loss=0.00135, L1 Loss=0.00044, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00083, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 160/10000 [00:02<02:10, 75.13it/s, Loss=0.00113, L1 Loss=0.00025, Depth Loss=0.00108, Normal Loss=0.00000, Seg Loss=0.00061, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 170/10000 [00:02<02:10, 75.36it/s, Loss=0.00113, L1 Loss=0.00025, Depth Loss=0.00108, Normal Loss=0.00000, Seg Loss=0.00061, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 170/10000 [00:02<02:10, 75.36it/s, Loss=0.00129, L1 Loss=0.00042, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00078, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 180/10000 [00:02<02:10, 75.48it/s, Loss=0.00129, L1 Loss=0.00042, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00078, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 180/10000 [00:02<02:10, 75.48it/s, Loss=0.00122, L1 Loss=0.00040, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 190/10000 [00:02<02:09, 75.58it/s, Loss=0.00122, L1 Loss=0.00040, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 190/10000 [00:03<02:09, 75.58it/s, Loss=0.00109, L1 Loss=0.00019, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 200/10000 [00:03<02:09, 75.51it/s, Loss=0.00109, L1 Loss=0.00019, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 200/10000 [00:03<02:09, 75.51it/s, Loss=0.00096, L1 Loss=0.00020, Depth Loss=0.00096, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 210/10000 [00:03<02:09, 75.37it/s, Loss=0.00096, L1 Loss=0.00020, Depth Loss=0.00096, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 210/10000 [00:03<02:09, 75.37it/s, Loss=0.00117, L1 Loss=0.00040, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 220/10000 [00:03<02:09, 75.52it/s, Loss=0.00117, L1 Loss=0.00040, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00072, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 220/10000 [00:03<02:09, 75.52it/s, Loss=0.00113, L1 Loss=0.00023, Depth Loss=0.00099, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 230/10000 [00:03<02:09, 75.41it/s, Loss=0.00113, L1 Loss=0.00023, Depth Loss=0.00099, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 230/10000 [00:03<02:09, 75.41it/s, Loss=0.00095, L1 Loss=0.00018, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 240/10000 [00:03<02:09, 75.31it/s, Loss=0.00095, L1 Loss=0.00018, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   2%|▏         | 240/10000 [00:03<02:09, 75.31it/s, Loss=0.00093, L1 Loss=0.00022, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   2%|▎         | 250/10000 [00:03<02:10, 74.98it/s, Loss=0.00093, L1 Loss=0.00022, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   2%|▎         | 250/10000 [00:03<02:10, 74.98it/s, Loss=0.00110, L1 Loss=0.00021, Depth Loss=0.00104, Normal Loss=0.00000, Seg Loss=0.00056, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 260/10000 [00:03<02:09, 75.40it/s, Loss=0.00110, L1 Loss=0.00021, Depth Loss=0.00104, Normal Loss=0.00000, Seg Loss=0.00056, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 260/10000 [00:03<02:09, 75.40it/s, Loss=0.00115, L1 Loss=0.00038, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00065, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 270/10000 [00:03<02:08, 75.53it/s, Loss=0.00115, L1 Loss=0.00038, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00065, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 270/10000 [00:04<02:08, 75.53it/s, Loss=0.00107, L1 Loss=0.00024, Depth Loss=0.00099, Normal Loss=0.00000, Seg Loss=0.00055, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 280/10000 [00:04<02:08, 75.56it/s, Loss=0.00107, L1 Loss=0.00024, Depth Loss=0.00099, Normal Loss=0.00000, Seg Loss=0.00055, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 280/10000 [00:04<02:08, 75.56it/s, Loss=0.00095, L1 Loss=0.00022, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 290/10000 [00:04<02:08, 75.43it/s, Loss=0.00095, L1 Loss=0.00022, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 290/10000 [00:04<02:08, 75.43it/s, Loss=0.00102, L1 Loss=0.00024, Depth Loss=0.00100, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 300/10000 [00:04<02:08, 75.70it/s, Loss=0.00102, L1 Loss=0.00024, Depth Loss=0.00100, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 300/10000 [00:04<02:08, 75.70it/s, Loss=0.00090, L1 Loss=0.00019, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 310/10000 [00:04<02:08, 75.50it/s, Loss=0.00090, L1 Loss=0.00019, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 310/10000 [00:04<02:08, 75.50it/s, Loss=0.00107, L1 Loss=0.00038, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00068, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 320/10000 [00:04<02:08, 75.37it/s, Loss=0.00107, L1 Loss=0.00038, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00068, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 320/10000 [00:04<02:08, 75.37it/s, Loss=0.00105, L1 Loss=0.00021, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 330/10000 [00:04<02:07, 75.55it/s, Loss=0.00105, L1 Loss=0.00021, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 330/10000 [00:04<02:07, 75.55it/s, Loss=0.00099, L1 Loss=0.00022, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 340/10000 [00:04<02:07, 75.55it/s, Loss=0.00099, L1 Loss=0.00022, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   3%|▎         | 340/10000 [00:05<02:07, 75.55it/s, Loss=0.00104, L1 Loss=0.00035, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00060, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 350/10000 [00:05<02:07, 75.72it/s, Loss=0.00104, L1 Loss=0.00035, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00060, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 350/10000 [00:05<02:07, 75.72it/s, Loss=0.00094, L1 Loss=0.00019, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 360/10000 [00:05<02:07, 75.42it/s, Loss=0.00094, L1 Loss=0.00019, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 360/10000 [00:05<02:07, 75.42it/s, Loss=0.00084, L1 Loss=0.00020, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 370/10000 [00:05<02:07, 75.51it/s, Loss=0.00084, L1 Loss=0.00020, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   4%|▎         | 370/10000 [00:05<02:07, 75.51it/s, Loss=0.00083, L1 Loss=0.00017, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 380/10000 [00:05<02:07, 75.50it/s, Loss=0.00083, L1 Loss=0.00017, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 380/10000 [00:05<02:07, 75.50it/s, Loss=0.00099, L1 Loss=0.00017, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 390/10000 [00:05<02:06, 75.90it/s, Loss=0.00099, L1 Loss=0.00017, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 390/10000 [00:05<02:06, 75.90it/s, Loss=0.00091, L1 Loss=0.00022, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 400/10000 [00:05<02:06, 75.63it/s, Loss=0.00091, L1 Loss=0.00022, Depth Loss=0.00095, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 400/10000 [00:05<02:06, 75.63it/s, Loss=0.00084, L1 Loss=0.00018, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 410/10000 [00:05<02:06, 75.66it/s, Loss=0.00084, L1 Loss=0.00018, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 410/10000 [00:05<02:06, 75.66it/s, Loss=0.00087, L1 Loss=0.00022, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 420/10000 [00:05<02:06, 75.81it/s, Loss=0.00087, L1 Loss=0.00022, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00047, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 420/10000 [00:06<02:06, 75.81it/s, Loss=0.00094, L1 Loss=0.00020, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 430/10000 [00:06<02:06, 75.84it/s, Loss=0.00094, L1 Loss=0.00020, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 430/10000 [00:06<02:06, 75.84it/s, Loss=0.00090, L1 Loss=0.00017, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 440/10000 [00:06<02:05, 75.95it/s, Loss=0.00090, L1 Loss=0.00017, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 440/10000 [00:06<02:05, 75.95it/s, Loss=0.00083, L1 Loss=0.00017, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 450/10000 [00:06<02:05, 75.92it/s, Loss=0.00083, L1 Loss=0.00017, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   4%|▍         | 450/10000 [00:06<02:05, 75.92it/s, Loss=0.00087, L1 Loss=0.00017, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 460/10000 [00:06<02:05, 75.97it/s, Loss=0.00087, L1 Loss=0.00017, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 460/10000 [00:06<02:05, 75.97it/s, Loss=0.00091, L1 Loss=0.00020, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 470/10000 [00:06<02:05, 76.02it/s, Loss=0.00091, L1 Loss=0.00020, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 470/10000 [00:06<02:05, 76.02it/s, Loss=0.00091, L1 Loss=0.00017, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 480/10000 [00:06<02:05, 76.00it/s, Loss=0.00091, L1 Loss=0.00017, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 480/10000 [00:06<02:05, 76.00it/s, Loss=0.00075, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 490/10000 [00:06<02:05, 76.02it/s, Loss=0.00075, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   5%|▍         | 490/10000 [00:07<02:05, 76.02it/s, Loss=0.00093, L1 Loss=0.00034, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00059, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 500/10000 [00:07<02:04, 76.24it/s, Loss=0.00093, L1 Loss=0.00034, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00059, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 500/10000 [00:07<02:04, 76.24it/s, Loss=0.00096, L1 Loss=0.00032, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 510/10000 [00:07<02:04, 76.39it/s, Loss=0.00096, L1 Loss=0.00032, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 510/10000 [00:07<02:04, 76.39it/s, Loss=0.00096, L1 Loss=0.00034, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00059, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 520/10000 [00:07<02:04, 75.96it/s, Loss=0.00096, L1 Loss=0.00034, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00059, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 520/10000 [00:07<02:04, 75.96it/s, Loss=0.00097, L1 Loss=0.00032, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 530/10000 [00:07<02:04, 75.98it/s, Loss=0.00097, L1 Loss=0.00032, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 530/10000 [00:07<02:04, 75.98it/s, Loss=0.00089, L1 Loss=0.00017, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 540/10000 [00:07<02:04, 75.91it/s, Loss=0.00089, L1 Loss=0.00017, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   5%|▌         | 540/10000 [00:07<02:04, 75.91it/s, Loss=0.00075, L1 Loss=0.00019, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 550/10000 [00:07<02:04, 75.70it/s, Loss=0.00075, L1 Loss=0.00019, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 550/10000 [00:07<02:04, 75.70it/s, Loss=0.00082, L1 Loss=0.00021, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 560/10000 [00:07<02:04, 76.00it/s, Loss=0.00082, L1 Loss=0.00021, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 560/10000 [00:07<02:04, 76.00it/s, Loss=0.00076, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 570/10000 [00:07<02:04, 75.96it/s, Loss=0.00076, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 570/10000 [00:08<02:04, 75.96it/s, Loss=0.00083, L1 Loss=0.00016, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 580/10000 [00:08<02:03, 76.14it/s, Loss=0.00083, L1 Loss=0.00016, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 580/10000 [00:08<02:03, 76.14it/s, Loss=0.00084, L1 Loss=0.00019, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 590/10000 [00:08<02:03, 76.26it/s, Loss=0.00084, L1 Loss=0.00019, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 590/10000 [00:08<02:03, 76.26it/s, Loss=0.00077, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 600/10000 [00:08<02:03, 76.31it/s, Loss=0.00077, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 600/10000 [00:08<02:03, 76.31it/s, Loss=0.00091, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 610/10000 [00:08<02:09, 72.25it/s, Loss=0.00091, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 610/10000 [00:08<02:09, 72.25it/s, Loss=0.00088, L1 Loss=0.00031, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 620/10000 [00:08<02:08, 73.12it/s, Loss=0.00088, L1 Loss=0.00031, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   6%|▌         | 620/10000 [00:08<02:08, 73.12it/s, Loss=0.00072, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 630/10000 [00:08<02:07, 73.69it/s, Loss=0.00072, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 630/10000 [00:08<02:07, 73.69it/s, Loss=0.00077, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 640/10000 [00:08<02:05, 74.68it/s, Loss=0.00077, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 640/10000 [00:09<02:05, 74.68it/s, Loss=0.00084, L1 Loss=0.00019, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 650/10000 [00:09<02:04, 75.28it/s, Loss=0.00084, L1 Loss=0.00019, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:   6%|▋         | 650/10000 [00:09<02:04, 75.28it/s, Loss=0.00089, L1 Loss=0.00019, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 660/10000 [00:09<02:03, 75.47it/s, Loss=0.00089, L1 Loss=0.00019, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 660/10000 [00:09<02:03, 75.47it/s, Loss=0.00084, L1 Loss=0.00017, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 670/10000 [00:09<02:03, 75.78it/s, Loss=0.00084, L1 Loss=0.00017, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 670/10000 [00:09<02:03, 75.78it/s, Loss=0.00087, L1 Loss=0.00023, Depth Loss=0.00097, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 680/10000 [00:09<02:03, 75.75it/s, Loss=0.00087, L1 Loss=0.00023, Depth Loss=0.00097, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 680/10000 [00:09<02:03, 75.75it/s, Loss=0.00089, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 690/10000 [00:09<02:03, 75.67it/s, Loss=0.00089, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 690/10000 [00:09<02:03, 75.67it/s, Loss=0.00081, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 700/10000 [00:09<02:02, 75.69it/s, Loss=0.00081, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 700/10000 [00:09<02:02, 75.69it/s, Loss=0.00078, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 710/10000 [00:09<02:06, 73.38it/s, Loss=0.00078, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 710/10000 [00:09<02:06, 73.38it/s, Loss=0.00088, L1 Loss=0.00031, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 720/10000 [00:09<02:05, 74.21it/s, Loss=0.00088, L1 Loss=0.00031, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00053, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 720/10000 [00:10<02:05, 74.21it/s, Loss=0.00085, L1 Loss=0.00029, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 730/10000 [00:10<02:03, 75.06it/s, Loss=0.00085, L1 Loss=0.00029, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00049, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 730/10000 [00:10<02:03, 75.06it/s, Loss=0.00078, L1 Loss=0.00020, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 740/10000 [00:10<02:02, 75.30it/s, Loss=0.00078, L1 Loss=0.00020, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   7%|▋         | 740/10000 [00:10<02:02, 75.30it/s, Loss=0.00070, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 750/10000 [00:10<02:02, 75.37it/s, Loss=0.00070, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 750/10000 [00:10<02:02, 75.37it/s, Loss=0.00074, L1 Loss=0.00016, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 760/10000 [00:10<02:01, 75.83it/s, Loss=0.00074, L1 Loss=0.00016, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 760/10000 [00:10<02:01, 75.83it/s, Loss=0.00083, L1 Loss=0.00016, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 770/10000 [00:10<02:01, 76.19it/s, Loss=0.00083, L1 Loss=0.00016, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 770/10000 [00:10<02:01, 76.19it/s, Loss=0.00075, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 780/10000 [00:10<02:01, 76.04it/s, Loss=0.00075, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 780/10000 [00:10<02:01, 76.04it/s, Loss=0.00075, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 790/10000 [00:10<02:00, 76.13it/s, Loss=0.00075, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 790/10000 [00:10<02:00, 76.13it/s, Loss=0.00083, L1 Loss=0.00030, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 800/10000 [00:10<02:00, 76.30it/s, Loss=0.00083, L1 Loss=0.00030, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 800/10000 [00:11<02:00, 76.30it/s, Loss=0.00088, L1 Loss=0.00030, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 810/10000 [00:11<02:04, 73.95it/s, Loss=0.00088, L1 Loss=0.00030, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 810/10000 [00:11<02:04, 73.95it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 820/10000 [00:11<02:02, 74.77it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 820/10000 [00:11<02:02, 74.77it/s, Loss=0.00074, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 830/10000 [00:11<02:01, 75.29it/s, Loss=0.00074, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 830/10000 [00:11<02:01, 75.29it/s, Loss=0.00083, L1 Loss=0.00019, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 840/10000 [00:11<02:01, 75.53it/s, Loss=0.00083, L1 Loss=0.00019, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 840/10000 [00:11<02:01, 75.53it/s, Loss=0.00072, L1 Loss=0.00016, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 850/10000 [00:11<02:01, 75.50it/s, Loss=0.00072, L1 Loss=0.00016, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:   8%|▊         | 850/10000 [00:11<02:01, 75.50it/s, Loss=0.00085, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   9%|▊         | 860/10000 [00:11<02:00, 75.98it/s, Loss=0.00085, L1 Loss=0.00020, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   9%|▊         | 860/10000 [00:11<02:00, 75.98it/s, Loss=0.00072, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   9%|▊         | 870/10000 [00:11<01:59, 76.21it/s, Loss=0.00072, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:   9%|▊         | 870/10000 [00:12<01:59, 76.21it/s, Loss=0.00081, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 880/10000 [00:12<02:00, 75.97it/s, Loss=0.00081, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 880/10000 [00:12<02:00, 75.97it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 890/10000 [00:12<01:59, 76.13it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 890/10000 [00:12<01:59, 76.13it/s, Loss=0.00087, L1 Loss=0.00031, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 900/10000 [00:12<01:59, 76.27it/s, Loss=0.00087, L1 Loss=0.00031, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00054, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 900/10000 [00:12<01:59, 76.27it/s, Loss=0.00090, L1 Loss=0.00030, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 910/10000 [00:12<02:03, 73.81it/s, Loss=0.00090, L1 Loss=0.00030, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 910/10000 [00:12<02:03, 73.81it/s, Loss=0.00085, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 920/10000 [00:12<02:02, 74.32it/s, Loss=0.00085, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00051, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 920/10000 [00:12<02:02, 74.32it/s, Loss=0.00088, L1 Loss=0.00020, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 930/10000 [00:12<02:01, 74.75it/s, Loss=0.00088, L1 Loss=0.00020, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 930/10000 [00:12<02:01, 74.75it/s, Loss=0.00073, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 940/10000 [00:12<02:00, 75.31it/s, Loss=0.00073, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:   9%|▉         | 940/10000 [00:12<02:00, 75.31it/s, Loss=0.00068, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 950/10000 [00:12<01:59, 75.60it/s, Loss=0.00068, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 950/10000 [00:13<01:59, 75.60it/s, Loss=0.00074, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 960/10000 [00:13<01:59, 75.56it/s, Loss=0.00074, L1 Loss=0.00018, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 960/10000 [00:13<01:59, 75.56it/s, Loss=0.00066, L1 Loss=0.00016, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 970/10000 [00:13<01:59, 75.70it/s, Loss=0.00066, L1 Loss=0.00016, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 970/10000 [00:13<01:59, 75.70it/s, Loss=0.00069, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 980/10000 [00:13<01:58, 75.88it/s, Loss=0.00069, L1 Loss=0.00015, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 980/10000 [00:13<01:58, 75.88it/s, Loss=0.00070, L1 Loss=0.00014, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 990/10000 [00:13<01:58, 75.98it/s, Loss=0.00070, L1 Loss=0.00014, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  10%|▉         | 990/10000 [00:13<01:58, 75.98it/s, Loss=0.00083, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1000/10000 [00:13<01:58, 76.17it/s, Loss=0.00083, L1 Loss=0.00030, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00052, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1000/10000 [00:13<01:58, 76.17it/s, Loss=0.00072, L1 Loss=0.00018, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1010/10000 [00:13<02:01, 73.79it/s, Loss=0.00072, L1 Loss=0.00018, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1010/10000 [00:13<02:01, 73.79it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1020/10000 [00:13<01:59, 74.94it/s, Loss=0.00080, L1 Loss=0.00016, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1020/10000 [00:14<01:59, 74.94it/s, Loss=0.00079, L1 Loss=0.00027, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1030/10000 [00:14<01:58, 75.55it/s, Loss=0.00079, L1 Loss=0.00027, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1030/10000 [00:14<01:58, 75.55it/s, Loss=0.00072, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1040/10000 [00:14<01:57, 75.95it/s, Loss=0.00072, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1040/10000 [00:14<01:57, 75.95it/s, Loss=0.00073, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1050/10000 [00:14<01:57, 76.30it/s, Loss=0.00073, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  10%|█         | 1050/10000 [00:14<01:57, 76.30it/s, Loss=0.00073, L1 Loss=0.00018, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1060/10000 [00:14<01:56, 76.50it/s, Loss=0.00073, L1 Loss=0.00018, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1060/10000 [00:14<01:56, 76.50it/s, Loss=0.00068, L1 Loss=0.00014, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1070/10000 [00:14<01:56, 76.46it/s, Loss=0.00068, L1 Loss=0.00014, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1070/10000 [00:14<01:56, 76.46it/s, Loss=0.00071, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1080/10000 [00:14<01:56, 76.74it/s, Loss=0.00071, L1 Loss=0.00015, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1080/10000 [00:14<01:56, 76.74it/s, Loss=0.00067, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1090/10000 [00:14<01:56, 76.78it/s, Loss=0.00067, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1090/10000 [00:14<01:56, 76.78it/s, Loss=0.00068, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1100/10000 [00:14<01:55, 76.84it/s, Loss=0.00068, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1100/10000 [00:15<01:55, 76.84it/s, Loss=0.00068, L1 Loss=0.00017, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1110/10000 [00:15<01:59, 74.35it/s, Loss=0.00068, L1 Loss=0.00017, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1110/10000 [00:15<01:59, 74.35it/s, Loss=0.00066, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1120/10000 [00:15<01:57, 75.30it/s, Loss=0.00066, L1 Loss=0.00014, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  11%|█         | 1120/10000 [00:15<01:57, 75.30it/s, Loss=0.00063, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  11%|█▏        | 1130/10000 [00:15<01:57, 75.76it/s, Loss=0.00063, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  11%|█▏        | 1130/10000 [00:15<01:57, 75.76it/s, Loss=0.00071, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:  11%|█▏        | 1140/10000 [00:15<01:56, 76.09it/s, Loss=0.00071, L1 Loss=0.00015, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:  11%|█▏        | 1140/10000 [00:15<01:56, 76.09it/s, Loss=0.00061, L1 Loss=0.00015, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1150/10000 [00:15<01:56, 76.22it/s, Loss=0.00061, L1 Loss=0.00015, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1150/10000 [00:15<01:56, 76.22it/s, Loss=0.00073, L1 Loss=0.00025, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1160/10000 [00:15<01:55, 76.35it/s, Loss=0.00073, L1 Loss=0.00025, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1160/10000 [00:15<01:55, 76.35it/s, Loss=0.00062, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1170/10000 [00:15<01:55, 76.27it/s, Loss=0.00062, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1170/10000 [00:16<01:55, 76.27it/s, Loss=0.00067, L1 Loss=0.00017, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1180/10000 [00:16<01:55, 76.49it/s, Loss=0.00067, L1 Loss=0.00017, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1180/10000 [00:16<01:55, 76.49it/s, Loss=0.00066, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1190/10000 [00:16<01:55, 76.31it/s, Loss=0.00066, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1190/10000 [00:16<01:55, 76.31it/s, Loss=0.00064, L1 Loss=0.00016, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1200/10000 [00:16<01:55, 76.38it/s, Loss=0.00064, L1 Loss=0.00016, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1200/10000 [00:16<01:55, 76.38it/s, Loss=0.00072, L1 Loss=0.00017, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1210/10000 [00:16<01:58, 74.08it/s, Loss=0.00072, L1 Loss=0.00017, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1210/10000 [00:16<01:58, 74.08it/s, Loss=0.00065, L1 Loss=0.00014, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1220/10000 [00:16<01:57, 74.54it/s, Loss=0.00065, L1 Loss=0.00014, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1220/10000 [00:16<01:57, 74.54it/s, Loss=0.00071, L1 Loss=0.00026, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1230/10000 [00:16<01:56, 75.03it/s, Loss=0.00071, L1 Loss=0.00026, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00045, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1230/10000 [00:16<01:56, 75.03it/s, Loss=0.00063, L1 Loss=0.00015, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1240/10000 [00:16<01:55, 75.65it/s, Loss=0.00063, L1 Loss=0.00015, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  12%|█▏        | 1240/10000 [00:16<01:55, 75.65it/s, Loss=0.00065, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  12%|█▎        | 1250/10000 [00:16<01:55, 76.07it/s, Loss=0.00065, L1 Loss=0.00013, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  12%|█▎        | 1250/10000 [00:17<01:55, 76.07it/s, Loss=0.00073, L1 Loss=0.00026, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1260/10000 [00:17<01:54, 76.49it/s, Loss=0.00073, L1 Loss=0.00026, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00046, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1260/10000 [00:17<01:54, 76.49it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1270/10000 [00:17<01:54, 76.55it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1270/10000 [00:17<01:54, 76.55it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1280/10000 [00:17<01:53, 76.88it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1280/10000 [00:17<01:53, 76.88it/s, Loss=0.00057, L1 Loss=0.00014, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1290/10000 [00:17<01:53, 76.62it/s, Loss=0.00057, L1 Loss=0.00014, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1290/10000 [00:17<01:53, 76.62it/s, Loss=0.00069, L1 Loss=0.00024, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1300/10000 [00:17<01:53, 76.81it/s, Loss=0.00069, L1 Loss=0.00024, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1300/10000 [00:17<01:53, 76.81it/s, Loss=0.00071, L1 Loss=0.00013, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1310/10000 [00:17<01:56, 74.35it/s, Loss=0.00071, L1 Loss=0.00013, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1310/10000 [00:17<01:56, 74.35it/s, Loss=0.00065, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1320/10000 [00:17<01:55, 75.13it/s, Loss=0.00065, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1320/10000 [00:17<01:55, 75.13it/s, Loss=0.00067, L1 Loss=0.00024, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1330/10000 [00:17<01:54, 75.63it/s, Loss=0.00067, L1 Loss=0.00024, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1330/10000 [00:18<01:54, 75.63it/s, Loss=0.00071, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1340/10000 [00:18<01:53, 76.14it/s, Loss=0.00071, L1 Loss=0.00016, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00041, Anisotropic Loss=0.00000]
Training progress:  13%|█▎        | 1340/10000 [00:18<01:53, 76.14it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1350/10000 [00:18<01:53, 76.40it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1350/10000 [00:18<01:53, 76.40it/s, Loss=0.00059, L1 Loss=0.00014, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1360/10000 [00:18<01:53, 76.38it/s, Loss=0.00059, L1 Loss=0.00014, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1360/10000 [00:18<01:53, 76.38it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1370/10000 [00:18<01:52, 76.44it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▎        | 1370/10000 [00:18<01:52, 76.44it/s, Loss=0.00064, L1 Loss=0.00016, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1380/10000 [00:18<01:52, 76.53it/s, Loss=0.00064, L1 Loss=0.00016, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1380/10000 [00:18<01:52, 76.53it/s, Loss=0.00057, L1 Loss=0.00012, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1390/10000 [00:18<01:52, 76.60it/s, Loss=0.00057, L1 Loss=0.00012, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1390/10000 [00:18<01:52, 76.60it/s, Loss=0.00076, L1 Loss=0.00028, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1400/10000 [00:18<01:52, 76.73it/s, Loss=0.00076, L1 Loss=0.00028, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1400/10000 [00:19<01:52, 76.73it/s, Loss=0.00064, L1 Loss=0.00014, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1410/10000 [00:19<01:55, 74.30it/s, Loss=0.00064, L1 Loss=0.00014, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1410/10000 [00:19<01:55, 74.30it/s, Loss=0.00063, L1 Loss=0.00013, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1420/10000 [00:19<01:53, 75.35it/s, Loss=0.00063, L1 Loss=0.00013, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1420/10000 [00:19<01:53, 75.35it/s, Loss=0.00062, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1430/10000 [00:19<01:53, 75.79it/s, Loss=0.00062, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1430/10000 [00:19<01:53, 75.79it/s, Loss=0.00068, L1 Loss=0.00016, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1440/10000 [00:19<01:52, 75.87it/s, Loss=0.00068, L1 Loss=0.00016, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1440/10000 [00:19<01:52, 75.87it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1450/10000 [00:19<01:52, 76.19it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  14%|█▍        | 1450/10000 [00:19<01:52, 76.19it/s, Loss=0.00061, L1 Loss=0.00016, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1460/10000 [00:19<01:52, 75.96it/s, Loss=0.00061, L1 Loss=0.00016, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1460/10000 [00:19<01:52, 75.96it/s, Loss=0.00070, L1 Loss=0.00013, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1470/10000 [00:19<01:51, 76.23it/s, Loss=0.00070, L1 Loss=0.00013, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1470/10000 [00:19<01:51, 76.23it/s, Loss=0.00071, L1 Loss=0.00025, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1480/10000 [00:19<01:51, 76.17it/s, Loss=0.00071, L1 Loss=0.00025, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00044, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1480/10000 [00:20<01:51, 76.17it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1490/10000 [00:20<01:51, 76.48it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  15%|█▍        | 1490/10000 [00:20<01:51, 76.48it/s, Loss=0.00068, L1 Loss=0.00023, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1500/10000 [00:20<01:50, 76.58it/s, Loss=0.00068, L1 Loss=0.00023, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00043, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1500/10000 [00:20<01:50, 76.58it/s, Loss=0.00059, L1 Loss=0.00013, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1510/10000 [00:20<01:54, 74.26it/s, Loss=0.00059, L1 Loss=0.00013, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1510/10000 [00:20<01:54, 74.26it/s, Loss=0.00060, L1 Loss=0.00013, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1520/10000 [00:20<01:52, 75.21it/s, Loss=0.00060, L1 Loss=0.00013, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1520/10000 [00:20<01:52, 75.21it/s, Loss=0.00056, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1530/10000 [00:20<01:51, 75.69it/s, Loss=0.00056, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1530/10000 [00:20<01:51, 75.69it/s, Loss=0.00056, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1540/10000 [00:20<01:51, 75.92it/s, Loss=0.00056, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  15%|█▌        | 1540/10000 [00:20<01:51, 75.92it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1550/10000 [00:20<01:50, 76.17it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1550/10000 [00:21<01:50, 76.17it/s, Loss=0.00060, L1 Loss=0.00014, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1560/10000 [00:21<01:50, 76.49it/s, Loss=0.00060, L1 Loss=0.00014, Depth Loss=0.00093, Normal Loss=0.00000, Seg Loss=0.00040, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1560/10000 [00:21<01:50, 76.49it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1570/10000 [00:21<01:50, 76.38it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1570/10000 [00:21<01:50, 76.38it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1580/10000 [00:21<01:50, 76.50it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1580/10000 [00:21<01:50, 76.50it/s, Loss=0.00053, L1 Loss=0.00012, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1590/10000 [00:21<01:50, 76.34it/s, Loss=0.00053, L1 Loss=0.00012, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1590/10000 [00:21<01:50, 76.34it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1600/10000 [00:21<01:49, 76.66it/s, Loss=0.00058, L1 Loss=0.00013, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1600/10000 [00:21<01:49, 76.66it/s, Loss=0.00070, L1 Loss=0.00022, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1610/10000 [00:21<01:52, 74.39it/s, Loss=0.00070, L1 Loss=0.00022, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1610/10000 [00:21<01:52, 74.39it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1620/10000 [00:21<01:51, 75.11it/s, Loss=0.00067, L1 Loss=0.00023, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  16%|█▌        | 1620/10000 [00:21<01:51, 75.11it/s, Loss=0.00052, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1630/10000 [00:21<01:51, 75.33it/s, Loss=0.00052, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1630/10000 [00:22<01:51, 75.33it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1640/10000 [00:22<01:50, 75.73it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1640/10000 [00:22<01:50, 75.73it/s, Loss=0.00051, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1650/10000 [00:22<01:49, 76.19it/s, Loss=0.00051, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  16%|█▋        | 1650/10000 [00:22<01:49, 76.19it/s, Loss=0.00062, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1660/10000 [00:22<01:48, 76.53it/s, Loss=0.00062, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1660/10000 [00:22<01:48, 76.53it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1670/10000 [00:22<01:48, 76.71it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1670/10000 [00:22<01:48, 76.71it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1680/10000 [00:22<01:48, 76.74it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1680/10000 [00:22<01:48, 76.74it/s, Loss=0.00063, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1690/10000 [00:22<01:47, 77.21it/s, Loss=0.00063, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1690/10000 [00:22<01:47, 77.21it/s, Loss=0.00064, L1 Loss=0.00023, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1700/10000 [00:22<01:47, 77.01it/s, Loss=0.00064, L1 Loss=0.00023, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00042, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1700/10000 [00:22<01:47, 77.01it/s, Loss=0.00065, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1710/10000 [00:22<01:50, 74.85it/s, Loss=0.00065, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1710/10000 [00:23<01:50, 74.85it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1720/10000 [00:23<01:49, 75.40it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1720/10000 [00:23<01:49, 75.40it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1730/10000 [00:23<01:48, 76.13it/s, Loss=0.00062, L1 Loss=0.00014, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1730/10000 [00:23<01:48, 76.13it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1740/10000 [00:23<01:48, 76.45it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  17%|█▋        | 1740/10000 [00:23<01:48, 76.45it/s, Loss=0.00062, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1750/10000 [00:23<01:47, 76.63it/s, Loss=0.00062, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1750/10000 [00:23<01:47, 76.63it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1760/10000 [00:23<01:47, 76.67it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1760/10000 [00:23<01:47, 76.67it/s, Loss=0.00051, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1770/10000 [00:23<01:47, 76.34it/s, Loss=0.00051, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1770/10000 [00:23<01:47, 76.34it/s, Loss=0.00060, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1780/10000 [00:23<01:47, 76.55it/s, Loss=0.00060, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1780/10000 [00:24<01:47, 76.55it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1790/10000 [00:24<01:47, 76.64it/s, Loss=0.00055, L1 Loss=0.00011, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1790/10000 [00:24<01:47, 76.64it/s, Loss=0.00051, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1800/10000 [00:24<01:46, 76.72it/s, Loss=0.00051, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1800/10000 [00:24<01:46, 76.72it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1810/10000 [00:24<01:49, 74.53it/s, Loss=0.00061, L1 Loss=0.00012, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1810/10000 [00:24<01:49, 74.53it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1820/10000 [00:24<01:48, 75.28it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1820/10000 [00:24<01:48, 75.28it/s, Loss=0.00053, L1 Loss=0.00012, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1830/10000 [00:24<01:47, 75.84it/s, Loss=0.00053, L1 Loss=0.00012, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1830/10000 [00:24<01:47, 75.84it/s, Loss=0.00064, L1 Loss=0.00013, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1840/10000 [00:24<01:46, 76.32it/s, Loss=0.00064, L1 Loss=0.00013, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1840/10000 [00:24<01:46, 76.32it/s, Loss=0.00060, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1850/10000 [00:24<01:46, 76.32it/s, Loss=0.00060, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  18%|█▊        | 1850/10000 [00:24<01:46, 76.32it/s, Loss=0.00063, L1 Loss=0.00015, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  19%|█▊        | 1860/10000 [00:24<01:46, 76.60it/s, Loss=0.00063, L1 Loss=0.00015, Depth Loss=0.00094, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  19%|█▊        | 1860/10000 [00:25<01:46, 76.60it/s, Loss=0.00065, L1 Loss=0.00022, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  19%|█▊        | 1870/10000 [00:25<01:46, 76.48it/s, Loss=0.00065, L1 Loss=0.00022, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  19%|█▊        | 1870/10000 [00:25<01:46, 76.48it/s, Loss=0.00057, L1 Loss=0.00012, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1880/10000 [00:25<01:46, 76.44it/s, Loss=0.00057, L1 Loss=0.00012, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1880/10000 [00:25<01:46, 76.44it/s, Loss=0.00056, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1890/10000 [00:25<01:47, 75.65it/s, Loss=0.00056, L1 Loss=0.00014, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1890/10000 [00:25<01:47, 75.65it/s, Loss=0.00056, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1900/10000 [00:25<01:47, 75.39it/s, Loss=0.00056, L1 Loss=0.00011, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1900/10000 [00:25<01:47, 75.39it/s, Loss=0.00061, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1910/10000 [00:25<01:50, 73.29it/s, Loss=0.00061, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1910/10000 [00:25<01:50, 73.29it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1920/10000 [00:25<01:49, 74.09it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1920/10000 [00:25<01:49, 74.09it/s, Loss=0.00059, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1930/10000 [00:25<01:47, 74.89it/s, Loss=0.00059, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1930/10000 [00:26<01:47, 74.89it/s, Loss=0.00060, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1940/10000 [00:26<01:47, 75.01it/s, Loss=0.00060, L1 Loss=0.00013, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  19%|█▉        | 1940/10000 [00:26<01:47, 75.01it/s, Loss=0.00061, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1950/10000 [00:26<01:47, 75.13it/s, Loss=0.00061, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1950/10000 [00:26<01:47, 75.13it/s, Loss=0.00054, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1960/10000 [00:26<01:46, 75.43it/s, Loss=0.00054, L1 Loss=0.00012, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1960/10000 [00:26<01:46, 75.43it/s, Loss=0.00059, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1970/10000 [00:26<01:46, 75.48it/s, Loss=0.00059, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1970/10000 [00:26<01:46, 75.48it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1980/10000 [00:26<01:46, 75.44it/s, Loss=0.00061, L1 Loss=0.00014, Depth Loss=0.00092, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1980/10000 [00:26<01:46, 75.44it/s, Loss=0.00055, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1990/10000 [00:26<01:46, 75.39it/s, Loss=0.00055, L1 Loss=0.00010, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  20%|█▉        | 1990/10000 [00:26<01:46, 75.39it/s, Loss=0.00061, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2000/10000 [00:26<01:45, 75.88it/s, Loss=0.00061, L1 Loss=0.00021, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2000/10000 [00:26<01:45, 75.88it/s, Loss=0.00065, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2010/10000 [00:26<01:48, 73.75it/s, Loss=0.00065, L1 Loss=0.00021, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00039, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2010/10000 [00:27<01:48, 73.75it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2020/10000 [00:27<01:47, 74.48it/s, Loss=0.00058, L1 Loss=0.00012, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2020/10000 [00:27<01:47, 74.48it/s, Loss=0.00065, L1 Loss=0.00020, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2030/10000 [00:27<01:46, 75.10it/s, Loss=0.00065, L1 Loss=0.00020, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2030/10000 [00:27<01:46, 75.10it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2040/10000 [00:27<01:46, 75.07it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2040/10000 [00:27<01:46, 75.07it/s, Loss=0.00052, L1 Loss=0.00010, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2050/10000 [00:27<01:45, 75.20it/s, Loss=0.00052, L1 Loss=0.00010, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  20%|██        | 2050/10000 [00:27<01:45, 75.20it/s, Loss=0.00058, L1 Loss=0.00011, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2060/10000 [00:27<01:45, 75.58it/s, Loss=0.00058, L1 Loss=0.00011, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2060/10000 [00:27<01:45, 75.58it/s, Loss=0.00060, L1 Loss=0.00011, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2070/10000 [00:27<01:44, 75.70it/s, Loss=0.00060, L1 Loss=0.00011, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2070/10000 [00:27<01:44, 75.70it/s, Loss=0.00048, L1 Loss=0.00010, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2080/10000 [00:27<01:44, 75.70it/s, Loss=0.00048, L1 Loss=0.00010, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2080/10000 [00:28<01:44, 75.70it/s, Loss=0.00050, L1 Loss=0.00011, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2090/10000 [00:28<01:44, 75.70it/s, Loss=0.00050, L1 Loss=0.00011, Depth Loss=0.00091, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2090/10000 [00:28<01:44, 75.70it/s, Loss=0.00053, L1 Loss=0.00019, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2100/10000 [00:28<01:44, 75.87it/s, Loss=0.00053, L1 Loss=0.00019, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2100/10000 [00:28<01:44, 75.87it/s, Loss=0.00055, L1 Loss=0.00010, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2110/10000 [00:28<01:46, 74.02it/s, Loss=0.00055, L1 Loss=0.00010, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2110/10000 [00:28<01:46, 74.02it/s, Loss=0.00058, L1 Loss=0.00019, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2120/10000 [00:28<01:45, 74.78it/s, Loss=0.00058, L1 Loss=0.00019, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00038, Anisotropic Loss=0.00000]
Training progress:  21%|██        | 2120/10000 [00:28<01:45, 74.78it/s, Loss=0.00053, L1 Loss=0.00018, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  21%|██▏       | 2130/10000 [00:28<01:44, 75.47it/s, Loss=0.00053, L1 Loss=0.00018, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  21%|██▏       | 2130/10000 [00:28<01:44, 75.47it/s, Loss=0.00053, L1 Loss=0.00018, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  21%|██▏       | 2140/10000 [00:28<01:44, 75.58it/s, Loss=0.00053, L1 Loss=0.00018, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00037, Anisotropic Loss=0.00000]
Training progress:  21%|██▏       | 2140/10000 [00:28<01:44, 75.58it/s, Loss=0.00045, L1 Loss=0.00008, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2150/10000 [00:28<01:43, 75.71it/s, Loss=0.00045, L1 Loss=0.00008, Depth Loss=0.00062, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2150/10000 [00:28<01:43, 75.71it/s, Loss=0.00049, L1 Loss=0.00010, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2160/10000 [00:28<01:42, 76.22it/s, Loss=0.00049, L1 Loss=0.00010, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2160/10000 [00:29<01:42, 76.22it/s, Loss=0.00054, L1 Loss=0.00016, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2170/10000 [00:29<01:42, 76.54it/s, Loss=0.00054, L1 Loss=0.00016, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2170/10000 [00:29<01:42, 76.54it/s, Loss=0.00046, L1 Loss=0.00011, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2180/10000 [00:29<01:42, 76.53it/s, Loss=0.00046, L1 Loss=0.00011, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2180/10000 [00:29<01:42, 76.53it/s, Loss=0.00042, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2190/10000 [00:29<01:42, 76.44it/s, Loss=0.00042, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2190/10000 [00:29<01:42, 76.44it/s, Loss=0.00045, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2200/10000 [00:29<01:41, 76.81it/s, Loss=0.00045, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2200/10000 [00:29<01:41, 76.81it/s, Loss=0.00055, L1 Loss=0.00016, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2210/10000 [00:29<01:44, 74.53it/s, Loss=0.00055, L1 Loss=0.00016, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2210/10000 [00:29<01:44, 74.53it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2220/10000 [00:29<01:43, 74.95it/s, Loss=0.00054, L1 Loss=0.00011, Depth Loss=0.00090, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2220/10000 [00:29<01:43, 74.95it/s, Loss=0.00040, L1 Loss=0.00009, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2230/10000 [00:29<01:43, 75.27it/s, Loss=0.00040, L1 Loss=0.00009, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2230/10000 [00:30<01:43, 75.27it/s, Loss=0.00050, L1 Loss=0.00012, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2240/10000 [00:30<01:42, 75.83it/s, Loss=0.00050, L1 Loss=0.00012, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  22%|██▏       | 2240/10000 [00:30<01:42, 75.83it/s, Loss=0.00047, L1 Loss=0.00009, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  22%|██▎       | 2250/10000 [00:30<01:41, 76.07it/s, Loss=0.00047, L1 Loss=0.00009, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  22%|██▎       | 2250/10000 [00:30<01:41, 76.07it/s, Loss=0.00043, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2260/10000 [00:30<01:41, 76.03it/s, Loss=0.00043, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2260/10000 [00:30<01:41, 76.03it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2270/10000 [00:30<01:41, 75.96it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2270/10000 [00:30<01:41, 75.96it/s, Loss=0.00050, L1 Loss=0.00016, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2280/10000 [00:30<01:41, 76.39it/s, Loss=0.00050, L1 Loss=0.00016, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2280/10000 [00:30<01:41, 76.39it/s, Loss=0.00041, L1 Loss=0.00009, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2290/10000 [00:30<01:40, 76.39it/s, Loss=0.00041, L1 Loss=0.00009, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2290/10000 [00:30<01:40, 76.39it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2300/10000 [00:30<01:40, 76.39it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2300/10000 [00:30<01:40, 76.39it/s, Loss=0.00043, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2310/10000 [00:30<01:43, 74.16it/s, Loss=0.00043, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2310/10000 [00:31<01:43, 74.16it/s, Loss=0.00042, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2320/10000 [00:31<01:42, 75.01it/s, Loss=0.00042, L1 Loss=0.00008, Depth Loss=0.00061, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2320/10000 [00:31<01:42, 75.01it/s, Loss=0.00045, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2330/10000 [00:31<01:41, 75.51it/s, Loss=0.00045, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2330/10000 [00:31<01:41, 75.51it/s, Loss=0.00049, L1 Loss=0.00017, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2340/10000 [00:31<01:40, 76.09it/s, Loss=0.00049, L1 Loss=0.00017, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00036, Anisotropic Loss=0.00000]
Training progress:  23%|██▎       | 2340/10000 [00:31<01:40, 76.09it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2350/10000 [00:31<01:40, 76.28it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2350/10000 [00:31<01:40, 76.28it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2360/10000 [00:31<01:39, 76.55it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2360/10000 [00:31<01:39, 76.55it/s, Loss=0.00042, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2370/10000 [00:31<01:39, 76.80it/s, Loss=0.00042, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▎       | 2370/10000 [00:31<01:39, 76.80it/s, Loss=0.00046, L1 Loss=0.00016, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2380/10000 [00:31<01:39, 76.85it/s, Loss=0.00046, L1 Loss=0.00016, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00034, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2380/10000 [00:31<01:39, 76.85it/s, Loss=0.00042, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2390/10000 [00:31<01:38, 77.05it/s, Loss=0.00042, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2390/10000 [00:32<01:38, 77.05it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2400/10000 [00:32<01:38, 77.03it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2400/10000 [00:32<01:38, 77.03it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2410/10000 [00:32<01:42, 74.02it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2410/10000 [00:32<01:42, 74.02it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2420/10000 [00:32<01:41, 74.61it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2420/10000 [00:32<01:41, 74.61it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2430/10000 [00:32<01:41, 74.82it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2430/10000 [00:32<01:41, 74.82it/s, Loss=0.00040, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2440/10000 [00:32<01:40, 75.35it/s, Loss=0.00040, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2440/10000 [00:32<01:40, 75.35it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2450/10000 [00:32<01:39, 75.65it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  24%|██▍       | 2450/10000 [00:32<01:39, 75.65it/s, Loss=0.00040, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2460/10000 [00:32<01:39, 76.08it/s, Loss=0.00040, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2460/10000 [00:33<01:39, 76.08it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2470/10000 [00:33<01:38, 76.18it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2470/10000 [00:33<01:38, 76.18it/s, Loss=0.00044, L1 Loss=0.00015, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2480/10000 [00:33<01:38, 76.30it/s, Loss=0.00044, L1 Loss=0.00015, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2480/10000 [00:33<01:38, 76.30it/s, Loss=0.00037, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2490/10000 [00:33<01:38, 76.26it/s, Loss=0.00037, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▍       | 2490/10000 [00:33<01:38, 76.26it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2500/10000 [00:33<01:38, 76.36it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2500/10000 [00:33<01:38, 76.36it/s, Loss=0.00044, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2510/10000 [00:33<01:41, 73.96it/s, Loss=0.00044, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2510/10000 [00:33<01:41, 73.96it/s, Loss=0.00045, L1 Loss=0.00015, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2520/10000 [00:33<01:40, 74.62it/s, Loss=0.00045, L1 Loss=0.00015, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00032, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2520/10000 [00:33<01:40, 74.62it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2530/10000 [00:33<01:39, 75.12it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2530/10000 [00:33<01:39, 75.12it/s, Loss=0.00039, L1 Loss=0.00008, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2540/10000 [00:33<01:38, 75.66it/s, Loss=0.00039, L1 Loss=0.00008, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  25%|██▌       | 2540/10000 [00:34<01:38, 75.66it/s, Loss=0.00035, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2550/10000 [00:34<01:38, 75.78it/s, Loss=0.00035, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2550/10000 [00:34<01:38, 75.78it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2560/10000 [00:34<01:37, 76.05it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2560/10000 [00:34<01:37, 76.05it/s, Loss=0.00038, L1 Loss=0.00008, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2570/10000 [00:34<01:37, 76.18it/s, Loss=0.00038, L1 Loss=0.00008, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2570/10000 [00:34<01:37, 76.18it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2580/10000 [00:34<01:37, 76.32it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2580/10000 [00:34<01:37, 76.32it/s, Loss=0.00041, L1 Loss=0.00013, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2590/10000 [00:34<01:37, 76.38it/s, Loss=0.00041, L1 Loss=0.00013, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2590/10000 [00:34<01:37, 76.38it/s, Loss=0.00041, L1 Loss=0.00008, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2600/10000 [00:34<01:36, 76.32it/s, Loss=0.00041, L1 Loss=0.00008, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2600/10000 [00:34<01:36, 76.32it/s, Loss=0.00038, L1 Loss=0.00008, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2610/10000 [00:34<01:39, 74.03it/s, Loss=0.00038, L1 Loss=0.00008, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2610/10000 [00:35<01:39, 74.03it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2620/10000 [00:35<01:38, 75.00it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00060, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  26%|██▌       | 2620/10000 [00:35<01:38, 75.00it/s, Loss=0.00040, L1 Loss=0.00012, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2630/10000 [00:35<01:37, 75.45it/s, Loss=0.00040, L1 Loss=0.00012, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2630/10000 [00:35<01:37, 75.45it/s, Loss=0.00042, L1 Loss=0.00014, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2640/10000 [00:35<01:36, 76.03it/s, Loss=0.00042, L1 Loss=0.00014, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2640/10000 [00:35<01:36, 76.03it/s, Loss=0.00043, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2650/10000 [00:35<01:36, 76.35it/s, Loss=0.00043, L1 Loss=0.00011, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00033, Anisotropic Loss=0.00000]
Training progress:  26%|██▋       | 2650/10000 [00:35<01:36, 76.35it/s, Loss=0.00043, L1 Loss=0.00015, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2660/10000 [00:35<01:35, 76.61it/s, Loss=0.00043, L1 Loss=0.00015, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00030, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2660/10000 [00:35<01:35, 76.61it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2670/10000 [00:35<01:35, 76.65it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2670/10000 [00:35<01:35, 76.65it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2680/10000 [00:35<01:35, 76.74it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2680/10000 [00:35<01:35, 76.74it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2690/10000 [00:35<01:35, 76.85it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2690/10000 [00:36<01:35, 76.85it/s, Loss=0.00037, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2700/10000 [00:36<01:34, 76.95it/s, Loss=0.00037, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2700/10000 [00:36<01:34, 76.95it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2710/10000 [00:36<01:37, 74.75it/s, Loss=0.00043, L1 Loss=0.00009, Depth Loss=0.00089, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2710/10000 [00:36<01:37, 74.75it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2720/10000 [00:36<01:36, 75.24it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2720/10000 [00:36<01:36, 75.24it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2730/10000 [00:36<01:36, 75.64it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2730/10000 [00:36<01:36, 75.64it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2740/10000 [00:36<01:35, 76.24it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00088, Normal Loss=0.00000, Seg Loss=0.00028, Anisotropic Loss=0.00000]
Training progress:  27%|██▋       | 2740/10000 [00:36<01:35, 76.24it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2750/10000 [00:36<01:34, 76.58it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2750/10000 [00:36<01:34, 76.58it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2760/10000 [00:36<01:34, 76.73it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2760/10000 [00:36<01:34, 76.73it/s, Loss=0.00041, L1 Loss=0.00012, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2770/10000 [00:36<01:34, 76.85it/s, Loss=0.00041, L1 Loss=0.00012, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2770/10000 [00:37<01:34, 76.85it/s, Loss=0.00039, L1 Loss=0.00013, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2780/10000 [00:37<01:33, 76.92it/s, Loss=0.00039, L1 Loss=0.00013, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2780/10000 [00:37<01:33, 76.92it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2790/10000 [00:37<01:33, 76.80it/s, Loss=0.00039, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2790/10000 [00:37<01:33, 76.80it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2800/10000 [00:37<01:33, 76.82it/s, Loss=0.00037, L1 Loss=0.00008, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2800/10000 [00:37<01:33, 76.82it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2810/10000 [00:37<01:36, 74.42it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00059, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2810/10000 [00:37<01:36, 74.42it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2820/10000 [00:37<01:35, 75.43it/s, Loss=0.00038, L1 Loss=0.00007, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2820/10000 [00:37<01:35, 75.43it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2830/10000 [00:37<01:34, 75.93it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2830/10000 [00:37<01:34, 75.93it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2840/10000 [00:37<01:33, 76.25it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2840/10000 [00:38<01:33, 76.25it/s, Loss=0.00032, L1 Loss=0.00005, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2850/10000 [00:38<01:33, 76.51it/s, Loss=0.00032, L1 Loss=0.00005, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  28%|██▊       | 2850/10000 [00:38<01:33, 76.51it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  29%|██▊       | 2860/10000 [00:38<01:33, 76.42it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  29%|██▊       | 2860/10000 [00:38<01:33, 76.42it/s, Loss=0.00036, L1 Loss=0.00011, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  29%|██▊       | 2870/10000 [00:38<01:33, 76.50it/s, Loss=0.00036, L1 Loss=0.00011, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  29%|██▊       | 2870/10000 [00:38<01:33, 76.50it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2880/10000 [00:38<01:32, 76.59it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2880/10000 [00:38<01:32, 76.59it/s, Loss=0.00038, L1 Loss=0.00011, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2890/10000 [00:38<01:32, 76.89it/s, Loss=0.00038, L1 Loss=0.00011, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2890/10000 [00:38<01:32, 76.89it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2900/10000 [00:38<01:32, 76.75it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2900/10000 [00:38<01:32, 76.75it/s, Loss=0.00038, L1 Loss=0.00011, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2910/10000 [00:38<01:35, 74.61it/s, Loss=0.00038, L1 Loss=0.00011, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2910/10000 [00:38<01:35, 74.61it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2920/10000 [00:38<01:34, 75.11it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2920/10000 [00:39<01:34, 75.11it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2930/10000 [00:39<01:33, 75.58it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2930/10000 [00:39<01:33, 75.58it/s, Loss=0.00035, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2940/10000 [00:39<01:32, 76.04it/s, Loss=0.00035, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  29%|██▉       | 2940/10000 [00:39<01:32, 76.04it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2950/10000 [00:39<01:32, 76.35it/s, Loss=0.00036, L1 Loss=0.00006, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2950/10000 [00:39<01:32, 76.35it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2960/10000 [00:39<01:32, 76.18it/s, Loss=0.00034, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2960/10000 [00:39<01:32, 76.18it/s, Loss=0.00034, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2970/10000 [00:39<01:32, 76.40it/s, Loss=0.00034, L1 Loss=0.00006, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2970/10000 [00:39<01:32, 76.40it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2980/10000 [00:39<01:32, 76.30it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00085, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2980/10000 [00:39<01:32, 76.30it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2990/10000 [00:39<01:31, 76.36it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00084, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  30%|██▉       | 2990/10000 [00:40<01:31, 76.36it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3000/10000 [00:40<01:31, 76.41it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00058, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3000/10000 [00:40<01:31, 76.41it/s, Loss=0.00375, L1 Loss=0.00035, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00172, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3010/10000 [00:40<01:34, 73.88it/s, Loss=0.00375, L1 Loss=0.00035, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00172, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3010/10000 [00:40<01:34, 73.88it/s, Loss=0.00159, L1 Loss=0.00020, Depth Loss=0.00111, Normal Loss=0.00000, Seg Loss=0.00065, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3020/10000 [00:40<01:33, 74.49it/s, Loss=0.00159, L1 Loss=0.00020, Depth Loss=0.00111, Normal Loss=0.00000, Seg Loss=0.00065, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3020/10000 [00:40<01:33, 74.49it/s, Loss=0.00088, L1 Loss=0.00017, Depth Loss=0.00122, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3030/10000 [00:40<01:33, 74.77it/s, Loss=0.00088, L1 Loss=0.00017, Depth Loss=0.00122, Normal Loss=0.00000, Seg Loss=0.00050, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3030/10000 [00:40<01:33, 74.77it/s, Loss=0.00069, L1 Loss=0.00020, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00048, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3040/10000 [00:40<01:32, 74.98it/s, Loss=0.00069, L1 Loss=0.00020, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00048, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3040/10000 [00:40<01:32, 74.98it/s, Loss=0.00041, L1 Loss=0.00008, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3050/10000 [00:40<01:32, 75.14it/s, Loss=0.00041, L1 Loss=0.00008, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  30%|███       | 3050/10000 [00:40<01:32, 75.14it/s, Loss=0.00050, L1 Loss=0.00016, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3060/10000 [00:40<01:31, 75.54it/s, Loss=0.00050, L1 Loss=0.00016, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00035, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3060/10000 [00:40<01:31, 75.54it/s, Loss=0.00046, L1 Loss=0.00014, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3070/10000 [00:40<01:31, 75.48it/s, Loss=0.00046, L1 Loss=0.00014, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00031, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3070/10000 [00:41<01:31, 75.48it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3080/10000 [00:41<01:31, 75.60it/s, Loss=0.00041, L1 Loss=0.00007, Depth Loss=0.00086, Normal Loss=0.00000, Seg Loss=0.00026, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3080/10000 [00:41<01:31, 75.60it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3090/10000 [00:41<01:31, 75.62it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3090/10000 [00:41<01:31, 75.62it/s, Loss=0.00041, L1 Loss=0.00009, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3100/10000 [00:41<01:31, 75.64it/s, Loss=0.00041, L1 Loss=0.00009, Depth Loss=0.00087, Normal Loss=0.00000, Seg Loss=0.00029, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3100/10000 [00:41<01:31, 75.64it/s, Loss=0.00040, L1 Loss=0.00011, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3110/10000 [00:41<01:33, 73.59it/s, Loss=0.00040, L1 Loss=0.00011, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3110/10000 [00:41<01:33, 73.59it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3120/10000 [00:41<01:33, 73.93it/s, Loss=0.00036, L1 Loss=0.00007, Depth Loss=0.00057, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  31%|███       | 3120/10000 [00:41<01:33, 73.93it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  31%|███▏      | 3130/10000 [00:41<01:32, 74.59it/s, Loss=0.00033, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  31%|███▏      | 3130/10000 [00:41<01:32, 74.59it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  31%|███▏      | 3140/10000 [00:41<01:31, 74.97it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  31%|███▏      | 3140/10000 [00:42<01:31, 74.97it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3150/10000 [00:42<01:30, 75.29it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3150/10000 [00:42<01:30, 75.29it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00055, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3160/10000 [00:42<01:30, 75.51it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00055, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3160/10000 [00:42<01:30, 75.51it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3170/10000 [00:42<01:30, 75.55it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3170/10000 [00:42<01:30, 75.55it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3180/10000 [00:42<01:29, 75.99it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3180/10000 [00:42<01:29, 75.99it/s, Loss=0.00033, L1 Loss=0.00010, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3190/10000 [00:42<01:29, 76.01it/s, Loss=0.00033, L1 Loss=0.00010, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3190/10000 [00:42<01:29, 76.01it/s, Loss=0.00037, L1 Loss=0.00012, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3200/10000 [00:42<01:29, 76.14it/s, Loss=0.00037, L1 Loss=0.00012, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3200/10000 [00:42<01:29, 76.14it/s, Loss=0.00040, L1 Loss=0.00013, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3210/10000 [00:42<01:31, 74.00it/s, Loss=0.00040, L1 Loss=0.00013, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3210/10000 [00:42<01:31, 74.00it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00055, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3220/10000 [00:42<01:31, 74.41it/s, Loss=0.00037, L1 Loss=0.00007, Depth Loss=0.00055, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3220/10000 [00:43<01:31, 74.41it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3230/10000 [00:43<01:30, 74.90it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3230/10000 [00:43<01:30, 74.90it/s, Loss=0.00032, L1 Loss=0.00007, Depth Loss=0.00054, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3240/10000 [00:43<01:29, 75.17it/s, Loss=0.00032, L1 Loss=0.00007, Depth Loss=0.00054, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  32%|███▏      | 3240/10000 [00:43<01:29, 75.17it/s, Loss=0.00033, L1 Loss=0.00009, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  32%|███▎      | 3250/10000 [00:43<01:29, 75.46it/s, Loss=0.00033, L1 Loss=0.00009, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  32%|███▎      | 3250/10000 [00:43<01:29, 75.46it/s, Loss=0.00032, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3260/10000 [00:43<01:29, 75.66it/s, Loss=0.00032, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3260/10000 [00:43<01:29, 75.66it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3270/10000 [00:43<01:28, 75.82it/s, Loss=0.00033, L1 Loss=0.00007, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3270/10000 [00:43<01:28, 75.82it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3280/10000 [00:43<01:28, 75.96it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3280/10000 [00:43<01:28, 75.96it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3290/10000 [00:43<01:28, 75.77it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3290/10000 [00:43<01:28, 75.77it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3300/10000 [00:43<01:28, 76.08it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3300/10000 [00:44<01:28, 76.08it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3310/10000 [00:44<01:30, 73.87it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3310/10000 [00:44<01:30, 73.87it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3320/10000 [00:44<01:29, 74.60it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3320/10000 [00:44<01:29, 74.60it/s, Loss=0.00030, L1 Loss=0.00008, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3330/10000 [00:44<01:28, 75.15it/s, Loss=0.00030, L1 Loss=0.00008, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3330/10000 [00:44<01:28, 75.15it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3340/10000 [00:44<01:28, 75.32it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  33%|███▎      | 3340/10000 [00:44<01:28, 75.32it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3350/10000 [00:44<01:27, 75.75it/s, Loss=0.00032, L1 Loss=0.00006, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3350/10000 [00:44<01:27, 75.75it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3360/10000 [00:44<01:27, 75.62it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3360/10000 [00:44<01:27, 75.62it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3370/10000 [00:44<01:27, 75.60it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▎      | 3370/10000 [00:45<01:27, 75.60it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3380/10000 [00:45<01:27, 75.69it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3380/10000 [00:45<01:27, 75.69it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3390/10000 [00:45<01:27, 75.82it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3390/10000 [00:45<01:27, 75.82it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3400/10000 [00:45<01:26, 75.93it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3400/10000 [00:45<01:26, 75.93it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3410/10000 [00:45<01:29, 73.75it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3410/10000 [00:45<01:29, 73.75it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3420/10000 [00:45<01:28, 74.45it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3420/10000 [00:45<01:28, 74.45it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3430/10000 [00:45<01:27, 74.91it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3430/10000 [00:45<01:27, 74.91it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3440/10000 [00:45<01:26, 75.52it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00082, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3440/10000 [00:45<01:26, 75.52it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3450/10000 [00:45<01:26, 75.91it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  34%|███▍      | 3450/10000 [00:46<01:26, 75.91it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3460/10000 [00:46<01:25, 76.10it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3460/10000 [00:46<01:25, 76.10it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3470/10000 [00:46<01:25, 76.32it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3470/10000 [00:46<01:25, 76.32it/s, Loss=0.00032, L1 Loss=0.00009, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3480/10000 [00:46<01:25, 76.53it/s, Loss=0.00032, L1 Loss=0.00009, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3480/10000 [00:46<01:25, 76.53it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3490/10000 [00:46<01:24, 76.63it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  35%|███▍      | 3490/10000 [00:46<01:24, 76.63it/s, Loss=0.00031, L1 Loss=0.00009, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3500/10000 [00:46<01:25, 76.26it/s, Loss=0.00031, L1 Loss=0.00009, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3500/10000 [00:46<01:25, 76.26it/s, Loss=0.00031, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3510/10000 [00:46<01:27, 73.98it/s, Loss=0.00031, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3510/10000 [00:46<01:27, 73.98it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3520/10000 [00:46<01:26, 74.72it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3520/10000 [00:47<01:26, 74.72it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3530/10000 [00:47<01:25, 75.52it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3530/10000 [00:47<01:25, 75.52it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3540/10000 [00:47<01:25, 75.66it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  35%|███▌      | 3540/10000 [00:47<01:25, 75.66it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3550/10000 [00:47<01:24, 75.95it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3550/10000 [00:47<01:24, 75.95it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3560/10000 [00:47<01:24, 76.25it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3560/10000 [00:47<01:24, 76.25it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3570/10000 [00:47<01:24, 76.39it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3570/10000 [00:47<01:24, 76.39it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3580/10000 [00:47<01:24, 76.39it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3580/10000 [00:47<01:24, 76.39it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3590/10000 [00:47<01:23, 76.50it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3590/10000 [00:47<01:23, 76.50it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3600/10000 [00:47<01:23, 76.76it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3600/10000 [00:48<01:23, 76.76it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3610/10000 [00:48<01:25, 74.32it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3610/10000 [00:48<01:25, 74.32it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3620/10000 [00:48<01:25, 74.80it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  36%|███▌      | 3620/10000 [00:48<01:25, 74.80it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3630/10000 [00:48<01:24, 75.24it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3630/10000 [00:48<01:24, 75.24it/s, Loss=0.00035, L1 Loss=0.00009, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3640/10000 [00:48<01:24, 75.56it/s, Loss=0.00035, L1 Loss=0.00009, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00027, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3640/10000 [00:48<01:24, 75.56it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3650/10000 [00:48<01:23, 75.84it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  36%|███▋      | 3650/10000 [00:48<01:23, 75.84it/s, Loss=0.00030, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3660/10000 [00:48<01:23, 75.91it/s, Loss=0.00030, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3660/10000 [00:48<01:23, 75.91it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3670/10000 [00:48<01:23, 76.08it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3670/10000 [00:49<01:23, 76.08it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3680/10000 [00:49<01:22, 76.26it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3680/10000 [00:49<01:22, 76.26it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3690/10000 [00:49<01:22, 76.30it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3690/10000 [00:49<01:22, 76.30it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3700/10000 [00:49<01:22, 76.27it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3700/10000 [00:49<01:22, 76.27it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3710/10000 [00:49<01:25, 73.96it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00080, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3710/10000 [00:49<01:25, 73.96it/s, Loss=0.00030, L1 Loss=0.00009, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3720/10000 [00:49<01:24, 74.72it/s, Loss=0.00030, L1 Loss=0.00009, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3720/10000 [00:49<01:24, 74.72it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3730/10000 [00:49<01:23, 75.34it/s, Loss=0.00031, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3730/10000 [00:49<01:23, 75.34it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3740/10000 [00:49<01:22, 75.80it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  37%|███▋      | 3740/10000 [00:49<01:22, 75.80it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3750/10000 [00:49<01:22, 76.15it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00083, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3750/10000 [00:50<01:22, 76.15it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3760/10000 [00:50<01:21, 76.20it/s, Loss=0.00035, L1 Loss=0.00006, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00023, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3760/10000 [00:50<01:21, 76.20it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3770/10000 [00:50<01:21, 76.30it/s, Loss=0.00031, L1 Loss=0.00006, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3770/10000 [00:50<01:21, 76.30it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3780/10000 [00:50<01:21, 76.45it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3780/10000 [00:50<01:21, 76.45it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3790/10000 [00:50<01:21, 76.61it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3790/10000 [00:50<01:21, 76.61it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3800/10000 [00:50<01:21, 76.42it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00053, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3800/10000 [00:50<01:21, 76.42it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3810/10000 [00:50<01:23, 74.37it/s, Loss=0.00029, L1 Loss=0.00004, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3810/10000 [00:50<01:23, 74.37it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3820/10000 [00:50<01:22, 74.88it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3820/10000 [00:51<01:22, 74.88it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3830/10000 [00:51<01:22, 75.22it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3830/10000 [00:51<01:22, 75.22it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3840/10000 [00:51<01:21, 75.71it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3840/10000 [00:51<01:21, 75.71it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3850/10000 [00:51<01:20, 75.99it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  38%|███▊      | 3850/10000 [00:51<01:20, 75.99it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  39%|███▊      | 3860/10000 [00:51<01:20, 76.09it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  39%|███▊      | 3860/10000 [00:51<01:20, 76.09it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  39%|███▊      | 3870/10000 [00:51<01:20, 76.34it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  39%|███▊      | 3870/10000 [00:51<01:20, 76.34it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3880/10000 [00:51<01:20, 76.40it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3880/10000 [00:51<01:20, 76.40it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3890/10000 [00:51<01:19, 76.54it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3890/10000 [00:51<01:19, 76.54it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3900/10000 [00:51<01:19, 76.73it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3900/10000 [00:52<01:19, 76.73it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3910/10000 [00:52<01:22, 74.12it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3910/10000 [00:52<01:22, 74.12it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3920/10000 [00:52<01:21, 75.03it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00081, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3920/10000 [00:52<01:21, 75.03it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3930/10000 [00:52<01:20, 75.65it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3930/10000 [00:52<01:20, 75.65it/s, Loss=0.00035, L1 Loss=0.00011, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3940/10000 [00:52<01:19, 76.02it/s, Loss=0.00035, L1 Loss=0.00011, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00025, Anisotropic Loss=0.00000]
Training progress:  39%|███▉      | 3940/10000 [00:52<01:19, 76.02it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3950/10000 [00:52<01:19, 76.47it/s, Loss=0.00031, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3950/10000 [00:52<01:19, 76.47it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3960/10000 [00:52<01:19, 76.32it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3960/10000 [00:52<01:19, 76.32it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3970/10000 [00:52<01:18, 76.34it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3970/10000 [00:52<01:18, 76.34it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3980/10000 [00:52<01:18, 76.68it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3980/10000 [00:53<01:18, 76.68it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3990/10000 [00:53<01:18, 76.69it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  40%|███▉      | 3990/10000 [00:53<01:18, 76.69it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4000/10000 [00:53<01:18, 76.75it/s, Loss=0.00030, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4000/10000 [00:53<01:18, 76.75it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4010/10000 [00:53<01:20, 74.33it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4010/10000 [00:53<01:20, 74.33it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4020/10000 [00:53<01:19, 75.33it/s, Loss=0.00028, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4020/10000 [00:53<01:19, 75.33it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4030/10000 [00:53<01:18, 75.91it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4030/10000 [00:53<01:18, 75.91it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4040/10000 [00:53<01:18, 76.23it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4040/10000 [00:53<01:18, 76.23it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4050/10000 [00:53<01:17, 76.38it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  40%|████      | 4050/10000 [00:54<01:17, 76.38it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4060/10000 [00:54<01:17, 76.46it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4060/10000 [00:54<01:17, 76.46it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4070/10000 [00:54<01:17, 76.57it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4070/10000 [00:54<01:17, 76.57it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4080/10000 [00:54<01:17, 76.61it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4080/10000 [00:54<01:17, 76.61it/s, Loss=0.00029, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4090/10000 [00:54<01:16, 76.90it/s, Loss=0.00029, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4090/10000 [00:54<01:16, 76.90it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4100/10000 [00:54<01:16, 76.89it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4100/10000 [00:54<01:16, 76.89it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4110/10000 [00:54<01:18, 74.66it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4110/10000 [00:54<01:18, 74.66it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4120/10000 [00:54<01:18, 75.12it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  41%|████      | 4120/10000 [00:54<01:18, 75.12it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  41%|████▏     | 4130/10000 [00:54<01:17, 75.73it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  41%|████▏     | 4130/10000 [00:55<01:17, 75.73it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  41%|████▏     | 4140/10000 [00:55<01:16, 76.12it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  41%|████▏     | 4140/10000 [00:55<01:16, 76.12it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4150/10000 [00:55<01:16, 76.43it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4150/10000 [00:55<01:16, 76.43it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4160/10000 [00:55<01:16, 76.46it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4160/10000 [00:55<01:16, 76.46it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4170/10000 [00:55<01:16, 76.55it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4170/10000 [00:55<01:16, 76.55it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4180/10000 [00:55<01:15, 76.60it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4180/10000 [00:55<01:15, 76.60it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4190/10000 [00:55<01:15, 76.77it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4190/10000 [00:55<01:15, 76.77it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4200/10000 [00:55<01:15, 76.73it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4200/10000 [00:56<01:15, 76.73it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4210/10000 [00:56<01:17, 74.41it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4210/10000 [00:56<01:17, 74.41it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4220/10000 [00:56<01:17, 74.98it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4220/10000 [00:56<01:17, 74.98it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4230/10000 [00:56<01:16, 75.67it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4230/10000 [00:56<01:16, 75.67it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4240/10000 [00:56<01:15, 76.05it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  42%|████▏     | 4240/10000 [00:56<01:15, 76.05it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  42%|████▎     | 4250/10000 [00:56<01:15, 76.41it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  42%|████▎     | 4250/10000 [00:56<01:15, 76.41it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4260/10000 [00:56<01:14, 76.61it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4260/10000 [00:56<01:14, 76.61it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4270/10000 [00:56<01:14, 76.83it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4270/10000 [00:56<01:14, 76.83it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4280/10000 [00:56<01:14, 76.68it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4280/10000 [00:57<01:14, 76.68it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4290/10000 [00:57<01:14, 76.68it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4290/10000 [00:57<01:14, 76.68it/s, Loss=0.00027, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4300/10000 [00:57<01:14, 77.02it/s, Loss=0.00027, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4300/10000 [00:57<01:14, 77.02it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4310/10000 [00:57<01:16, 74.59it/s, Loss=0.00029, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4310/10000 [00:57<01:16, 74.59it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4320/10000 [00:57<01:15, 74.79it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4320/10000 [00:57<01:15, 74.79it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4330/10000 [00:57<01:15, 75.48it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4330/10000 [00:57<01:15, 75.48it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4340/10000 [00:57<01:14, 75.78it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  43%|████▎     | 4340/10000 [00:57<01:14, 75.78it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4350/10000 [00:57<01:14, 76.27it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4350/10000 [00:57<01:14, 76.27it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4360/10000 [00:57<01:13, 76.60it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4360/10000 [00:58<01:13, 76.60it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4370/10000 [00:58<01:13, 76.77it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▎     | 4370/10000 [00:58<01:13, 76.77it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4380/10000 [00:58<01:12, 77.00it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4380/10000 [00:58<01:12, 77.00it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4390/10000 [00:58<01:12, 77.08it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4390/10000 [00:58<01:12, 77.08it/s, Loss=0.00031, L1 Loss=0.00009, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4400/10000 [00:58<01:12, 76.78it/s, Loss=0.00031, L1 Loss=0.00009, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00024, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4400/10000 [00:58<01:12, 76.78it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4410/10000 [00:58<01:15, 74.26it/s, Loss=0.00029, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4410/10000 [00:58<01:15, 74.26it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4420/10000 [00:58<01:14, 75.20it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4420/10000 [00:58<01:14, 75.20it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4430/10000 [00:58<01:13, 75.81it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00074, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4430/10000 [00:59<01:13, 75.81it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4440/10000 [00:59<01:13, 76.12it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00073, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4440/10000 [00:59<01:13, 76.12it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4450/10000 [00:59<01:12, 76.22it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  44%|████▍     | 4450/10000 [00:59<01:12, 76.22it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4460/10000 [00:59<01:12, 76.48it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4460/10000 [00:59<01:12, 76.48it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4470/10000 [00:59<01:12, 76.72it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4470/10000 [00:59<01:12, 76.72it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4480/10000 [00:59<01:11, 76.72it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4480/10000 [00:59<01:11, 76.72it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4490/10000 [00:59<01:11, 76.94it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  45%|████▍     | 4490/10000 [00:59<01:11, 76.94it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4500/10000 [00:59<01:11, 76.86it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4500/10000 [00:59<01:11, 76.86it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4510/10000 [00:59<01:13, 74.66it/s, Loss=0.00030, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4510/10000 [01:00<01:13, 74.66it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4520/10000 [01:00<01:12, 75.16it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4520/10000 [01:00<01:12, 75.16it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4530/10000 [01:00<01:12, 75.83it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4530/10000 [01:00<01:12, 75.83it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4540/10000 [01:00<01:11, 76.12it/s, Loss=0.00030, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  45%|████▌     | 4540/10000 [01:00<01:11, 76.12it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4550/10000 [01:00<01:11, 76.64it/s, Loss=0.00029, L1 Loss=0.00005, Depth Loss=0.00079, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4550/10000 [01:00<01:11, 76.64it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4560/10000 [01:00<01:10, 76.66it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00072, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4560/10000 [01:00<01:10, 76.66it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4570/10000 [01:00<01:10, 76.85it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4570/10000 [01:00<01:10, 76.85it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4580/10000 [01:00<01:10, 76.86it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4580/10000 [01:00<01:10, 76.86it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4590/10000 [01:00<01:10, 76.96it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4590/10000 [01:01<01:10, 76.96it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4600/10000 [01:01<01:09, 77.17it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4600/10000 [01:01<01:09, 77.17it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4610/10000 [01:01<01:11, 74.95it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4610/10000 [01:01<01:11, 74.95it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4620/10000 [01:01<01:11, 75.44it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00052, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  46%|████▌     | 4620/10000 [01:01<01:11, 75.44it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4630/10000 [01:01<01:10, 75.92it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4630/10000 [01:01<01:10, 75.92it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4640/10000 [01:01<01:10, 76.10it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4640/10000 [01:01<01:10, 76.10it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4650/10000 [01:01<01:10, 76.39it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  46%|████▋     | 4650/10000 [01:01<01:10, 76.39it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4660/10000 [01:01<01:09, 76.49it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4660/10000 [01:02<01:09, 76.49it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4670/10000 [01:02<01:09, 76.72it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4670/10000 [01:02<01:09, 76.72it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4680/10000 [01:02<01:09, 76.95it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4680/10000 [01:02<01:09, 76.95it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4690/10000 [01:02<01:08, 77.10it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00067, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4690/10000 [01:02<01:08, 77.10it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4700/10000 [01:02<01:08, 77.10it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4700/10000 [01:02<01:08, 77.10it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4710/10000 [01:02<01:10, 74.79it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4710/10000 [01:02<01:10, 74.79it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4720/10000 [01:02<01:09, 75.57it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4720/10000 [01:02<01:09, 75.57it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4730/10000 [01:02<01:08, 76.38it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4730/10000 [01:02<01:08, 76.38it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4740/10000 [01:02<01:08, 76.57it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  47%|████▋     | 4740/10000 [01:03<01:08, 76.57it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4750/10000 [01:03<01:08, 76.94it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4750/10000 [01:03<01:08, 76.94it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4760/10000 [01:03<01:08, 76.91it/s, Loss=0.00028, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4760/10000 [01:03<01:08, 76.91it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4770/10000 [01:03<01:08, 76.86it/s, Loss=0.00028, L1 Loss=0.00006, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4770/10000 [01:03<01:08, 76.86it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4780/10000 [01:03<01:07, 76.88it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4780/10000 [01:03<01:07, 76.88it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4790/10000 [01:03<01:07, 77.29it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4790/10000 [01:03<01:07, 77.29it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4800/10000 [01:03<01:07, 77.10it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4800/10000 [01:03<01:07, 77.10it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4810/10000 [01:03<01:09, 74.59it/s, Loss=0.00028, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4810/10000 [01:04<01:09, 74.59it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4820/10000 [01:04<01:08, 75.22it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4820/10000 [01:04<01:08, 75.22it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4830/10000 [01:04<01:08, 75.96it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00078, Normal Loss=0.00000, Seg Loss=0.00022, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4830/10000 [01:04<01:08, 75.96it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4840/10000 [01:04<01:07, 76.46it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4840/10000 [01:04<01:07, 76.46it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4850/10000 [01:04<01:07, 76.68it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  48%|████▊     | 4850/10000 [01:04<01:07, 76.68it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  49%|████▊     | 4860/10000 [01:04<01:06, 76.79it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  49%|████▊     | 4860/10000 [01:04<01:06, 76.79it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▊     | 4870/10000 [01:04<01:06, 77.11it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▊     | 4870/10000 [01:04<01:06, 77.11it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4880/10000 [01:04<01:06, 76.95it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4880/10000 [01:04<01:06, 76.95it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4890/10000 [01:04<01:06, 77.24it/s, Loss=0.00028, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4890/10000 [01:05<01:06, 77.24it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4900/10000 [01:05<01:06, 77.07it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4900/10000 [01:05<01:06, 77.07it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4910/10000 [01:05<01:08, 74.70it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4910/10000 [01:05<01:08, 74.70it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4920/10000 [01:05<01:07, 75.44it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4920/10000 [01:05<01:07, 75.44it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4930/10000 [01:05<01:06, 76.09it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4930/10000 [01:05<01:06, 76.09it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4940/10000 [01:05<01:06, 76.33it/s, Loss=0.00027, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  49%|████▉     | 4940/10000 [01:05<01:06, 76.33it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4950/10000 [01:05<01:05, 76.64it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4950/10000 [01:05<01:05, 76.64it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4960/10000 [01:05<01:05, 76.87it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4960/10000 [01:05<01:05, 76.87it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4970/10000 [01:05<01:05, 77.06it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4970/10000 [01:06<01:05, 77.06it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4980/10000 [01:06<01:05, 77.13it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4980/10000 [01:06<01:05, 77.13it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4990/10000 [01:06<01:04, 77.19it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  50%|████▉     | 4990/10000 [01:06<01:04, 77.19it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5000/10000 [01:06<01:04, 77.21it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5000/10000 [01:06<01:04, 77.21it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5010/10000 [01:06<01:06, 74.74it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5010/10000 [01:06<01:06, 74.74it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5020/10000 [01:06<01:06, 75.30it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5020/10000 [01:06<01:06, 75.30it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5030/10000 [01:06<01:05, 75.93it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5030/10000 [01:06<01:05, 75.93it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5040/10000 [01:06<01:04, 76.42it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5040/10000 [01:07<01:04, 76.42it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5050/10000 [01:07<01:04, 76.57it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  50%|█████     | 5050/10000 [01:07<01:04, 76.57it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5060/10000 [01:07<01:04, 76.86it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5060/10000 [01:07<01:04, 76.86it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5070/10000 [01:07<01:04, 76.59it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5070/10000 [01:07<01:04, 76.59it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5080/10000 [01:07<01:03, 76.98it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00051, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5080/10000 [01:07<01:03, 76.98it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5090/10000 [01:07<01:03, 77.14it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5090/10000 [01:07<01:03, 77.14it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5100/10000 [01:07<01:03, 77.16it/s, Loss=0.00026, L1 Loss=0.00005, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5100/10000 [01:07<01:03, 77.16it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5110/10000 [01:07<01:05, 74.86it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5110/10000 [01:07<01:05, 74.86it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5120/10000 [01:07<01:04, 75.62it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  51%|█████     | 5120/10000 [01:08<01:04, 75.62it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  51%|█████▏    | 5130/10000 [01:08<01:04, 75.90it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00077, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  51%|█████▏    | 5130/10000 [01:08<01:04, 75.90it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████▏    | 5140/10000 [01:08<01:03, 76.08it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  51%|█████▏    | 5140/10000 [01:08<01:03, 76.08it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5150/10000 [01:08<01:03, 76.59it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5150/10000 [01:08<01:03, 76.59it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5160/10000 [01:08<01:03, 76.72it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5160/10000 [01:08<01:03, 76.72it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5170/10000 [01:08<01:02, 77.16it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5170/10000 [01:08<01:02, 77.16it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5180/10000 [01:08<01:02, 77.03it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5180/10000 [01:08<01:02, 77.03it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5190/10000 [01:08<01:02, 77.17it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5190/10000 [01:08<01:02, 77.17it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5200/10000 [01:08<01:02, 77.36it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5200/10000 [01:09<01:02, 77.36it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5210/10000 [01:09<01:03, 75.12it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5210/10000 [01:09<01:03, 75.12it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5220/10000 [01:09<01:02, 75.89it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5220/10000 [01:09<01:02, 75.89it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5230/10000 [01:09<01:02, 76.44it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5230/10000 [01:09<01:02, 76.44it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5240/10000 [01:09<01:02, 76.73it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  52%|█████▏    | 5240/10000 [01:09<01:02, 76.73it/s, Loss=0.00024, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  52%|█████▎    | 5250/10000 [01:09<01:01, 77.04it/s, Loss=0.00024, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  52%|█████▎    | 5250/10000 [01:09<01:01, 77.04it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5260/10000 [01:09<01:01, 77.17it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5260/10000 [01:09<01:01, 77.17it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5270/10000 [01:09<01:00, 77.54it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5270/10000 [01:10<01:00, 77.54it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5280/10000 [01:10<01:00, 77.63it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5280/10000 [01:10<01:00, 77.63it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5290/10000 [01:10<01:00, 77.87it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5290/10000 [01:10<01:00, 77.87it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5300/10000 [01:10<01:00, 77.78it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5300/10000 [01:10<01:00, 77.78it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5310/10000 [01:10<01:02, 75.13it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5310/10000 [01:10<01:02, 75.13it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5320/10000 [01:10<01:01, 75.75it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5320/10000 [01:10<01:01, 75.75it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5330/10000 [01:10<01:01, 76.27it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5330/10000 [01:10<01:01, 76.27it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5340/10000 [01:10<01:00, 76.55it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  53%|█████▎    | 5340/10000 [01:10<01:00, 76.55it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5350/10000 [01:10<01:00, 76.75it/s, Loss=0.00027, L1 Loss=0.00005, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5350/10000 [01:11<01:00, 76.75it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5360/10000 [01:11<01:00, 77.05it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5360/10000 [01:11<01:00, 77.05it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5370/10000 [01:11<00:59, 77.34it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  54%|█████▎    | 5370/10000 [01:11<00:59, 77.34it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5380/10000 [01:11<00:59, 77.21it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00071, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5380/10000 [01:11<00:59, 77.21it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5390/10000 [01:11<00:59, 77.23it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5390/10000 [01:11<00:59, 77.23it/s, Loss=0.00026, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5400/10000 [01:11<00:59, 77.26it/s, Loss=0.00026, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5400/10000 [01:11<00:59, 77.26it/s, Loss=0.00029, L1 Loss=0.00009, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5410/10000 [01:11<01:01, 74.88it/s, Loss=0.00029, L1 Loss=0.00009, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00021, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5410/10000 [01:11<01:01, 74.88it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5420/10000 [01:11<01:00, 75.78it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5420/10000 [01:11<01:00, 75.78it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5430/10000 [01:11<00:59, 76.30it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5430/10000 [01:12<00:59, 76.30it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5440/10000 [01:12<00:59, 76.64it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  54%|█████▍    | 5440/10000 [01:12<00:59, 76.64it/s, Loss=0.00022, L1 Loss=0.00003, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00014, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5450/10000 [01:12<00:59, 76.92it/s, Loss=0.00022, L1 Loss=0.00003, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00014, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5450/10000 [01:12<00:59, 76.92it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5460/10000 [01:12<00:58, 77.23it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5460/10000 [01:12<00:58, 77.23it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5470/10000 [01:12<00:58, 77.09it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5470/10000 [01:12<00:58, 77.09it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5480/10000 [01:12<00:58, 77.26it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5480/10000 [01:12<00:58, 77.26it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5490/10000 [01:12<00:58, 77.33it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  55%|█████▍    | 5490/10000 [01:12<00:58, 77.33it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5500/10000 [01:12<00:58, 77.42it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00020, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5500/10000 [01:13<00:58, 77.42it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5510/10000 [01:13<00:59, 74.99it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5510/10000 [01:13<00:59, 74.99it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5520/10000 [01:13<00:59, 75.83it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5520/10000 [01:13<00:59, 75.83it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5530/10000 [01:13<00:58, 76.39it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5530/10000 [01:13<00:58, 76.39it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5540/10000 [01:13<00:58, 76.80it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  55%|█████▌    | 5540/10000 [01:13<00:58, 76.80it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5550/10000 [01:13<00:57, 77.12it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5550/10000 [01:13<00:57, 77.12it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5560/10000 [01:13<00:57, 77.04it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5560/10000 [01:13<00:57, 77.04it/s, Loss=0.00022, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00014, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5570/10000 [01:13<00:57, 76.86it/s, Loss=0.00022, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00014, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5570/10000 [01:13<00:57, 76.86it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5580/10000 [01:13<00:57, 77.41it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5580/10000 [01:14<00:57, 77.41it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5590/10000 [01:14<00:56, 77.50it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00068, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5590/10000 [01:14<00:56, 77.50it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5600/10000 [01:14<00:56, 77.66it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5600/10000 [01:14<00:56, 77.66it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5610/10000 [01:14<00:58, 75.32it/s, Loss=0.00027, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5610/10000 [01:14<00:58, 75.32it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5620/10000 [01:14<00:57, 75.85it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  56%|█████▌    | 5620/10000 [01:14<00:57, 75.85it/s, Loss=0.00023, L1 Loss=0.00005, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5630/10000 [01:14<00:57, 76.28it/s, Loss=0.00023, L1 Loss=0.00005, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5630/10000 [01:14<00:57, 76.28it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5640/10000 [01:14<00:56, 76.76it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00064, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5640/10000 [01:14<00:56, 76.76it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5650/10000 [01:14<00:56, 76.96it/s, Loss=0.00026, L1 Loss=0.00007, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  56%|█████▋    | 5650/10000 [01:14<00:56, 76.96it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5660/10000 [01:14<00:56, 77.08it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5660/10000 [01:15<00:56, 77.08it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5670/10000 [01:15<00:56, 77.07it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5670/10000 [01:15<00:56, 77.07it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5680/10000 [01:15<00:55, 77.34it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5680/10000 [01:15<00:55, 77.34it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5690/10000 [01:15<00:55, 77.54it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5690/10000 [01:15<00:55, 77.54it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5700/10000 [01:15<00:55, 77.47it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00069, Normal Loss=0.00000, Seg Loss=0.00015, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5700/10000 [01:15<00:55, 77.47it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5710/10000 [01:15<00:57, 74.94it/s, Loss=0.00023, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5710/10000 [01:15<00:57, 74.94it/s, Loss=0.00026, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5720/10000 [01:15<00:56, 75.88it/s, Loss=0.00026, L1 Loss=0.00008, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5720/10000 [01:15<00:56, 75.88it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5730/10000 [01:15<00:55, 76.50it/s, Loss=0.00026, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5730/10000 [01:16<00:55, 76.50it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5740/10000 [01:16<00:55, 76.77it/s, Loss=0.00026, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  57%|█████▋    | 5740/10000 [01:16<00:55, 76.77it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▊    | 5750/10000 [01:16<00:55, 76.91it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  57%|█████▊    | 5750/10000 [01:16<00:55, 76.91it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5760/10000 [01:16<00:54, 77.16it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5760/10000 [01:16<00:54, 77.16it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5770/10000 [01:16<00:54, 77.22it/s, Loss=0.00023, L1 Loss=0.00004, Depth Loss=0.00050, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5770/10000 [01:16<00:54, 77.22it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5780/10000 [01:16<00:54, 77.51it/s, Loss=0.00025, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5780/10000 [01:16<00:54, 77.51it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5790/10000 [01:16<00:54, 77.47it/s, Loss=0.00024, L1 Loss=0.00004, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5790/10000 [01:16<00:54, 77.47it/s, Loss=0.00027, L1 Loss=0.00008, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5800/10000 [01:16<00:54, 77.63it/s, Loss=0.00027, L1 Loss=0.00008, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5800/10000 [01:16<00:54, 77.63it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5810/10000 [01:16<00:55, 75.40it/s, Loss=0.00027, L1 Loss=0.00004, Depth Loss=0.00075, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5810/10000 [01:17<00:55, 75.40it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5820/10000 [01:17<00:55, 75.84it/s, Loss=0.00025, L1 Loss=0.00005, Depth Loss=0.00070, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5820/10000 [01:17<00:55, 75.84it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5830/10000 [01:17<00:54, 76.43it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5830/10000 [01:17<00:54, 76.43it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5840/10000 [01:17<00:54, 76.85it/s, Loss=0.00025, L1 Loss=0.00004, Depth Loss=0.00076, Normal Loss=0.00000, Seg Loss=0.00019, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5840/10000 [01:17<00:54, 76.85it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5850/10000 [01:17<00:53, 77.25it/s, Loss=0.00024, L1 Loss=0.00003, Depth Loss=0.00049, Normal Loss=0.00000, Seg Loss=0.00016, Anisotropic Loss=0.00000]
Training progress:  58%|█████▊    | 5850/10000 [01:17<00:53, 77.25it/s, Loss=0.00024, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  59%|█████▊    | 5860/10000 [01:17<00:53, 77.36it/s, Loss=0.00024, L1 Loss=0.00006, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  59%|█████▊    | 5860/10000 [01:17<00:53, 77.36it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  59%|█████▊    | 5870/10000 [01:17<00:53, 77.47it/s, Loss=0.00025, L1 Loss=0.00006, Depth Loss=0.00066, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  59%|█████▊    | 5870/10000 [01:17<00:53, 77.47it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  59%|█████▉    | 5880/10000 [01:17<00:53, 77.58it/s, Loss=0.00025, L1 Loss=0.00007, Depth Loss=0.00063, Normal Loss=0.00000, Seg Loss=0.00018, Anisotropic Loss=0.00000]
Training progress:  59%|█████▉    | 5880/10000 [01:17<00:53, 77.58it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
Training progress:  59%|█████▉    | 5890/10000 [01:17<00:52, 77.59it/s, Loss=0.00024, L1 Loss=0.00005, Depth Loss=0.00065, Normal Loss=0.00000, Seg Loss=0.00017, Anisotropic Loss=0.00000]
\ No newline at end of file
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..b4e9ec551405e67dc458af9389f3d3b05e8f7152
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/appearance_train_v1_argv.json
@@ -0,0 +1,18 @@
+{
+  "argv": [
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_gs_train_env/.venv/bin/python",
+    "scripts/real_rgbd_appearance_train.py",
+    "train",
+    "--prior",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "--accepted-prior-manifest-sha256",
+    "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/appearance_train_v1"
+  ],
+  "cwd": "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027",
+  "environment": {
+    "PYTHONDONTWRITEBYTECODE": "1",
+    "MPLBACKEND": "Agg"
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors.log
new file mode 100644
index 0000000000000000000000000000000000000000..74f7e5f9ca1e2b0b0a46d0d1c584efc553dc0348
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors.log
@@ -0,0 +1 @@
+colors: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/colors.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..4ef0debca883c9db90258f101242bde0fca0ae5c
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "colors",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:05:17.830380+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..e222192a6d747e27724b6167a23ddc36b3623aa3
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_colors_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 2.6732046604156494,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_complete.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_complete.json
new file mode 100644
index 0000000000000000000000000000000000000000..c0e4393c34b45e2d2dbc17b2a2d9783b8e4c3c6a
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_complete.json
@@ -0,0 +1,5 @@
+{
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "phase": "prior",
+  "completed_utc": "2026-10-28T22:05:36.192812+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth.log
new file mode 100644
index 0000000000000000000000000000000000000000..a2715bd6abcd2f8d1b5024dea5b911a773cf991f
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth.log
@@ -0,0 +1 @@
+diagnose-depth: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/diagnose-depth.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f2c77c2603bfd54b739fdc7f9ae5a9d9250c153
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "diagnose-depth",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:04:49.176796+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..44dd9c72ae54487e2477a2ee5cce4f8ca160ab70
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_diagnose_depth_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 1.5730030536651611,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit.log
new file mode 100644
index 0000000000000000000000000000000000000000..d06a8df9ae2a857299762964cbd9ac22c827f926
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit.log
@@ -0,0 +1,43 @@
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+compiling!
+fit: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/fit.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..f053e837b3f1a846ec495cf1e2d59b7c5f8b02b6
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_argv.json
@@ -0,0 +1,22 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "fit",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1",
+    "--outlier-filter",
+    "statistical:20:2.0",
+    "--max-rounds",
+    "400",
+    "--reprojection-weight",
+    "4"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:04:50.752168+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..9b81136a73af602fab6b7446f9f27513b0a6ec53
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_fit_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 22.10499095916748,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid.log
new file mode 100644
index 0000000000000000000000000000000000000000..70c948776d30a3529eb2bd9b31fcf2c427998245
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid.log
@@ -0,0 +1 @@
+hybrid: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/hybrid.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..61a9f7c68c156fa5a0cf526e7e323426069aa72e
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "hybrid",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:05:20.506710+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..4c550fadb8dca417766ac4344a4d80bf7649a250
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_hybrid_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 2.3043668270111084,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs.log
new file mode 100644
index 0000000000000000000000000000000000000000..d73bdff443f7058e80e9ecbe7181378cffa456a4
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs.log
@@ -0,0 +1 @@
+inputs: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/inputs.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..79852cac58972d5aa111787222af6c08c7e53f70
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_argv.json
@@ -0,0 +1,29 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "inputs",
+    "--result-root",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--object",
+    "bike_chain",
+    "--unfitted-inputs",
+    "--fitting-bundle",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/rt3_lifts/fitting/fitting_observations.npz",
+    "--frames-bundle",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--depth-source",
+    "waft_stereo_dav2l5",
+    "--mask-erosion-px",
+    "2",
+    "--out",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:04:19.722117+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..f460f7c864b99fd08e9cfc9de3c696eb05120d8b
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_inputs_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 29.434255599975586,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest.log
new file mode 100644
index 0000000000000000000000000000000000000000..8d4d4db6b9f8c421de88e228e4c515b25fd75741
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest.log
@@ -0,0 +1 @@
+/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/manifest.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..5eb66be492a0883daea7fc8a7f4765911cc64b2e
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "manifest",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:05:34.865937+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..646afbb55ee81a832a150419a3b8f67eda86857d
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_manifest_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 1.3242645263671875,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review.log
new file mode 100644
index 0000000000000000000000000000000000000000..9a5569b7f30511431e296558d708f61dac52a56c
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review.log
@@ -0,0 +1 @@
+review: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/review.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..f5102ea4de92cccf8e1268007bc37ffe0bc32675
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "review",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:05:22.819841+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..5321f3a7a5ffcd3f408a91a1de11bcb4853a43f8
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_review_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 12.027328491210938,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube.log
new file mode 100644
index 0000000000000000000000000000000000000000..c0cbed767bb9582d2988d8b1cc581d4590d755ae
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube.log
@@ -0,0 +1 @@
+tube: ok -> /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1/steps/tube.json
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..55f0650c1eb72cbe39b8ce1d1b230f935ae36dc0
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_argv.json
@@ -0,0 +1,16 @@
+{
+  "argv": [
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/real_rgbd_appearance_spline_prior.py",
+    "tube",
+    "--run",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T22:05:12.862789+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..dcc0797447f6533c6f18fa5deac2f6ccd67e5cfc
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_prior_tube_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 4.964536905288696,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_complete.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_complete.json
new file mode 100644
index 0000000000000000000000000000000000000000..e10b3c67a16e44f0660aadb46eb72f07d0128f00
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_complete.json
@@ -0,0 +1,5 @@
+{
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "phase": "reference",
+  "completed_utc": "2026-10-28T21:42:09.178798+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..6f98773ce9c202600ab19fcc4d4a492733805018
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_argv.json
@@ -0,0 +1,33 @@
+{
+  "argv": [
+    "flock",
+    "-x",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/prepare_rt3_lifting.py",
+    "--result-root",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--selected-session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--selected-frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--tap-tracks",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks",
+    "--cotracker-tracks",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/query_tracks",
+    "--support-rule",
+    "rt_d7_depth_consistency_v1",
+    "--observations-only",
+    "--participating-hand",
+    "hand_A",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/rt3_lifts"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T21:31:30.368090+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..35358c3bba4251d092154e8d81d9b04ac0b895b7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_lift_depth_consistency_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 276.4631986618042,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks.log
new file mode 100644
index 0000000000000000000000000000000000000000..df635dd8dcbc60b27fba3e6d6546269dc6b6af33
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks.log
@@ -0,0 +1,8 @@
+12540311: tracked 5000 queries
+12933256: tracked 5000 queries
+17736741: tracked 5000 queries
+23007103: tracked 5000 queries
+12540311: members {'object': 1598, 'hand_A': 3360, 'hand_B': 0}
+12933256: members {'object': 1831, 'hand_A': 3074, 'hand_B': 0}
+17736741: members {'object': 1339, 'hand_A': 3611, 'hand_B': 0}
+23007103: members {'object': 1973, 'hand_A': 2904, 'hand_B': 0}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..a979519102124e8b9b5946eee44944fe1ecb7e7b
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_argv.json
@@ -0,0 +1,29 @@
+{
+  "argv": [
+    "flock",
+    "-x",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/phystwin_real_rgbd.py",
+    "track",
+    "--session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/native_inputs/per_camera_inputs.json",
+    "--checkpoint",
+    "/mnt/data0/magna/.cache/torch/hub/checkpoints/scaled_online.pth",
+    "--seed",
+    "20260923",
+    "--device",
+    "cuda",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/query_tracks"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T21:20:55.770237+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..2672360cf7aa1c373d32f5aab3bd9d467804102f
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_query_tracks_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 327.64011883735657,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review.log
new file mode 100644
index 0000000000000000000000000000000000000000..44a565d8129f2ec409d3583bb8557acbd3c94ef1
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review.log
@@ -0,0 +1,49 @@
+12540311: diagnostics 0/405
+12540311: diagnostics 100/405
+12540311: diagnostics 200/405
+12540311: diagnostics 300/405
+12540311: diagnostics 400/405
+12540311: diagnostics 405/405
+12540311: visual frame 0/405
+12540311: visual frame 100/405
+12540311: visual frame 200/405
+12540311: visual frame 300/405
+12540311: visual frame 400/405
+12540311: visual frame 405/405
+12933256: diagnostics 0/405
+12933256: diagnostics 100/405
+12933256: diagnostics 200/405
+12933256: diagnostics 300/405
+12933256: diagnostics 400/405
+12933256: diagnostics 405/405
+12933256: visual frame 0/405
+12933256: visual frame 100/405
+12933256: visual frame 200/405
+12933256: visual frame 300/405
+12933256: visual frame 400/405
+12933256: visual frame 405/405
+17736741: diagnostics 0/405
+17736741: diagnostics 100/405
+17736741: diagnostics 200/405
+17736741: diagnostics 300/405
+17736741: diagnostics 400/405
+17736741: diagnostics 405/405
+17736741: visual frame 0/405
+17736741: visual frame 100/405
+17736741: visual frame 200/405
+17736741: visual frame 300/405
+17736741: visual frame 400/405
+17736741: visual frame 405/405
+23007103: diagnostics 0/406
+23007103: diagnostics 100/406
+23007103: diagnostics 200/406
+23007103: diagnostics 300/406
+23007103: diagnostics 400/406
+23007103: diagnostics 406/406
+23007103: visual frame 0/406
+23007103: visual frame 100/406
+23007103: visual frame 200/406
+23007103: visual frame 300/406
+23007103: visual frame 400/406
+23007103: visual frame 406/406
+{"views": {"12540311": {"T": 406, "queries": 5000, "keyframes": [0, 50, 101, 151, 160, 202, 253, 293, 303, 354, 405], "source_seed_counts": {"object": 1602, "hand_A": 3398, "hand_B": 0}, "tap_visible_fraction_min_mean_max": [0.6356, 0.7556093596059114, 1.0], "accepted_object_min_max": [0, 1177], "paired_common_min_max": [3195, 4564]}, "12933256": {"T": 406, "queries": 5000, "keyframes": [0, 50, 101, 151, 202, 214, 253, 303, 354, 405], "source_seed_counts": {"object": 1867, "hand_A": 3133, "hand_B": 0}, "tap_visible_fraction_min_mean_max": [0.6628, 0.9445802955665025, 1.0], "accepted_object_min_max": [1159, 1267], "paired_common_min_max": [1777, 4374]}, "17736741": {"T": 406, "queries": 5000, "keyframes": [0, 50, 101, 119, 121, 151, 202, 253, 303, 354, 405], "source_seed_counts": {"object": 1348, "hand_A": 3652, "hand_B": 0}, "tap_visible_fraction_min_mean_max": [0.8368, 0.953579802955665, 1.0], "accepted_object_min_max": [386, 897], "paired_common_min_max": [3762, 4527]}, "23007103": {"T": 407, "queries": 5000, "keyframes": [0, 50, 101, 152, 203, 217, 253, 304, 355, 406], "source_seed_counts": {"object": 1992, "hand_A": 3008, "hand_B": 0}, "tap_visible_fraction_min_mean_max": [0.8322, 0.9410083538083538, 1.0], "accepted_object_min_max": [911, 1267], "paired_common_min_max": [2618, 4179]}}, "complete": true, "observations_only": true, "controller_pool": {"controller_gate_m": 0.04, "participating_hands": ["hand_A"], "survivors": {"12540311": {"hand_A": 1749, "hand_B": 0}, "12933256": {"hand_A": 739, "hand_B": 0}, "17736741": {"hand_A": 1766, "hand_B": 0}, "23007103": {"hand_A": 1044, "hand_B": 0}}, "pool": {"12540311": {"hand_A": 0, "hand_B": 0}, "12933256": {"hand_A": 64, "hand_B": 0}, "17736741": {"hand_A": 20, "hand_B": 0}, "23007103": {"hand_A": 121, "hand_B": 0}}, "survivor_distance_to_belt_mm_p5_p50_p95": {"hand_A": [43.4, 104.3, 173.7]}, "selected_distance_to_belt_mm_max": 39.95138212593075}, "participating_hands": ["hand_A"], "derived_samples": 406, "observed_object_points": 1113, "controls": 30, "controller_initial_distance_m_max": 0.03995138212593075, "selected_controls_per_hand": {"hand_A": 30}}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..7b459cc5b9b1eb49e20fdbbdc76497b7dd3e67bd
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_argv.json
@@ -0,0 +1,28 @@
+{
+  "argv": [
+    "flock",
+    "-x",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/agent_drafts/accepted_batch_recipe/review_reference_case.py",
+    "--case",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--prep",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1",
+    "--participating-hand",
+    "hand_A",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T21:36:06.902893+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..84ec6ac6e05857fe7fdb9fcd20889d61362832dc
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_review_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 362.2325701713562,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks.log b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks.log
new file mode 100644
index 0000000000000000000000000000000000000000..f8e42ccd34e533b922e848d511cb0f071fc82a63
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks.log
@@ -0,0 +1,6 @@
+/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/.venv/lib/python3.11/site-packages/torch/functional.py:505: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4381.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+12540311: saved 406 native frames and 5000 joint queries to /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks/tracks_12540311.npz
+12933256: saved 406 native frames and 5000 joint queries to /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks/tracks_12933256.npz
+17736741: saved 406 native frames and 5000 joint queries to /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks/tracks_17736741.npz
+23007103: saved 407 native frames and 5000 joint queries to /mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks/tracks_23007103.npz
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..8551b568f8aa4fa69049510158fed7842883e3ee
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_argv.json
@@ -0,0 +1,30 @@
+{
+  "argv": [
+    "flock",
+    "-x",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+    "uv",
+    "run",
+    "--no-sync",
+    "python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/run_rt2_tapnextpp_tracks.py",
+    "--result-root",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--selected-session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--selected-frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--saved-query-tracks",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/query_tracks",
+    "--checkpoint",
+    "/mnt/data2/magna/belt_perception/results/phystwin_real_rgbd/session_20260921_225630_786148/rt_waft_tapnextpp_refit_01/checkpoints/tapnextpp_512.ckpt",
+    "--device",
+    "cuda",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks"
+  ],
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+  "owner_acceptance": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_mask_acceptance_c6a_selected_20261028.json",
+  "started_utc": "2026-10-28T21:26:24.252705+00:00"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..541c9836f621201681d5fec141c900ac1cde04c1
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/batch_reference_tapnextpp_tracks_result.json
@@ -0,0 +1,5 @@
+{
+  "returncode": 0,
+  "wall_seconds": 305.98933458328247,
+  "code": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/chain_prior_revocation_preparation_state_20261028.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/chain_prior_revocation_preparation_state_20261028.json
new file mode 100644
index 0000000000000000000000000000000000000000..3c6e18153b038fa5d96d641892854eedb6730f34
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/chain_prior_revocation_preparation_state_20261028.json
@@ -0,0 +1,32 @@
+{
+  "owner_revocation": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/owner_prior_revocation_20261028.json",
+  "stage_snapshot": {
+    "appearance_check_v1": {
+      "returncode": 0,
+      "elapsed_s": 52.67811157100368
+    },
+    "appearance_train_v1": {
+      "state": "started or queued; no completed exit record; held after revocation"
+    },
+    "appearance_review_v1": {
+      "state": "not started"
+    },
+    "mech_v2": {
+      "returncode": 0,
+      "elapsed_s": 129.46192697600054
+    },
+    "mech_v3": {
+      "state": "started or queued; no completed exit record; held after revocation"
+    },
+    "mech_center_v3": {
+      "state": "not started"
+    },
+    "mech_supervision": {
+      "state": "not started"
+    },
+    "mech_center_supervision": {
+      "state": "not started"
+    }
+  },
+  "completed_and_partial_outputs": "preserved without deletion or retry"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12540311/inspection/frames.npz b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12540311/inspection/frames.npz
new file mode 100644
index 0000000000000000000000000000000000000000..828bda499cae07bb9e8a081a41625a82289e44df
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12540311/inspection/frames.npz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3596e6b3c0a6de2203154ec01cb1cdf1ce09021984be1c6a0772f9bd7eaad0e6
+size 100988
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12933256/inspection/frames.npz b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12933256/inspection/frames.npz
new file mode 100644
index 0000000000000000000000000000000000000000..f180acc71cb951cbaf9d1e9d075f755480f30fb2
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/12933256/inspection/frames.npz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36045b98d67d907627df66018bf594592bc0dd5bf42864172da62daab0573614
+size 100988
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/17736741/inspection/frames.npz b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/17736741/inspection/frames.npz
new file mode 100644
index 0000000000000000000000000000000000000000..67282bcf24d4473458f42e26cd019733e4b68f03
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/17736741/inspection/frames.npz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e0ac4e8b99751cc361320b6c86d34074e5912036a0dba7c1c96bcc7fba3c712
+size 100988
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/23007103/inspection/frames.npz b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/23007103/inspection/frames.npz
new file mode 100644
index 0000000000000000000000000000000000000000..0370314b1c4002fa4a7c26e4f7fff4fafeab7e8c
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/cameras/23007103/inspection/frames.npz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0737a2adf4a697588ad6b8f2d3fac514f680bd919257a3c0640fee312f6b3d05
+size 101200
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/per_camera_inputs.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/per_camera_inputs.json
new file mode 100644
index 0000000000000000000000000000000000000000..fbba79d874913ac7d17115886028d9af7a24f64d
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/frames_waft/per_camera_inputs.json
@@ -0,0 +1,33 @@
+{
+  "annotation_ns": [
+    1790047938700151856,
+    1790047952247381484
+  ],
+  "anchor_view": 0,
+  "cameras": [
+    {
+      "view_index": 0,
+      "serial": "12540311",
+      "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/cameras/12540311/inspection/frames.npz",
+      "result_root": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/native_inputs/cameras/12540311"
+    },
+    {
+      "view_index": 1,
+      "serial": "12933256",
+      "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/cameras/12933256/inspection/frames.npz",
+      "result_root": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/native_inputs/cameras/12933256"
+    },
+    {
+      "view_index": 2,
+      "serial": "17736741",
+      "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/cameras/17736741/inspection/frames.npz",
+      "result_root": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/native_inputs/cameras/17736741"
+    },
+    {
+      "view_index": 3,
+      "serial": "23007103",
+      "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/cameras/23007103/inspection/frames.npz",
+      "result_root": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/native_inputs/cameras/23007103"
+    }
+  ]
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..11688f67affb8cf079d1dbb10f4d626e8ef00aff
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_argv.json
@@ -0,0 +1,34 @@
+{
+  "argv": [
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/.venv/bin/python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/prepare_rt3_lifting.py",
+    "--result-root",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--selected-session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--selected-frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--tap-tracks",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks",
+    "--retained-lifts",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/rt3_lifts/lifted",
+    "--accepted-mesh-sha256",
+    "6684bd8b1a51a67bc36b716fdd41cc0ad017a5b38be32c22936b4d2df016194f",
+    "--accepted-prior-manifest-sha256",
+    "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "--participating-hand",
+    "hand_A",
+    "--mechanical-prior",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "--mesh-rules",
+    "m2_10mm_centerline_core_v2",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/mechanics_v2"
+  ],
+  "environment": {
+    "CUDA_VISIBLE_DEVICES": "",
+    "PYTHONDONTWRITEBYTECODE": "1"
+  },
+  "resource_lock": "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_result.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_result.json
new file mode 100644
index 0000000000000000000000000000000000000000..905d8f8967f09baa271639ae4ee26e77f0603154
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v2_result.json
@@ -0,0 +1,4 @@
+{
+  "returncode": 0,
+  "elapsed_s": 129.46192697600054
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v3_argv.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v3_argv.json
new file mode 100644
index 0000000000000000000000000000000000000000..bb2f0d127ec8b6fc54b61b91eeeb18d1bd61996d
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/mech_v3_argv.json
@@ -0,0 +1,36 @@
+{
+  "argv": [
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/.venv/bin/python",
+    "-B",
+    "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/scripts/prepare_rt3_lifting.py",
+    "--result-root",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "--selected-session",
+    "/mnt/data2/magna/belt_perception/converted_datasets/sessions/session_20260921_233144_367389",
+    "--selected-frames",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json",
+    "--tap-tracks",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/tapnextpp_tracks",
+    "--retained-lifts",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/rt3_lifts/lifted",
+    "--accepted-mesh-sha256",
+    "6684bd8b1a51a67bc36b716fdd41cc0ad017a5b38be32c22936b4d2df016194f",
+    "--accepted-prior-manifest-sha256",
+    "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "--participating-hand",
+    "hand_A",
+    "--mechanical-prior",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "--mesh-rules",
+    "m2_10mm_centerline_voxels_v3",
+    "--retained-samples-binding",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/mechanics_v2/mesh/binding.json",
+    "--output",
+    "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/mechanics_v3"
+  ],
+  "environment": {
+    "CUDA_VISIBLE_DEVICES": "",
+    "PYTHONDONTWRITEBYTECODE": "1"
+  },
+  "resource_lock": "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/AGENT_FINDINGS.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/AGENT_FINDINGS.md
new file mode 100644
index 0000000000000000000000000000000000000000..b77cfd41286057e3a413c73f7dc670d54f92b0f6
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/AGENT_FINDINGS.md
@@ -0,0 +1,44 @@
+# 233144 — lower curvature candidate review
+
+The completed candidate passes the saved numerical, input-preservation and geometry checks. **21 panels and nine orbit samples were directly inspected. Owner acceptance remains pending; previous chain-prior acceptances remain revoked.** This review grants no mechanics, appearance-training or physical-fitting authorization.
+
+Goal: review this full initial prior and compare it with the preserved 18-control baseline. Done when: settings/input/numeric checks and the actual visual inspection are complete and saved. Non-goals: refitting, changing observations or recording approval.
+
+The requested settings are **18 controls, curvature weight 5e-10, reprojection weight 4**. All other scientific settings exactly match the baseline, including the solver, input cloud/masks, 4.75 mm radius and tube sampling. All six preparation stages and the single packaging/check invocation returned zero. All 27 true observation/filter/step files remain byte-equal; original baseline fit/manifest and this saved fit remain unchanged. The derived Gaussian shape_prior.glb was newly generated and equals this candidate's tube GLB.
+
+[Numerical checks](numeric_prior_review.json) · [Exact visual record](agent_visual_review.json) · [Metrics and camera links](PRIOR_REVIEW_c6a_v1.md) · [Common comparison values](baseline_comparison.json)
+
+| Common diagnostic | Baseline: curvature 5e-9 / reprojection 4 | This candidate |
+|---|---:|---:|
+| Mean point-to-centerline distance (mm) | 7.389500 | 7.388422 |
+| Mean of camera mean reprojection distances (mm) | 3.676484 | 3.676523 |
+| Length (mm) | 587.702 | 587.697 |
+| Distance outside soft length bounds (mm) | 2.702 | 2.697 |
+| Maximum curvature (/m) | 37.928 | 38.438 |
+| Median curvature (/m) | 9.860 | 9.838 |
+| Outer rounds | 1 | 1 |
+| Final gradient norm | 9.39435e-05 | 8.99446e-05 |
+
+Against the preserved M18 baseline, the visible silhouettes are effectively unchanged at the saved image scale. Lowering the curvature weight from 5e-9 to 5e-10 slightly increases maximum curvature from 37.928/m to 38.438/m, but does not produce a clear visible corner correction or resolve the hidden grip arc.
+
+The common 3D mean decreases only from 7.389500 to 7.388422 mm, while the mean of camera reprojection means is essentially unchanged, 3.676484 to 3.676523 mm. Length remains 587.697 mm, about 2.697 mm above the soft upper bound. These changes are too small to support a substantive quality improvement claim from this comparison.
+
+The total objectives are **not directly comparable** after a coefficient changes. Common residuals are diagnostic, not ground-truth error; the stored comparison also gives reprojection loss before its weight. All runs retain the absolute 1e-4 gradient tolerance, which can change effective precision when a term is reweighted. This candidate stopped on `gradient_tolerance`; one outer round permits up to 25 L-BFGS updates, whose actual count is not saved. The 400-round value is only a budget. The single-round convergence line plot appears blank because one point is plotted without a marker; the saved round record equals the final result. [Saved rounds](../prior_chain_lower_curvature_v2/spline/rounds.jsonl).
+
+The four input panels show the same rounded, approximately quadrilateral chain loop on the table with the hand covering part of the grip arc. Object masks retain the visible loop and exclude the occluding hand; no conspicuous initial invalid-depth region is visible. Mask continuity does not resolve the geometry underneath the hand.
+
+All four masked centerlines and unmasked tube silhouettes follow the broad visible loop. The exposed corners remain rounded and small edge offsets remain, especially near the upper-left turn in camera 17736741 and the left/bottom arc in camera 12933256. The unmasked red tube crosses the hand in views 0 and 2 because it displays the complete inferred loop; this is not evidence of a physically observed path or attachment.
+
+The scale, 3D-fit and hybrid panels show the tube within the broad observed loop envelope, but the observed cloud is wider and vertically dispersed. The mesh seeds fill the hidden arc and sit lower than many observations. The depth diagnostic remains roughly 9–12 mm above the local table across views, and views 2 and 3 retain larger centerline residuals. No depth or contact correction is implied.
+
+Visibility/fill views show median-filled vertices concentrated along the underside and the occluded arc; 575 of 1,728 vertices are filled. Curvature peaks near 38.4/m. The single-round line plots are blank because one sample is plotted without a point marker; the numerical round record is available. Nine individual orbit samples show one closed rounded loop without an obvious fold or crossing; axis-label overlap limits their fine vertical readability, so the scale/profile plots supply that evidence.
+
+Saved geometry checks report **144 rings, 1,728 vertices, 3,456 triangles**, one consistently wound watertight component and **zero confirmed surface intersections**. Maximum recomputed radial error is **8.15e-09 m**. Minimum nonadjacent ring-center distance is **6.670 mm**; this remains a diagnostic, not a substitute for the surface-intersection rule. Tube z spans **0.044 to 17.253 mm** relative to nominal table z=0. No table snapping, depth correction or contact validation was performed.
+
+Colors are observed for **1153 vertices** and median-filled for **575**. Saved GLB roundtrip checks report exact float32 positions and colors, and colored-GLB hashes match current files. Hybrid initialization contains **26,339 observed + 103,680 mesh-derived = 130,019 seeds**. These populations match the baseline counts because all runs use 18 controls and the same observation bank/sampling settings. They are appearance seeds, not mechanical particles or mass. Seven checked NPZ families are finite; per-view residual quantiles recompute to their recorded values.
+
+Orbit frames **0, 4, 9, 13, 18, 22, 26, 31, 35 of 36** were inspected individually. [Rotation](../prior_chain_lower_curvature_v2/review/tube/orbit.mp4) · [Inspected orbit frames](orbit_frames.json). The 21 panels comprise four input/mask sets, four masked centerlines, four unmasked silhouettes and nine diagnostic panels. The smooth tube does not model individual chain links or prove the shape behind the hand.
+
+This is a hand_A-only sequence. Camera 12540311 has zero admitted object support at native-relative rows 143–238, 240, 243 and 262–326; 116 undefined object-mask coverage values remain NaNs. Finite coverage has median 0.479 and maximum 0.900. The derived middle/lowering views retain isolated points below or beside the main loop. Paired native-minus-WAFT medians are 0/0/−8/+2 mm and per-frame p95 maxima are 27/27/33/28 mm in camera order. These are common-support diagnostics, not ground-truth errors. Initial-prior changes do not clear later tracking or physical fitting. [Preserved temporal review](../review_v1/REVIEW.md).
+
+Only this new packet and scoped review scratch were written by the reviewer. No prior has been accepted by this review.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/PRIOR_REVIEW_c6a_v1.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/PRIOR_REVIEW_c6a_v1.md
new file mode 100644
index 0000000000000000000000000000000000000000..6ea8ef11af1b7a6b8c98a614d8ae28189921f0f8
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/PRIOR_REVIEW_c6a_v1.md
@@ -0,0 +1,39 @@
+# session_20260921_233144_367389 — prior approval pending
+
+Saved C2 **bike_chain** prior evidence (radius 4.75 mm). Agent visual review is complete; owner acceptance remains pending, and earlier chain acceptances remain revoked. See [findings and baseline comparison](AGENT_FINDINGS.md). Saved values and media are linked here; the agent assessment is recorded separately. This packet records no acceptance.
+
+Initial reference frames are the saved nearest-camera observations of the first 30 Hz sample. The local row, converted frame, native-container frame and capture clock remain distinct in [the metrics JSON](<prior_metrics.json>).
+
+| Camera | Local / converted / native | Centerline / geometry / native input |
+|---|---|---|
+| 12540311 | 1 / 269 / 270 | [centerline](<../prior_chain_lower_curvature_v2/review/spline/centerline_0_with_masks.png>) · [tube](<../prior_chain_lower_curvature_v2/review/tube/silhouette_0_unmasked.png>) · [geometry + mask](<../prior_chain_lower_curvature_v2/review/tube/silhouette_0_with_masks.png>) · [RGB/depth](<../prior_chain_lower_curvature_v2/review/inputs/view_0_12540311.png>) |
+| 12933256 | 0 / 268 / 269 | [centerline](<../prior_chain_lower_curvature_v2/review/spline/centerline_1_with_masks.png>) · [tube](<../prior_chain_lower_curvature_v2/review/tube/silhouette_1_unmasked.png>) · [geometry + mask](<../prior_chain_lower_curvature_v2/review/tube/silhouette_1_with_masks.png>) · [RGB/depth](<../prior_chain_lower_curvature_v2/review/inputs/view_1_12933256.png>) |
+| 17736741 | 0 / 268 / 269 | [centerline](<../prior_chain_lower_curvature_v2/review/spline/centerline_2_with_masks.png>) · [tube](<../prior_chain_lower_curvature_v2/review/tube/silhouette_2_unmasked.png>) · [geometry + mask](<../prior_chain_lower_curvature_v2/review/tube/silhouette_2_with_masks.png>) · [RGB/depth](<../prior_chain_lower_curvature_v2/review/inputs/view_2_17736741.png>) |
+| 23007103 | 1 / 268 / 269 | [centerline](<../prior_chain_lower_curvature_v2/review/spline/centerline_3_with_masks.png>) · [tube](<../prior_chain_lower_curvature_v2/review/tube/silhouette_3_unmasked.png>) · [geometry + mask](<../prior_chain_lower_curvature_v2/review/tube/silhouette_3_with_masks.png>) · [RGB/depth](<../prior_chain_lower_curvature_v2/review/inputs/view_3_23007103.png>) |
+
+[Full rotation video](<../prior_chain_lower_curvature_v2/review/tube/orbit.mp4>) · [Geometry and depth cloud](<../prior_chain_lower_curvature_v2/review/tube/tube_scale_views.png>) · [Residual histograms](<../prior_chain_lower_curvature_v2/review/spline/residual_histograms.png>) · [Depth offset diagnostic](<../prior_chain_lower_curvature_v2/review/inputs/depth_offset.png>)
+
+Saved fit: **1 stage-1 rounds**, stop `gradient_tolerance`; final length **587.697 mm**, maximum curvature **38.438/m**, final gradient norm **8.99446e-05**. Soft length bounds are 583.000–585.000 mm; distance outside those bounds is 2.697 mm.
+
+Saved geometry checks report `passed=True`: **1728 mesh vertices / 3456 triangles**, 1 component(s), `watertight=True`, `self_intersecting=False`. [Full geometry checks](<../prior_chain_lower_curvature_v2/tube/tube_checks.json>).
+
+The following 3D residuals are lifted-depth point distances to the **centerline**, not measured reconstruction errors. The centerline-in-mask fraction is a separate saved diagnostic; it is not silhouette IoU or a visual quality decision.
+
+| Camera | Median / p90 3D residual (mm) | Visible centerline inside mask | Native depth and visibility evidence |
+|---|---|---|---|
+| 12540311 | 6.862 / 10.056 | 99.01% | [Full native video](<../review_v1/12540311/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12540311/12540311_visibility_certainty.png>) · [Support/rejections](<../review_v1/12540311/support_and_rejections.png>) · [Paired depth](<../review_v1/12540311/paired_depth.png>) |
+| 12933256 | 5.855 / 10.509 | 96.66% | [Full native video](<../review_v1/12933256/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12933256/12933256_visibility_certainty.png>) · [Support/rejections](<../review_v1/12933256/support_and_rejections.png>) · [Paired depth](<../review_v1/12933256/paired_depth.png>) |
+| 17736741 | 8.463 / 11.138 | 98.41% | [Full native video](<../review_v1/17736741/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/17736741/17736741_visibility_certainty.png>) · [Support/rejections](<../review_v1/17736741/support_and_rejections.png>) · [Paired depth](<../review_v1/17736741/paired_depth.png>) |
+| 23007103 | 9.288 / 12.152 | 98.60% | [Full native video](<../review_v1/23007103/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/23007103/23007103_visibility_certainty.png>) · [Support/rejections](<../review_v1/23007103/support_and_rejections.png>) · [Paired depth](<../review_v1/23007103/paired_depth.png>) |
+
+The saved statistical filter used 20 neighbours / 2 sigma and kept **26339 of 27239** eroded points. The run used `waft_stereo_dav2l5`, 2 px erosion, a 400-round budget and reprojection weight **4**.
+
+Appearance initialization contains **26339 observed + 103680 mesh-derived = 130019 seeds**. Vertex colors are observed for 1153 mesh vertices and median-filled for 575. Mesh vertices and appearance seeds are not mechanical particles. No post-prior mechanical population or unit-mass total is supplied by this prior run; trained appearance and physical fitting are separate stages.
+
+[Appearance seed populations](<../prior_chain_lower_curvature_v2/review/hybrid/combined.png>) · [Observed and filled colors](<../prior_chain_lower_curvature_v2/review/colors/visibility_and_fill.png>) · [30 Hz observation coverage](<../review_v1/derived/derived_coverage.png>)
+
+[Full saved 30 Hz observations](<../review_v1/derived/derived_30hz_xyz.mp4>)
+
+[Manifest](<../prior_chain_lower_curvature_v2/manifest.json>) · [Fit settings and metrics](<../prior_chain_lower_curvature_v2/steps/fit.json>) · [Reference observations](<../prior_chain_lower_curvature_v2/inputs/reference_views.json>) · [Tracking/lift summary](<../review_v1/summary.json>)
+
+The agent inspected 21 panels and 9 orbit frames; see [the exact visual record](agent_visual_review.json) and [findings](AGENT_FINDINGS.md). Owner prior approval remains pending. No case status, prior acceptance, mechanics approval or trained-appearance approval is changed here.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/agent_visual_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/agent_visual_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..28d5278ff4eef07eb8d730204c96f1cb944fed93
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/agent_visual_review.json
@@ -0,0 +1,159 @@
+{
+  "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+  "candidate": "lower_curvature",
+  "prior_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2",
+  "reviewed_at_utc": "2026-10-28T22:55:44.691553+00:00",
+  "review_kind": "bounded numeric and visual review of completed owner-requested chain weight candidate",
+  "readiness": "ready_for_owner_comparison_with_disclosed_limits",
+  "blocking_findings_for_review_completion": [],
+  "owner_prior_acceptance": "pending; all earlier chain prior acceptances remain revoked",
+  "scope": {
+    "goal": "Review the completed candidate and compare common diagnostics against the preserved 18-control baseline.",
+    "done_when": "Six successful stages, single scoped packaging/check invocation,21 actual panels and 9 orbit samples inspected, numerical/settings/input checks and findings saved.",
+    "non_goals": [
+      "No scientific execution, fitting or source changes",
+      "No global or approval changes",
+      "No mechanics, appearance training or physical-fitting approval"
+    ]
+  },
+  "job": {
+    "case_id": "233144",
+    "session_id": "session_20260921_233144_367389",
+    "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "prep": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1",
+    "candidate": "lower_curvature",
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2",
+    "review": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_chain_lower_curvature_v2",
+    "baseline": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "settings": {
+      "control_count": 18,
+      "lambda_curvature_cost": 5e-10,
+      "reprojection_weight": 4.0
+    },
+    "baseline_manifest_sha256": "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "baseline_fit_sha256": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455"
+  },
+  "packaging_and_check_result": {
+    "returncode": 0,
+    "elapsed_s": 11.915011679011513,
+    "completed_utc": "2026-10-28T22:51:00.205820+00:00",
+    "argv": [
+      "flock",
+      "-x",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/.venv/bin/python",
+      "-B",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/agent_drafts/chain_weight_pair_review_233045_233144_20261028/package_and_check.py",
+      "--case",
+      "233144",
+      "--candidate",
+      "lower_curvature"
+    ]
+  },
+  "numeric_review_path": "numeric_prior_review.json",
+  "baseline_comparison_path": "baseline_comparison.json",
+  "unchanged_input_count": 27,
+  "geometry_checks": {
+    "euler_characteristic": 0,
+    "watertight": true,
+    "components": 1,
+    "winding_consistent": true,
+    "signed_volume_m3": 3.9760913468027164e-05,
+    "minimum_outward_normal_dot_m": 0.004579301530990692,
+    "degenerate_triangles": 0,
+    "self_intersecting": false,
+    "maximum_radius_error_m": 8.147490698599824e-09,
+    "maximum_diameter_error_m": 1.604742104130419e-08,
+    "bishop_seam_error": 1.1920928955078125e-07,
+    "minimum_nonadjacent_centerline_distance_m": 0.006669520568941395,
+    "triangle_area_m2": {
+      "min": 3.3687073323917815e-06,
+      "median": 4.9904552902008765e-06,
+      "mean": 5.015839394544405e-06,
+      "max": 6.25396980998803e-06,
+      "density_max_over_min": 1.8564895055895856
+    },
+    "bounds_m": [
+      [
+        -0.0369243286550045,
+        -0.1740739345550537,
+        4.4061802327632904e-05
+      ],
+      [
+        0.17170047760009766,
+        -0.0010024229995906353,
+        0.017253154888749123
+      ]
+    ],
+    "radius_m": 0.00475,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "overlap_rule": "surface_intersection",
+    "open3d_reported_intersecting_pairs": 0,
+    "confirmed_intersecting_pairs": [],
+    "passed": true,
+    "orientation_flipped": true
+  },
+  "glb_roundtrip_checks": {
+    "float32_positions_exact": true,
+    "vertex_colors_exact": true,
+    "uv_count": 0,
+    "vertex_reordered_on_open3d_load": true
+  },
+  "observed_counts_unchanged": true,
+  "inspected_panels_relative_to_run": [
+    "review/inputs/view_0_12540311.png",
+    "review/inputs/view_1_12933256.png",
+    "review/inputs/view_2_17736741.png",
+    "review/inputs/view_3_23007103.png",
+    "review/spline/centerline_0_with_masks.png",
+    "review/spline/centerline_1_with_masks.png",
+    "review/spline/centerline_2_with_masks.png",
+    "review/spline/centerline_3_with_masks.png",
+    "review/tube/silhouette_0_unmasked.png",
+    "review/tube/silhouette_1_unmasked.png",
+    "review/tube/silhouette_2_unmasked.png",
+    "review/tube/silhouette_3_unmasked.png",
+    "review/tube/tube_scale_views.png",
+    "review/spline/residual_histograms.png",
+    "review/inputs/depth_offset.png",
+    "review/colors/visibility_and_fill.png",
+    "review/hybrid/combined.png",
+    "review/spline/fit_3d.png",
+    "review/spline/curvature_profile.png",
+    "review/spline/round_curves.png",
+    "review/tube/tube_profiles.png"
+  ],
+  "inspected_orbit_frames_relative_to_review": [
+    "orbit_frames/frame_000.png",
+    "orbit_frames/frame_004.png",
+    "orbit_frames/frame_009.png",
+    "orbit_frames/frame_013.png",
+    "orbit_frames/frame_018.png",
+    "orbit_frames/frame_022.png",
+    "orbit_frames/frame_026.png",
+    "orbit_frames/frame_031.png",
+    "orbit_frames/frame_035.png"
+  ],
+  "inspection_method": "Direct individual image reads of all 21 panels and nine sampled orbit frames; four baseline silhouettes were also inspected for comparison.",
+  "blocking_findings": [],
+  "observations": [
+    "The four input panels show the same rounded, approximately quadrilateral chain loop on the table with the hand covering part of the grip arc. Object masks retain the visible loop and exclude the occluding hand; no conspicuous initial invalid-depth region is visible. Mask continuity does not resolve the geometry underneath the hand.",
+    "All four masked centerlines and unmasked tube silhouettes follow the broad visible loop. The exposed corners remain rounded and small edge offsets remain, especially near the upper-left turn in camera 17736741 and the left/bottom arc in camera 12933256. The unmasked red tube crosses the hand in views 0 and 2 because it displays the complete inferred loop; this is not evidence of a physically observed path or attachment.",
+    "The scale, 3D-fit and hybrid panels show the tube within the broad observed loop envelope, but the observed cloud is wider and vertically dispersed. The mesh seeds fill the hidden arc and sit lower than many observations. The depth diagnostic remains roughly 9\u201312 mm above the local table across views, and views 2 and 3 retain larger centerline residuals. No depth or contact correction is implied.",
+    "Visibility/fill views show median-filled vertices concentrated along the underside and the occluded arc; 575 of 1,728 vertices are filled. Curvature peaks near 38.4/m. The single-round line plots are blank because one sample is plotted without a point marker; the numerical round record is available. Nine individual orbit samples show one closed rounded loop without an obvious fold or crossing; axis-label overlap limits their fine vertical readability, so the scale/profile plots supply that evidence."
+  ],
+  "comparison_observations": [
+    "Against the preserved M18 baseline, the visible silhouettes are effectively unchanged at the saved image scale. Lowering the curvature weight from 5e-9 to 5e-10 slightly increases maximum curvature from 37.928/m to 38.438/m, but does not produce a clear visible corner correction or resolve the hidden grip arc.",
+    "The common 3D mean decreases only from 7.389500 to 7.388422 mm, while the mean of camera reprojection means is essentially unchanged, 3.676484 to 3.676523 mm. Length remains 587.697 mm, about 2.697 mm above the soft upper bound. These changes are too small to support a substantive quality improvement claim from this comparison."
+  ],
+  "limitations": [
+    "The 4.75 mm radius continuous circular tube does not model individual links, gaps, rollers, plates or articulation; hidden geometry is inferred.",
+    "Length 587.697268 mm is 2.697289 mm outside the soft interval. The 10/m threshold and 80/m smooth saturation are penalty parameters, not hard curvature bounds.",
+    "Tube z bounds are 0.044062 to 17.253155 mm relative to nominal table z=0. Depth/table offsets and contact uncertainty remain unchanged.",
+    "A lower common residual does not establish ground-truth shape accuracy. Changed objective totals are not ranked; fixed absolute gradient tolerance can affect effective precision when weights change.",
+    "This is a hand_A-only sequence. Camera 12540311 has zero admitted object support at native-relative rows 143\u2013238, 240, 243 and 262\u2013326; 116 undefined object-mask coverage values remain NaNs. Finite coverage has median 0.479 and maximum 0.900. The derived middle/lowering views retain isolated points below or beside the main loop. Paired native-minus-WAFT medians are 0/0/\u22128/+2 mm and per-frame p95 maxima are 27/27/33/28 mm in camera order. These are common-support diagnostics, not ground-truth errors. Initial-prior changes do not clear later tracking or physical fitting.",
+    "Inspection covers 21 saved panels and nine orbit frames, not every video frame. No mechanics, appearance training or physical fit is assessed or accepted."
+  ],
+  "approval_effect": "No approval is recorded; execution authorization is not acceptance."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/baseline_comparison.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/baseline_comparison.json
new file mode 100644
index 0000000000000000000000000000000000000000..1eb796e799f70cf6994a8853854d1e3ac5340bff
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/baseline_comparison.json
@@ -0,0 +1,47 @@
+{
+  "baseline": {
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "control_count": 18,
+    "curvature_weight": 5e-09,
+    "reprojection_weight": 4.0,
+    "mean_point_to_centerline_mm": 7.3894998910938074,
+    "mean_of_camera_mean_reprojection_mm": 3.6764842167760454,
+    "length_mm": 587.7017974853516,
+    "length_outside_soft_bounds_mm": 2.7018189430236816,
+    "maximum_curvature_per_m": 37.92765808105469,
+    "median_curvature_per_m": 9.860498428344727,
+    "common_3d_loss": 2.406455132586416e-05,
+    "unweighted_reprojection_loss": 8.715408512216527e-06,
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "gradient_norm": 9.39435267355293e-05,
+    "wall_time_s": 21.2205913066864
+  },
+  "candidate": {
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2",
+    "control_count": 18,
+    "curvature_weight": 5e-10,
+    "reprojection_weight": 4.0,
+    "mean_point_to_centerline_mm": 7.388422403286789,
+    "mean_of_camera_mean_reprojection_mm": 3.6765234728730625,
+    "length_mm": 587.6972675323486,
+    "length_outside_soft_bounds_mm": 2.697288990020752,
+    "maximum_curvature_per_m": 38.438140869140625,
+    "median_curvature_per_m": 9.838243007659912,
+    "common_3d_loss": 2.4059820134425536e-05,
+    "unweighted_reprojection_loss": 8.715906005818397e-06,
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "gradient_norm": 8.994463860290125e-05,
+    "wall_time_s": 20.75391411781311
+  },
+  "same_observed_inputs": true,
+  "only_scientific_setting_change": {
+    "control_count": 18,
+    "lambda_curvature_cost": 5e-10,
+    "reprojection_weight": 4.0
+  },
+  "all_other_baseline_settings_match": true,
+  "total_objectives_comparable": false,
+  "interpretation": "Common raw residuals and unweighted reprojection losses are compared. The coefficient changes the objective; absolute gradient stopping also changes effective precision. No optimum or ground-truth accuracy claim."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/numeric_prior_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/numeric_prior_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..2dd548297f3cdb9a2f7d892ece228f8a1dbc608c
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/numeric_prior_review.json
@@ -0,0 +1,824 @@
+{
+  "scope": "Saved 233144 lower_curvature prior review only; no fitting, sampling or scientific correction executed by this reviewer.",
+  "job": {
+    "case_id": "233144",
+    "session_id": "session_20260921_233144_367389",
+    "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "prep": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1",
+    "candidate": "lower_curvature",
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2",
+    "review": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_chain_lower_curvature_v2",
+    "baseline": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "settings": {
+      "control_count": 18,
+      "lambda_curvature_cost": 5e-10,
+      "reprojection_weight": 4.0
+    },
+    "baseline_manifest_sha256": "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "baseline_fit_sha256": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455"
+  },
+  "stage_results": {
+    "fit": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 22.790251764992718,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "tube": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 4.972538448011619,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "colors": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 2.573285260994453,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "hybrid": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 2.255043400989962,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "review": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 12.419266690005315,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "manifest": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 1.3612327229930088,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    }
+  },
+  "saved_fit_identity_unchanged_during_review": true,
+  "actual_settings_match_owner_request_and_other_baseline_settings": true,
+  "baseline_fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 5.9778743889182806e-05,
+      "components": {
+        "distance": 2.406455132586416e-05,
+        "curvature": 1.2257775949819916e-07,
+        "arclength": 7.299825597328891e-07,
+        "mask_reprojection_2d": 3.486163404886611e-05
+      },
+      "grad_norm": 9.39435267355293e-05,
+      "arclength_m": 0.5877017974853516,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 37.92765808105469,
+      "curvature_p50_p90_per_m": [
+        9.860498428344727,
+        22.216008949279782
+      ],
+      "curvature_above_max_fraction": 0.492,
+      "mean_distance_m": 0.007389499891093807,
+      "max_distance_m": 0.017994370311498642,
+      "beyond_2delta_fraction": 0.19343938646114128,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.00687211100012064,
+          "p90_m": 0.010084983985871076,
+          "max_m": 0.013782148249447346,
+          "beyond_2delta_fraction": 0.10613207547169812,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0058602094650268555,
+          "p90_m": 0.010533764958381653,
+          "max_m": 0.017994370311498642,
+          "beyond_2delta_fraction": 0.11087387524782676,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008450270630419254,
+          "p90_m": 0.011142997816205025,
+          "max_m": 0.014050017111003399,
+          "beyond_2delta_fraction": 0.2523956951201533,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009255929850041866,
+          "p90_m": 0.012146585062146187,
+          "max_m": 0.015855133533477783,
+          "beyond_2delta_fraction": 0.38858154458951094,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.981192264140135
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 18,
+      "curvature_weight": 5e-09,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5848049521446228,
+      "mean_centerline_distance_to_truth_m": 0.003304384713212965,
+      "max_centerline_distance_to_truth_m": 0.0057414166823648,
+      "corner_truth_to_fit_mean_max_m": [
+        0.0034466403731989066,
+        0.004110262860931054
+      ],
+      "straight_truth_to_fit_mean_m": 0.003288980337124959,
+      "fit_max_curvature_per_m": 123.24449920654297,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        49.5619010925293,
+        123.24449920654297
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "909fa4c5991e924658e230d891b42027f9454e23297b69e7b0f18ee5600faa9a",
+      "spline/residuals.npz": "0de90046aee43ae7251db5e2a174d68c80e3d4317ea3f3a8fec38c76b47e2046",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "19d1b7bfa714a4afe18a0831dc78c1b42c6d087a91605aaf2ee11fd8ed6af064",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "9021e8d6d6e019fa4adc745ee9b3620ff75852583b523c7918ec6ea9b265ba6a",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455",
+      "spline/reprojection_comparison.json": "b40f5414d66a7047642440527c63a08d74c1b63b7bdfc3f3492451de0821fa8a",
+      "spline/rounds.jsonl": "a1dacca4ea7a924588f3b9713279c4834a7a8809e38b13c8334200e93a87dcd5",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "a2423fa09343e404bf108c2411ed62da67f3dd7477679ebf3d9855e99fe12223"
+    }
+  },
+  "input_copy_derived_output_policy": {
+    "path": "gaussian_data/session_20260921_233144_367389/shape_prior.glb",
+    "copied_from_baseline": false,
+    "reason": "Only the 27 true observation/filter/step inputs were copied. The new colors stage generated this derived GLB.",
+    "final_bytes_equal_new_tube_glb": true
+  },
+  "fit_sha256": "bc046638ca56b123bcce372e540192cd3cebee6fdb06e4d3b1868b114b3010aa",
+  "baseline_fit_and_manifest_unchanged": true,
+  "approval_status": "No prior is accepted; old approvals remain revoked.",
+  "reused_input_bytes_equal": {
+    "inputs/reference_views.json": true,
+    "inputs/observed_cloud.npz": true,
+    "inputs/depth_offset_diagnostic.json": true,
+    "spline/filter.json": true,
+    "spline/filter_removed.npz": true,
+    "spline/filtered_cloud.npz": true,
+    "steps/inputs.json": true,
+    "steps/diagnose-depth.json": true,
+    "gaussian_data/session_20260921_233144_367389/0.png": true,
+    "gaussian_data/session_20260921_233144_367389/0_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/1.png": true,
+    "gaussian_data/session_20260921_233144_367389/1_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/2.png": true,
+    "gaussian_data/session_20260921_233144_367389/2_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/3.png": true,
+    "gaussian_data/session_20260921_233144_367389/3_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/camera_meta.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/interp_poses.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/mask_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/observation.ply": true
+  },
+  "saved_array_checks": {
+    "spline/filtered_cloud.npz": {
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/residuals.npz": {
+      "distance_m": {
+        "shape": [
+          26339
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/reprojection_residuals.npz": {
+      "pixel_uv": {
+        "shape": [
+          39337,
+          2
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          39337
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "distance_m": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "distance_px": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/geometry.npz": {
+      "vertices": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "faces": {
+        "shape": [
+          3456,
+          3
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "centers": {
+        "shape": [
+          144,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "normals": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/triangle_areas.npz": {
+      "area_m2": {
+        "shape": [
+          3456
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "colors/vertex_colors.npz": {
+      "colors_rgb": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors_u8": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "uint8",
+        "finite": true
+      },
+      "observed_by_view": {
+        "shape": [
+          4,
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      },
+      "filled": {
+        "shape": [
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      }
+    },
+    "hybrid/populations.npz": {
+      "observed_points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "observed_colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "mesh_points": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_colors": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_normals": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      }
+    }
+  },
+  "last_round_matches_saved_final": true,
+  "per_view_residual_recomputed": {
+    "0": {
+      "count": 8480,
+      "median_m": 0.006861987989395857,
+      "p90_m": 0.0100558502599597,
+      "quantiles_match_saved": true
+    },
+    "1": {
+      "count": 6557,
+      "median_m": 0.005855448078364134,
+      "p90_m": 0.010509432852268222,
+      "quantiles_match_saved": true
+    },
+    "2": {
+      "count": 6783,
+      "median_m": 0.008462602272629738,
+      "p90_m": 0.011138082109391689,
+      "quantiles_match_saved": true
+    },
+    "3": {
+      "count": 4519,
+      "median_m": 0.009287589229643345,
+      "p90_m": 0.012151789292693138,
+      "quantiles_match_saved": true
+    }
+  },
+  "settings": {
+    "object": {
+      "name": "bike_chain",
+      "representation": "tube",
+      "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+      "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+      "radius_m": 0.00475,
+      "ribbon_width_m": null,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "u_resolution": null,
+      "overlap_rule": "surface_intersection",
+      "twist_penalty": null
+    },
+    "max_rounds": 400,
+    "control_count": 18,
+    "relative_loss_tolerance": 1e-07,
+    "reprojection": {
+      "weight": 4.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1"
+    },
+    "loss_config": {
+      "eval_distances": true,
+      "max_saturation_distance": 0.035,
+      "distance_huber_delta": 0.005,
+      "eval_curvature_cost": true,
+      "num_eval_reg_points": 250,
+      "lambda_curvature_cost": 5e-10,
+      "max_curvature": 10.0,
+      "curvature_huber_delta": 30.0,
+      "max_saturation_curvature": 80.0,
+      "eval_curvilinear_energy": false,
+      "lambda_curvilinear_energy": 0.001,
+      "eval_length_constraint": true,
+      "lambda_length_constraint": 0.1,
+      "arclength_bounds": [
+        0.5830000042915344,
+        0.5849999785423279
+      ]
+    },
+    "solver_config": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 5.9663550928235054e-05,
+      "components": {
+        "distance": 2.4059820134425536e-05,
+        "curvature": 1.2570627205832352e-08,
+        "arclength": 7.275368147929839e-07,
+        "mask_reprojection_2d": 3.486362402327359e-05
+      },
+      "grad_norm": 8.994463860290125e-05,
+      "arclength_m": 0.5876972675323486,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 38.438140869140625,
+      "curvature_p50_p90_per_m": [
+        9.838243007659912,
+        22.204584503173823
+      ],
+      "curvature_above_max_fraction": 0.488,
+      "mean_distance_m": 0.007388422403286789,
+      "max_distance_m": 0.017924964427947998,
+      "beyond_2delta_fraction": 0.19393295113709708,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.006861987989395857,
+          "p90_m": 0.0100558502599597,
+          "max_m": 0.013722896575927734,
+          "beyond_2delta_fraction": 0.10424528301886793,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.005855448078364134,
+          "p90_m": 0.010509432852268222,
+          "max_m": 0.017924964427947998,
+          "beyond_2delta_fraction": 0.11041634894006405,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008462602272629738,
+          "p90_m": 0.011138082109391689,
+          "max_m": 0.014056223444640636,
+          "beyond_2delta_fraction": 0.2528379772961816,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009287589229643345,
+          "p90_m": 0.012151789292693138,
+          "max_m": 0.015877872705459595,
+          "beyond_2delta_fraction": 0.39499889356052226,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.9812273925508821
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 18,
+      "curvature_weight": 5e-10,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5844525098800659,
+      "mean_centerline_distance_to_truth_m": 0.0032534316357658395,
+      "max_centerline_distance_to_truth_m": 0.005682110872846478,
+      "corner_truth_to_fit_mean_max_m": [
+        0.0033831706792910405,
+        0.004025381679852744
+      ],
+      "straight_truth_to_fit_mean_m": 0.0032304298767154866,
+      "fit_max_curvature_per_m": 127.63349151611328,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        51.74705123901367,
+        127.63349151611328
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "2cd8a2fa4305cf80b1d9c036c7f81889a3d4031e22ef66d2d31b1d1e8afb519a",
+      "spline/residuals.npz": "8159d786198decd56d8def2f9b3655a3173ec258047cad964dd47e01f0146c1b",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "19d1b7bfa714a4afe18a0831dc78c1b42c6d087a91605aaf2ee11fd8ed6af064",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "2ac95373504a4c61a1537eb2d67618b11c0071b7ddbd1631a70bab56095fc2b8",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "bc046638ca56b123bcce372e540192cd3cebee6fdb06e4d3b1868b114b3010aa",
+      "spline/reprojection_comparison.json": "23ab2e6bebeb66090c27444c6ca30171bac318492edb701be36f191748078d7b",
+      "spline/rounds.jsonl": "743067da945e62fb95572aa37db107f8264a12981873032242d352bf08f7e0f6",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "ff2dd099065a631994ae759a00bfa037afdd57cd4e45e66a7e73069ce8e81ab4"
+    }
+  },
+  "geometry_checks": {
+    "euler_characteristic": 0,
+    "watertight": true,
+    "components": 1,
+    "winding_consistent": true,
+    "signed_volume_m3": 3.9760913468027164e-05,
+    "minimum_outward_normal_dot_m": 0.004579301530990692,
+    "degenerate_triangles": 0,
+    "self_intersecting": false,
+    "maximum_radius_error_m": 8.147490698599824e-09,
+    "maximum_diameter_error_m": 1.604742104130419e-08,
+    "bishop_seam_error": 1.1920928955078125e-07,
+    "minimum_nonadjacent_centerline_distance_m": 0.006669520568941395,
+    "triangle_area_m2": {
+      "min": 3.3687073323917815e-06,
+      "median": 4.9904552902008765e-06,
+      "mean": 5.015839394544405e-06,
+      "max": 6.25396980998803e-06,
+      "density_max_over_min": 1.8564895055895856
+    },
+    "bounds_m": [
+      [
+        -0.0369243286550045,
+        -0.1740739345550537,
+        4.4061802327632904e-05
+      ],
+      [
+        0.17170047760009766,
+        -0.0010024229995906353,
+        0.017253154888749123
+      ]
+    ],
+    "radius_m": 0.00475,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "overlap_rule": "surface_intersection",
+    "open3d_reported_intersecting_pairs": 0,
+    "confirmed_intersecting_pairs": [],
+    "passed": true,
+    "orientation_flipped": true
+  },
+  "actual_geometry_counts": {
+    "rings": 144,
+    "vertices": 1728,
+    "triangles": 3456
+  },
+  "recomputed_max_radius_error_m": 8.147490698599824e-09,
+  "vertex_colors": {
+    "rule": "native sRGB/255, cosine weighted over mask/depth/first-hit visible views",
+    "two_sided": false,
+    "depth_tolerance_m": 0.025,
+    "first_hit_tolerance_m": 0.0002,
+    "per_view": [
+      {
+        "view": 0,
+        "serial": "12540311",
+        "observed_vertices": 611,
+        "projected_inside": 1728,
+        "first_hit": 922
+      },
+      {
+        "view": 1,
+        "serial": "12933256",
+        "observed_vertices": 686,
+        "projected_inside": 1728,
+        "first_hit": 909
+      },
+      {
+        "view": 2,
+        "serial": "17736741",
+        "observed_vertices": 628,
+        "projected_inside": 1728,
+        "first_hit": 908
+      },
+      {
+        "view": 3,
+        "serial": "23007103",
+        "observed_vertices": 749,
+        "projected_inside": 1728,
+        "first_hit": 940
+      }
+    ],
+    "observed_vertices": 1153,
+    "filled_vertices": 575,
+    "fill_color_rgb": [
+      0.20641317520444788,
+      0.16862745098039214,
+      0.1572585014457562
+    ],
+    "glb_roundtrip": {
+      "float32_positions_exact": true,
+      "vertex_colors_exact": true,
+      "uv_count": 0,
+      "vertex_reordered_on_open3d_load": true
+    }
+  },
+  "hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "baseline_hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "panels_to_inspect_relative_to_run": [
+    "review/inputs/view_0_12540311.png",
+    "review/inputs/view_1_12933256.png",
+    "review/inputs/view_2_17736741.png",
+    "review/inputs/view_3_23007103.png",
+    "review/spline/centerline_0_with_masks.png",
+    "review/spline/centerline_1_with_masks.png",
+    "review/spline/centerline_2_with_masks.png",
+    "review/spline/centerline_3_with_masks.png",
+    "review/tube/silhouette_0_unmasked.png",
+    "review/tube/silhouette_1_unmasked.png",
+    "review/tube/silhouette_2_unmasked.png",
+    "review/tube/silhouette_3_unmasked.png",
+    "review/tube/tube_scale_views.png",
+    "review/spline/residual_histograms.png",
+    "review/inputs/depth_offset.png",
+    "review/colors/visibility_and_fill.png",
+    "review/hybrid/combined.png",
+    "review/spline/fit_3d.png",
+    "review/spline/curvature_profile.png",
+    "review/spline/round_curves.png",
+    "review/tube/tube_profiles.png"
+  ],
+  "owner_prior_approval": "pending"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames.json
new file mode 100644
index 0000000000000000000000000000000000000000..cf85c3625d43845eda6ef42013610a449a2900cd
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames.json
@@ -0,0 +1,30 @@
+{
+  "source": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/orbit.mp4",
+  "video_frame_count": 36,
+  "fps": 12.0,
+  "selected_frame_indices": [
+    0,
+    4,
+    9,
+    13,
+    18,
+    22,
+    26,
+    31,
+    35
+  ],
+  "extracted_frames_relative_to_review": [
+    "orbit_frames/frame_000.png",
+    "orbit_frames/frame_004.png",
+    "orbit_frames/frame_009.png",
+    "orbit_frames/frame_013.png",
+    "orbit_frames/frame_018.png",
+    "orbit_frames/frame_022.png",
+    "orbit_frames/frame_026.png",
+    "orbit_frames/frame_031.png",
+    "orbit_frames/frame_035.png"
+  ],
+  "visual_inspection_pending": false,
+  "visual_inspection_completed": true,
+  "method": "Individual actual image reads of the nine extracted frames."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_000.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_000.png
new file mode 100644
index 0000000000000000000000000000000000000000..ad0bcdf9f31a9e88bf62f784670dd87d29ffae8f
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_000.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73c333d8e7acd7a58c804738a02dac9505a3653d075a54cb9a4328f9645489a7
+size 118608
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_004.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_004.png
new file mode 100644
index 0000000000000000000000000000000000000000..42b1e6932954bc49b86a9bd84cf92e5353ce6126
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_004.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a5dba6cc4c95b4d88528209bfc599674ef0d2add98f5c0c80542c5ef4828d48
+size 157056
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_009.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_009.png
new file mode 100644
index 0000000000000000000000000000000000000000..c07191fa7052bc531c5db31047da594c5e02327a
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_009.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:842284535822a387e62059fd4bfd52ee26a3dec5ba5a1eed933274ec96afe562
+size 132411
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_013.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_013.png
new file mode 100644
index 0000000000000000000000000000000000000000..06e7c55c986e5786cf3062091cf1c3f0dc87fc44
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_013.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa04940be1a0a3f268cf7a192fefbdadd9457edc5bb27426c765413edc32827c
+size 153637
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_018.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_018.png
new file mode 100644
index 0000000000000000000000000000000000000000..6440730e5f7c38f6e49e3b34b5f2f3fe945f8a73
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_018.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d29922fa48c25ba5aae7e335c03311f4c1b9dd5b93c47913f10af76ea6273d0
+size 131183
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_022.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_022.png
new file mode 100644
index 0000000000000000000000000000000000000000..a2e54dea47991a1d5813f2af3feae58be21ef357
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_022.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7551b6d5728d67853929a755954824a6d897358e96d24910f7c89367a8acc663
+size 156009
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_026.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_026.png
new file mode 100644
index 0000000000000000000000000000000000000000..7f54e6cf3f05c7acffb54ba9244b3e1a595763c7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_026.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af3ed39fd48b2bafb6520b3be59e9aa5abec6da86477d3e9e256f8caf24fbc50
+size 140003
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_031.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_031.png
new file mode 100644
index 0000000000000000000000000000000000000000..f605597dbec1db513d8454a353f391f2f3fbb38a
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_031.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02247ab573ad0bbc97e4f41b4b16f43dce9898dfffcb4f4bb0daa6d66dc55deb
+size 155160
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_035.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_035.png
new file mode 100644
index 0000000000000000000000000000000000000000..68bcff523f609a623430c22a3776bf5f1f1160d8
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/orbit_frames/frame_035.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4864f5e0e8a01032df8166a19f5c6c7c57bc69ae4f1c90793e39a42b0b0bd08e
+size 141064
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/prior_metrics.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/prior_metrics.json
new file mode 100644
index 0000000000000000000000000000000000000000..81e435e5a8ab1b8a3287317cf4f3fb243b8a31d0
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_lower_curvature_v2/prior_metrics.json
@@ -0,0 +1,491 @@
+{
+  "session_id": "session_20260921_233144_367389",
+  "prior_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2",
+  "owner_prior_approval": "pending",
+  "visual_quality_assessment": "Completed; see AGENT_FINDINGS.md and agent_visual_review.json. Prior acceptance pending; earlier chain acceptances remain revoked.",
+  "object": {
+    "name": "bike_chain",
+    "representation": "tube",
+    "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+    "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+    "radius_m": 0.00475,
+    "ribbon_width_m": null,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "u_resolution": null,
+    "overlap_rule": "surface_intersection",
+    "twist_penalty": null
+  },
+  "initial_grid_sample": 0,
+  "views": [
+    {
+      "serial": "12540311",
+      "view_index": 0,
+      "selected_window_local_index": 1,
+      "converted_frame": 269,
+      "native_container_frame": 270,
+      "official_frame": 269,
+      "depth_frame": 269,
+      "capture_timestamp_ns": 1790047938734565100,
+      "device_timestamp_ns": 1790047938691799126,
+      "initial_grid_offset_ms": -15.595843,
+      "native_count": 406,
+      "initial_observed_depth_points": 8480,
+      "median_3d_residual_m": 0.006861987989395857,
+      "p90_3d_residual_m": 0.0100558502599597,
+      "visible_centerline_inside_mask_fraction": 0.9901315789473685,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          160,
+          202,
+          253,
+          293,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1602,
+          "hand_A": 3398,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6356,
+          0.7556093596059114,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          0,
+          1177
+        ],
+        "paired_common_min_max": [
+          3195,
+          4564
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/spline/centerline_0_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_0_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_0_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/inputs/view_0_12540311.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/12540311_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/paired_depth.png"
+      }
+    },
+    {
+      "serial": "12933256",
+      "view_index": 1,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938718969257,
+      "device_timestamp_ns": 1790047938670420020,
+      "initial_grid_offset_ms": 0.0,
+      "native_count": 406,
+      "initial_observed_depth_points": 6557,
+      "median_3d_residual_m": 0.005855448078364134,
+      "p90_3d_residual_m": 0.010509432852268222,
+      "visible_centerline_inside_mask_fraction": 0.9665738161559888,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          202,
+          214,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1867,
+          "hand_A": 3133,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6628,
+          0.9445802955665025,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          1159,
+          1267
+        ],
+        "paired_common_min_max": [
+          1777,
+          4374
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/spline/centerline_1_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_1_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_1_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/inputs/view_1_12933256.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/12933256_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/paired_depth.png"
+      }
+    },
+    {
+      "serial": "17736741",
+      "view_index": 2,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938706187642,
+      "device_timestamp_ns": 1790047938663525231,
+      "initial_grid_offset_ms": 12.781615,
+      "native_count": 406,
+      "initial_observed_depth_points": 6783,
+      "median_3d_residual_m": 0.008462602272629738,
+      "p90_3d_residual_m": 0.011138082109391689,
+      "visible_centerline_inside_mask_fraction": 0.9841269841269842,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          119,
+          121,
+          151,
+          202,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1348,
+          "hand_A": 3652,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8368,
+          0.953579802955665,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          386,
+          897
+        ],
+        "paired_common_min_max": [
+          3762,
+          4527
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/spline/centerline_2_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_2_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_2_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/inputs/view_2_17736741.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/17736741_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/paired_depth.png"
+      }
+    },
+    {
+      "serial": "23007103",
+      "view_index": 3,
+      "selected_window_local_index": 1,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938720158122,
+      "device_timestamp_ns": 1790047938674181398,
+      "initial_grid_offset_ms": -1.188865,
+      "native_count": 407,
+      "initial_observed_depth_points": 4519,
+      "median_3d_residual_m": 0.009287589229643345,
+      "p90_3d_residual_m": 0.012151789292693138,
+      "visible_centerline_inside_mask_fraction": 0.9860335195530726,
+      "track_diagnostics": {
+        "T": 407,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          152,
+          203,
+          217,
+          253,
+          304,
+          355,
+          406
+        ],
+        "source_seed_counts": {
+          "object": 1992,
+          "hand_A": 3008,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8322,
+          0.9410083538083538,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          911,
+          1267
+        ],
+        "paired_common_min_max": [
+          2618,
+          4179
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/spline/centerline_3_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_3_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/silhouette_3_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/inputs/view_3_23007103.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/23007103_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/paired_depth.png"
+      }
+    }
+  ],
+  "recipe": {
+    "depth_source": "waft_stereo_dav2l5",
+    "mask_erosion_px": 2,
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "max_rounds": 400,
+    "reprojection": {
+      "weight": 4.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1"
+    },
+    "solver": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit": {
+    "stage1_rounds": 1,
+    "stage1_stop": "gradient_tolerance",
+    "stage1_final": null,
+    "twist": null,
+    "final": {
+      "loss": 5.9663550928235054e-05,
+      "components": {
+        "distance": 2.4059820134425536e-05,
+        "curvature": 1.2570627205832352e-08,
+        "arclength": 7.275368147929839e-07,
+        "mask_reprojection_2d": 3.486362402327359e-05
+      },
+      "grad_norm": 8.994463860290125e-05,
+      "arclength_m": 0.5876972675323486,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 38.438140869140625,
+      "curvature_p50_p90_per_m": [
+        9.838243007659912,
+        22.204584503173823
+      ],
+      "curvature_above_max_fraction": 0.488,
+      "mean_distance_m": 0.007388422403286789,
+      "max_distance_m": 0.017924964427947998,
+      "beyond_2delta_fraction": 0.19393295113709708,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.006861987989395857,
+          "p90_m": 0.0100558502599597,
+          "max_m": 0.013722896575927734,
+          "beyond_2delta_fraction": 0.10424528301886793,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.005855448078364134,
+          "p90_m": 0.010509432852268222,
+          "max_m": 0.017924964427947998,
+          "beyond_2delta_fraction": 0.11041634894006405,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008462602272629738,
+          "p90_m": 0.011138082109391689,
+          "max_m": 0.014056223444640636,
+          "beyond_2delta_fraction": 0.2528379772961816,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009287589229643345,
+          "p90_m": 0.012151789292693138,
+          "max_m": 0.015877872705459595,
+          "beyond_2delta_fraction": 0.39499889356052226,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.9812273925508821
+    },
+    "arclength_soft_bounds_m": [
+      0.5830000042915344,
+      0.5849999785423279
+    ],
+    "arclength_outside_soft_bounds_m": 0.002697288990020752
+  },
+  "geometry": {
+    "vertex_count": 1728,
+    "triangle_count": 3456,
+    "checks": {
+      "euler_characteristic": 0,
+      "watertight": true,
+      "components": 1,
+      "winding_consistent": true,
+      "signed_volume_m3": 3.9760913468027164e-05,
+      "minimum_outward_normal_dot_m": 0.004579301530990692,
+      "degenerate_triangles": 0,
+      "self_intersecting": false,
+      "maximum_radius_error_m": 8.147490698599824e-09,
+      "maximum_diameter_error_m": 1.604742104130419e-08,
+      "bishop_seam_error": 1.1920928955078125e-07,
+      "minimum_nonadjacent_centerline_distance_m": 0.006669520568941395,
+      "triangle_area_m2": {
+        "min": 3.3687073323917815e-06,
+        "median": 4.9904552902008765e-06,
+        "mean": 5.015839394544405e-06,
+        "max": 6.25396980998803e-06,
+        "density_max_over_min": 1.8564895055895856
+      },
+      "bounds_m": [
+        [
+          -0.0369243286550045,
+          -0.1740739345550537,
+          4.4061802327632904e-05
+        ],
+        [
+          0.17170047760009766,
+          -0.0010024229995906353,
+          0.017253154888749123
+        ]
+      ],
+      "radius_m": 0.00475,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "overlap_rule": "surface_intersection",
+      "open3d_reported_intersecting_pairs": 0,
+      "confirmed_intersecting_pairs": [],
+      "passed": true,
+      "orientation_flipped": true
+    }
+  },
+  "appearance_seed_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "vertex_colors": {
+    "observed_vertices": 1153,
+    "filled_vertices": 575
+  },
+  "mechanical_population": {
+    "particle_count": null,
+    "total_unit_mass": null,
+    "status": "not produced by the prior run; no post-prior mechanics are evaluated here"
+  },
+  "source_code": {
+    "commit": "98ebffe13629edeb0a9b29df973bebc5730d7026",
+    "u_pin": "81c718790a37e5e0102eb77af2c6edd34a9db25f"
+  },
+  "media": {
+    "full_rotation": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/orbit.mp4",
+    "geometry_depth_cloud": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/tube/tube_scale_views.png",
+    "depth_offset_diagnostic": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/inputs/depth_offset.png",
+    "residual_histograms": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/spline/residual_histograms.png",
+    "color_visibility_fill": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/colors/visibility_and_fill.png",
+    "hybrid_seed_populations": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/review/hybrid/combined.png",
+    "derived_coverage": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_coverage.png",
+    "derived_30hz_movie": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_30hz_xyz.mp4"
+  },
+  "sources": {
+    "manifest": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/manifest.json",
+    "fit": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/steps/fit.json",
+    "geometry_checks": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/tube/tube_checks.json",
+    "reference_views": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/inputs/reference_views.json",
+    "reprojection": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_lower_curvature_v2/spline/reprojection.json",
+    "tracking_summary": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/summary.json",
+    "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json"
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/AGENT_FINDINGS.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/AGENT_FINDINGS.md
new file mode 100644
index 0000000000000000000000000000000000000000..506c738f7872c890f0a7e968aafddc39bf388cd9
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/AGENT_FINDINGS.md
@@ -0,0 +1,46 @@
+# 233144 — stronger reprojection candidate review
+
+The completed candidate passes the saved numerical, input-preservation and geometry checks. **21 panels and nine orbit samples were directly inspected. Owner acceptance remains pending; previous chain-prior acceptances remain revoked.** This review grants no mechanics, appearance-training or physical-fitting authorization.
+
+Goal: review this full initial prior and compare it with the preserved 18-control baseline. Done when: settings/input/numeric checks and the actual visual inspection are complete and saved. Non-goals: refitting, changing observations or recording approval.
+
+The requested settings are **18 controls, curvature weight 5e-09, reprojection weight 8**. All other scientific settings exactly match the baseline, including the solver, input cloud/masks, 4.75 mm radius and tube sampling. All six preparation stages and the single packaging/check invocation returned zero. All 27 true observation/filter/step files remain byte-equal; original baseline fit/manifest and this saved fit remain unchanged. The derived Gaussian shape_prior.glb was newly generated and equals this candidate's tube GLB.
+
+[Numerical checks](numeric_prior_review.json) · [Exact visual record](agent_visual_review.json) · [Metrics and camera links](PRIOR_REVIEW_c6a_v1.md) · [Common comparison values](baseline_comparison.json)
+
+| Common diagnostic | Baseline: curvature 5e-9 / reprojection 4 | This candidate |
+|---|---:|---:|
+| Mean point-to-centerline distance (mm) | 7.389500 | 7.907245 |
+| Mean of camera mean reprojection distances (mm) | 3.676484 | 3.530591 |
+| Length (mm) | 587.702 | 589.568 |
+| Distance outside soft length bounds (mm) | 2.702 | 4.568 |
+| Maximum curvature (/m) | 37.928 | 39.228 |
+| Median curvature (/m) | 9.860 | 9.342 |
+| Outer rounds | 1 | 1 |
+| Final gradient norm | 9.39435e-05 | 7.99216e-05 |
+
+Doubling reprojection weight from 4 to 8 improves the common mean of camera mean reprojection distances from 3.676484 to 3.530591 mm, while worsening the common 3D mean from 7.389500 to 7.907245 mm. At the saved image scale, silhouette shifts are subtle and the same broad rounded loop remains; there is no clear visual resolution of the hidden grip arc.
+
+Length increases from 587.702 to 589.568 mm, placing it 4.568 mm above the soft upper bound, and maximum curvature increases from 37.928/m to 39.228/m. The tube moves lower relative to the nominal table. This candidate trades improved image-plane agreement for worse depth agreement and greater length deviation; it is not a uniform quality improvement or ground-truth accuracy result.
+
+The total objectives are **not directly comparable** after a coefficient changes. Common residuals are diagnostic, not ground-truth error; the stored comparison also gives reprojection loss before its weight. All runs retain the absolute 1e-4 gradient tolerance, which can change effective precision when a term is reweighted. This candidate stopped on `gradient_tolerance`; one outer round permits up to 25 L-BFGS updates, whose actual count is not saved. The 400-round value is only a budget. The single-round convergence line plot appears blank because one point is plotted without a marker; the saved round record equals the final result. [Saved rounds](../prior_chain_stronger_reprojection_v2/spline/rounds.jsonl).
+
+The four input panels preserve the rounded quadrilateral chain loop and the same hand-occluded grip arc. Visible object masks track the chain envelope, while hand union masks exclude the covered region. No conspicuous initial invalid-depth region is visible; this does not address the later temporal support gaps.
+
+The four centerline and silhouette views follow the broad visible loop, with small local edge offsets and rounded corners. The complete unmasked silhouette crosses the hand in views 0 and 2; the covered arc remains inferred. No new fold, extra loop or obvious self-crossing is visible in these images.
+
+The scale, 3D-fit and hybrid views show an extended observation cloud surrounding the tube; the mesh-derived seeds sit toward its lower portion. The profile reveals tube points below nominal table z=0, consistent with the numerical minimum −1.108 mm. This is an unresolved depth/contact limitation, not proof of a physical penetration depth. The local depth diagnostic is unchanged at roughly 9–12 mm above the table.
+
+Residual histograms shift toward larger 3D centerline distances, particularly views 2 and 3. Visibility/fill plots retain an underside and occluded-arc fill population: 579 of 1,728 vertices are median-filled. Curvature peaks near 39.2/m. The single-round convergence plots have no visible line segment because only one sample is available; numerical round data remain present.
+
+All nine inspected orbit samples show a single closed, gently nonplanar rounded loop without an obvious fold or crossing. Fine z interpretation comes from the scale/profile panels because orbit axis labels overlap. These sampled views do not establish the chain's link articulation or the shape behind the hand.
+
+Saved geometry checks report **144 rings, 1,728 vertices, 3,456 triangles**, one consistently wound watertight component and **zero confirmed surface intersections**. Maximum recomputed radial error is **7.36e-09 m**. Minimum nonadjacent ring-center distance is **6.633 mm**; this remains a diagnostic, not a substitute for the surface-intersection rule. Tube z spans **-1.108 to 16.361 mm** relative to nominal table z=0. No table snapping, depth correction or contact validation was performed.
+
+Colors are observed for **1149 vertices** and median-filled for **579**. Saved GLB roundtrip checks report exact float32 positions and colors, and colored-GLB hashes match current files. Hybrid initialization contains **26,339 observed + 103,680 mesh-derived = 130,019 seeds**. These populations match the baseline counts because all runs use 18 controls and the same observation bank/sampling settings. They are appearance seeds, not mechanical particles or mass. Seven checked NPZ families are finite; per-view residual quantiles recompute to their recorded values.
+
+Orbit frames **0, 4, 9, 13, 18, 22, 26, 31, 35 of 36** were inspected individually. [Rotation](../prior_chain_stronger_reprojection_v2/review/tube/orbit.mp4) · [Inspected orbit frames](orbit_frames.json). The 21 panels comprise four input/mask sets, four masked centerlines, four unmasked silhouettes and nine diagnostic panels. The smooth tube does not model individual chain links or prove the shape behind the hand.
+
+This is a hand_A-only sequence. Camera 12540311 has zero admitted object support at native-relative rows 143–238, 240, 243 and 262–326; 116 undefined object-mask coverage values remain NaNs. Finite coverage has median 0.479 and maximum 0.900. The derived middle/lowering views retain isolated points below or beside the main loop. Paired native-minus-WAFT medians are 0/0/−8/+2 mm and per-frame p95 maxima are 27/27/33/28 mm in camera order. These are common-support diagnostics, not ground-truth errors. Initial-prior changes do not clear later tracking or physical fitting. [Preserved temporal review](../review_v1/REVIEW.md).
+
+Only this new packet and scoped review scratch were written by the reviewer. No prior has been accepted by this review.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/PRIOR_REVIEW_c6a_v1.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/PRIOR_REVIEW_c6a_v1.md
new file mode 100644
index 0000000000000000000000000000000000000000..d616535b5a755c6eada0db99e9394f391eeb14c9
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/PRIOR_REVIEW_c6a_v1.md
@@ -0,0 +1,39 @@
+# session_20260921_233144_367389 — prior approval pending
+
+Saved C2 **bike_chain** prior evidence (radius 4.75 mm). Agent visual review is complete; owner acceptance remains pending, and earlier chain acceptances remain revoked. See [findings and baseline comparison](AGENT_FINDINGS.md). Saved values and media are linked here; the agent assessment is recorded separately. This packet records no acceptance.
+
+Initial reference frames are the saved nearest-camera observations of the first 30 Hz sample. The local row, converted frame, native-container frame and capture clock remain distinct in [the metrics JSON](<prior_metrics.json>).
+
+| Camera | Local / converted / native | Centerline / geometry / native input |
+|---|---|---|
+| 12540311 | 1 / 269 / 270 | [centerline](<../prior_chain_stronger_reprojection_v2/review/spline/centerline_0_with_masks.png>) · [tube](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_0_unmasked.png>) · [geometry + mask](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_0_with_masks.png>) · [RGB/depth](<../prior_chain_stronger_reprojection_v2/review/inputs/view_0_12540311.png>) |
+| 12933256 | 0 / 268 / 269 | [centerline](<../prior_chain_stronger_reprojection_v2/review/spline/centerline_1_with_masks.png>) · [tube](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_1_unmasked.png>) · [geometry + mask](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_1_with_masks.png>) · [RGB/depth](<../prior_chain_stronger_reprojection_v2/review/inputs/view_1_12933256.png>) |
+| 17736741 | 0 / 268 / 269 | [centerline](<../prior_chain_stronger_reprojection_v2/review/spline/centerline_2_with_masks.png>) · [tube](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_2_unmasked.png>) · [geometry + mask](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_2_with_masks.png>) · [RGB/depth](<../prior_chain_stronger_reprojection_v2/review/inputs/view_2_17736741.png>) |
+| 23007103 | 1 / 268 / 269 | [centerline](<../prior_chain_stronger_reprojection_v2/review/spline/centerline_3_with_masks.png>) · [tube](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_3_unmasked.png>) · [geometry + mask](<../prior_chain_stronger_reprojection_v2/review/tube/silhouette_3_with_masks.png>) · [RGB/depth](<../prior_chain_stronger_reprojection_v2/review/inputs/view_3_23007103.png>) |
+
+[Full rotation video](<../prior_chain_stronger_reprojection_v2/review/tube/orbit.mp4>) · [Geometry and depth cloud](<../prior_chain_stronger_reprojection_v2/review/tube/tube_scale_views.png>) · [Residual histograms](<../prior_chain_stronger_reprojection_v2/review/spline/residual_histograms.png>) · [Depth offset diagnostic](<../prior_chain_stronger_reprojection_v2/review/inputs/depth_offset.png>)
+
+Saved fit: **1 stage-1 rounds**, stop `gradient_tolerance`; final length **589.568 mm**, maximum curvature **39.228/m**, final gradient norm **7.99216e-05**. Soft length bounds are 583.000–585.000 mm; distance outside those bounds is 4.568 mm.
+
+Saved geometry checks report `passed=True`: **1728 mesh vertices / 3456 triangles**, 1 component(s), `watertight=True`, `self_intersecting=False`. [Full geometry checks](<../prior_chain_stronger_reprojection_v2/tube/tube_checks.json>).
+
+The following 3D residuals are lifted-depth point distances to the **centerline**, not measured reconstruction errors. The centerline-in-mask fraction is a separate saved diagnostic; it is not silhouette IoU or a visual quality decision.
+
+| Camera | Median / p90 3D residual (mm) | Visible centerline inside mask | Native depth and visibility evidence |
+|---|---|---|---|
+| 12540311 | 7.463 / 11.094 | 99.34% | [Full native video](<../review_v1/12540311/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12540311/12540311_visibility_certainty.png>) · [Support/rejections](<../review_v1/12540311/support_and_rejections.png>) · [Paired depth](<../review_v1/12540311/paired_depth.png>) |
+| 12933256 | 6.107 / 10.636 | 96.67% | [Full native video](<../review_v1/12933256/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12933256/12933256_visibility_certainty.png>) · [Support/rejections](<../review_v1/12933256/support_and_rejections.png>) · [Paired depth](<../review_v1/12933256/paired_depth.png>) |
+| 17736741 | 9.260 / 12.170 | 99.36% | [Full native video](<../review_v1/17736741/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/17736741/17736741_visibility_certainty.png>) · [Support/rejections](<../review_v1/17736741/support_and_rejections.png>) · [Paired depth](<../review_v1/17736741/paired_depth.png>) |
+| 23007103 | 9.861 / 12.667 | 98.60% | [Full native video](<../review_v1/23007103/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/23007103/23007103_visibility_certainty.png>) · [Support/rejections](<../review_v1/23007103/support_and_rejections.png>) · [Paired depth](<../review_v1/23007103/paired_depth.png>) |
+
+The saved statistical filter used 20 neighbours / 2 sigma and kept **26339 of 27239** eroded points. The run used `waft_stereo_dav2l5`, 2 px erosion, a 400-round budget and reprojection weight **8**.
+
+Appearance initialization contains **26339 observed + 103680 mesh-derived = 130019 seeds**. Vertex colors are observed for 1149 mesh vertices and median-filled for 579. Mesh vertices and appearance seeds are not mechanical particles. No post-prior mechanical population or unit-mass total is supplied by this prior run; trained appearance and physical fitting are separate stages.
+
+[Appearance seed populations](<../prior_chain_stronger_reprojection_v2/review/hybrid/combined.png>) · [Observed and filled colors](<../prior_chain_stronger_reprojection_v2/review/colors/visibility_and_fill.png>) · [30 Hz observation coverage](<../review_v1/derived/derived_coverage.png>)
+
+[Full saved 30 Hz observations](<../review_v1/derived/derived_30hz_xyz.mp4>)
+
+[Manifest](<../prior_chain_stronger_reprojection_v2/manifest.json>) · [Fit settings and metrics](<../prior_chain_stronger_reprojection_v2/steps/fit.json>) · [Reference observations](<../prior_chain_stronger_reprojection_v2/inputs/reference_views.json>) · [Tracking/lift summary](<../review_v1/summary.json>)
+
+The agent inspected 21 panels and 9 orbit frames; see [the exact visual record](agent_visual_review.json) and [findings](AGENT_FINDINGS.md). Owner prior approval remains pending. No case status, prior acceptance, mechanics approval or trained-appearance approval is changed here.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/agent_visual_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/agent_visual_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..4880abd6534816b34b96a6e6fbb0fa8b0fc6993e
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/agent_visual_review.json
@@ -0,0 +1,160 @@
+{
+  "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+  "candidate": "stronger_reprojection",
+  "prior_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2",
+  "reviewed_at_utc": "2026-10-28T22:57:32.356936+00:00",
+  "review_kind": "bounded numeric and visual review of completed owner-requested chain weight candidate",
+  "readiness": "ready_for_owner_comparison_with_disclosed_limits",
+  "blocking_findings_for_review_completion": [],
+  "owner_prior_acceptance": "pending; all earlier chain prior acceptances remain revoked",
+  "scope": {
+    "goal": "Review the completed candidate and compare common diagnostics against the preserved 18-control baseline.",
+    "done_when": "Six successful stages, single scoped packaging/check invocation,21 actual panels and 9 orbit samples inspected, numerical/settings/input checks and findings saved.",
+    "non_goals": [
+      "No scientific execution, fitting or source changes",
+      "No global or approval changes",
+      "No mechanics, appearance training or physical-fitting approval"
+    ]
+  },
+  "job": {
+    "case_id": "233144",
+    "session_id": "session_20260921_233144_367389",
+    "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "prep": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1",
+    "candidate": "stronger_reprojection",
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2",
+    "review": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_chain_stronger_reprojection_v2",
+    "baseline": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "settings": {
+      "control_count": 18,
+      "lambda_curvature_cost": 5e-09,
+      "reprojection_weight": 8.0
+    },
+    "baseline_manifest_sha256": "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "baseline_fit_sha256": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455"
+  },
+  "packaging_and_check_result": {
+    "returncode": 0,
+    "elapsed_s": 0.6984232090035221,
+    "completed_utc": "2026-10-28T22:52:10.014640+00:00",
+    "argv": [
+      "flock",
+      "-x",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/resource_lane.lock",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/.venv/bin/python",
+      "-B",
+      "/mnt/data0/magna/belt_perception_wt_c6a_lab_input_preparation_20261027/outputs/c6a_recovery/agent_drafts/chain_weight_pair_review_233045_233144_20261028/package_and_check.py",
+      "--case",
+      "233144",
+      "--candidate",
+      "stronger_reprojection"
+    ]
+  },
+  "numeric_review_path": "numeric_prior_review.json",
+  "baseline_comparison_path": "baseline_comparison.json",
+  "unchanged_input_count": 27,
+  "geometry_checks": {
+    "euler_characteristic": 0,
+    "watertight": true,
+    "components": 1,
+    "winding_consistent": true,
+    "signed_volume_m3": 3.988753950276397e-05,
+    "minimum_outward_normal_dot_m": 0.0045789361996072244,
+    "degenerate_triangles": 0,
+    "self_intersecting": false,
+    "maximum_radius_error_m": 7.36314771567631e-09,
+    "maximum_diameter_error_m": 1.4711541323694788e-08,
+    "bishop_seam_error": 5.960464477539063e-08,
+    "minimum_nonadjacent_centerline_distance_m": 0.006632810951101171,
+    "triangle_area_m2": {
+      "min": 3.3363548276930293e-06,
+      "median": 4.978674734057986e-06,
+      "mean": 5.031806731543179e-06,
+      "max": 6.320388433778611e-06,
+      "density_max_over_min": 1.8943993550437004
+    },
+    "bounds_m": [
+      [
+        -0.03722784295678139,
+        -0.17418868839740753,
+        -0.0011078841052949429
+      ],
+      [
+        0.17201127111911774,
+        -0.0006560203619301319,
+        0.01636137068271637
+      ]
+    ],
+    "radius_m": 0.00475,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "overlap_rule": "surface_intersection",
+    "open3d_reported_intersecting_pairs": 0,
+    "confirmed_intersecting_pairs": [],
+    "passed": true,
+    "orientation_flipped": true
+  },
+  "glb_roundtrip_checks": {
+    "float32_positions_exact": true,
+    "vertex_colors_exact": true,
+    "uv_count": 0,
+    "vertex_reordered_on_open3d_load": true
+  },
+  "observed_counts_unchanged": true,
+  "inspected_panels_relative_to_run": [
+    "review/inputs/view_0_12540311.png",
+    "review/inputs/view_1_12933256.png",
+    "review/inputs/view_2_17736741.png",
+    "review/inputs/view_3_23007103.png",
+    "review/spline/centerline_0_with_masks.png",
+    "review/spline/centerline_1_with_masks.png",
+    "review/spline/centerline_2_with_masks.png",
+    "review/spline/centerline_3_with_masks.png",
+    "review/tube/silhouette_0_unmasked.png",
+    "review/tube/silhouette_1_unmasked.png",
+    "review/tube/silhouette_2_unmasked.png",
+    "review/tube/silhouette_3_unmasked.png",
+    "review/tube/tube_scale_views.png",
+    "review/spline/residual_histograms.png",
+    "review/inputs/depth_offset.png",
+    "review/colors/visibility_and_fill.png",
+    "review/hybrid/combined.png",
+    "review/spline/fit_3d.png",
+    "review/spline/curvature_profile.png",
+    "review/spline/round_curves.png",
+    "review/tube/tube_profiles.png"
+  ],
+  "inspected_orbit_frames_relative_to_review": [
+    "orbit_frames/frame_000.png",
+    "orbit_frames/frame_004.png",
+    "orbit_frames/frame_009.png",
+    "orbit_frames/frame_013.png",
+    "orbit_frames/frame_018.png",
+    "orbit_frames/frame_022.png",
+    "orbit_frames/frame_026.png",
+    "orbit_frames/frame_031.png",
+    "orbit_frames/frame_035.png"
+  ],
+  "inspection_method": "Direct individual image reads of all 21 panels and nine sampled orbit frames; four baseline silhouettes were also inspected for comparison.",
+  "blocking_findings": [],
+  "observations": [
+    "The four input panels preserve the rounded quadrilateral chain loop and the same hand-occluded grip arc. Visible object masks track the chain envelope, while hand union masks exclude the covered region. No conspicuous initial invalid-depth region is visible; this does not address the later temporal support gaps.",
+    "The four centerline and silhouette views follow the broad visible loop, with small local edge offsets and rounded corners. The complete unmasked silhouette crosses the hand in views 0 and 2; the covered arc remains inferred. No new fold, extra loop or obvious self-crossing is visible in these images.",
+    "The scale, 3D-fit and hybrid views show an extended observation cloud surrounding the tube; the mesh-derived seeds sit toward its lower portion. The profile reveals tube points below nominal table z=0, consistent with the numerical minimum \u22121.108 mm. This is an unresolved depth/contact limitation, not proof of a physical penetration depth. The local depth diagnostic is unchanged at roughly 9\u201312 mm above the table.",
+    "Residual histograms shift toward larger 3D centerline distances, particularly views 2 and 3. Visibility/fill plots retain an underside and occluded-arc fill population: 579 of 1,728 vertices are median-filled. Curvature peaks near 39.2/m. The single-round convergence plots have no visible line segment because only one sample is available; numerical round data remain present.",
+    "All nine inspected orbit samples show a single closed, gently nonplanar rounded loop without an obvious fold or crossing. Fine z interpretation comes from the scale/profile panels because orbit axis labels overlap. These sampled views do not establish the chain's link articulation or the shape behind the hand."
+  ],
+  "comparison_observations": [
+    "Doubling reprojection weight from 4 to 8 improves the common mean of camera mean reprojection distances from 3.676484 to 3.530591 mm, while worsening the common 3D mean from 7.389500 to 7.907245 mm. At the saved image scale, silhouette shifts are subtle and the same broad rounded loop remains; there is no clear visual resolution of the hidden grip arc.",
+    "Length increases from 587.702 to 589.568 mm, placing it 4.568 mm above the soft upper bound, and maximum curvature increases from 37.928/m to 39.228/m. The tube moves lower relative to the nominal table. This candidate trades improved image-plane agreement for worse depth agreement and greater length deviation; it is not a uniform quality improvement or ground-truth accuracy result."
+  ],
+  "limitations": [
+    "The 4.75 mm radius continuous circular tube does not model individual links, gaps, rollers, plates or articulation; hidden geometry is inferred.",
+    "Length 589.568257 mm is 4.568279 mm outside the soft interval. The 10/m threshold and 80/m smooth saturation are penalty parameters, not hard curvature bounds.",
+    "Tube z bounds are -1.107884 to 16.361371 mm relative to nominal table z=0. Depth/table offsets and contact uncertainty remain unchanged.",
+    "A lower common residual does not establish ground-truth shape accuracy. Changed objective totals are not ranked; fixed absolute gradient tolerance can affect effective precision when weights change.",
+    "This is a hand_A-only sequence. Camera 12540311 has zero admitted object support at native-relative rows 143\u2013238, 240, 243 and 262\u2013326; 116 undefined object-mask coverage values remain NaNs. Finite coverage has median 0.479 and maximum 0.900. The derived middle/lowering views retain isolated points below or beside the main loop. Paired native-minus-WAFT medians are 0/0/\u22128/+2 mm and per-frame p95 maxima are 27/27/33/28 mm in camera order. These are common-support diagnostics, not ground-truth errors. Initial-prior changes do not clear later tracking or physical fitting.",
+    "Inspection covers 21 saved panels and nine orbit frames, not every video frame. No mechanics, appearance training or physical fit is assessed or accepted."
+  ],
+  "approval_effect": "No approval is recorded; execution authorization is not acceptance."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/baseline_comparison.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/baseline_comparison.json
new file mode 100644
index 0000000000000000000000000000000000000000..6b9357b35d5f07a9fc5224f6f93749705eec14dd
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/baseline_comparison.json
@@ -0,0 +1,47 @@
+{
+  "baseline": {
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "control_count": 18,
+    "curvature_weight": 5e-09,
+    "reprojection_weight": 4.0,
+    "mean_point_to_centerline_mm": 7.3894998910938074,
+    "mean_of_camera_mean_reprojection_mm": 3.6764842167760454,
+    "length_mm": 587.7017974853516,
+    "length_outside_soft_bounds_mm": 2.7018189430236816,
+    "maximum_curvature_per_m": 37.92765808105469,
+    "median_curvature_per_m": 9.860498428344727,
+    "common_3d_loss": 2.406455132586416e-05,
+    "unweighted_reprojection_loss": 8.715408512216527e-06,
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "gradient_norm": 9.39435267355293e-05,
+    "wall_time_s": 21.2205913066864
+  },
+  "candidate": {
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2",
+    "control_count": 18,
+    "curvature_weight": 5e-09,
+    "reprojection_weight": 8.0,
+    "mean_point_to_centerline_mm": 7.907245111366833,
+    "mean_of_camera_mean_reprojection_mm": 3.5305909241216424,
+    "length_mm": 589.5682573318481,
+    "length_outside_soft_bounds_mm": 4.568278789520264,
+    "maximum_curvature_per_m": 39.227989196777344,
+    "median_curvature_per_m": 9.342434406280518,
+    "common_3d_loss": 2.6436840926180594e-05,
+    "unweighted_reprojection_loss": 8.036608960537706e-06,
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "gradient_norm": 7.992162863956764e-05,
+    "wall_time_s": 20.597996711730957
+  },
+  "same_observed_inputs": true,
+  "only_scientific_setting_change": {
+    "control_count": 18,
+    "lambda_curvature_cost": 5e-09,
+    "reprojection_weight": 8.0
+  },
+  "all_other_baseline_settings_match": true,
+  "total_objectives_comparable": false,
+  "interpretation": "Common raw residuals and unweighted reprojection losses are compared. The coefficient changes the objective; absolute gradient stopping also changes effective precision. No optimum or ground-truth accuracy claim."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/numeric_prior_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/numeric_prior_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..c97ae2a2885ca8841f4a08c1e5fe3c1532f8ad3f
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/numeric_prior_review.json
@@ -0,0 +1,824 @@
+{
+  "scope": "Saved 233144 stronger_reprojection prior review only; no fitting, sampling or scientific correction executed by this reviewer.",
+  "job": {
+    "case_id": "233144",
+    "session_id": "session_20260921_233144_367389",
+    "case": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673",
+    "prep": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1",
+    "candidate": "stronger_reprojection",
+    "run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2",
+    "review": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_chain_stronger_reprojection_v2",
+    "baseline": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+    "settings": {
+      "control_count": 18,
+      "lambda_curvature_cost": 5e-09,
+      "reprojection_weight": 8.0
+    },
+    "baseline_manifest_sha256": "44d944a5d07bb7a385b346ba916e745083d00a032b3dd560a53a18e4afc16007",
+    "baseline_fit_sha256": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455"
+  },
+  "stage_results": {
+    "fit": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 22.655370295004104,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "tube": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 4.938233467997634,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "colors": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 2.5460264819994336,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "hybrid": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 2.2611869700049283,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "review": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 12.528403839998646,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    },
+    "manifest": {
+      "returncode": 0,
+      "elapsed_s_excluding_queue": 1.326893513003597,
+      "source_commit": "98ebffe13629edeb0a9b29df973bebc5730d7026"
+    }
+  },
+  "saved_fit_identity_unchanged_during_review": true,
+  "actual_settings_match_owner_request_and_other_baseline_settings": true,
+  "baseline_fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 5.9778743889182806e-05,
+      "components": {
+        "distance": 2.406455132586416e-05,
+        "curvature": 1.2257775949819916e-07,
+        "arclength": 7.299825597328891e-07,
+        "mask_reprojection_2d": 3.486163404886611e-05
+      },
+      "grad_norm": 9.39435267355293e-05,
+      "arclength_m": 0.5877017974853516,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 37.92765808105469,
+      "curvature_p50_p90_per_m": [
+        9.860498428344727,
+        22.216008949279782
+      ],
+      "curvature_above_max_fraction": 0.492,
+      "mean_distance_m": 0.007389499891093807,
+      "max_distance_m": 0.017994370311498642,
+      "beyond_2delta_fraction": 0.19343938646114128,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.00687211100012064,
+          "p90_m": 0.010084983985871076,
+          "max_m": 0.013782148249447346,
+          "beyond_2delta_fraction": 0.10613207547169812,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0058602094650268555,
+          "p90_m": 0.010533764958381653,
+          "max_m": 0.017994370311498642,
+          "beyond_2delta_fraction": 0.11087387524782676,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008450270630419254,
+          "p90_m": 0.011142997816205025,
+          "max_m": 0.014050017111003399,
+          "beyond_2delta_fraction": 0.2523956951201533,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009255929850041866,
+          "p90_m": 0.012146585062146187,
+          "max_m": 0.015855133533477783,
+          "beyond_2delta_fraction": 0.38858154458951094,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.981192264140135
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 18,
+      "curvature_weight": 5e-09,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5848049521446228,
+      "mean_centerline_distance_to_truth_m": 0.003304384713212965,
+      "max_centerline_distance_to_truth_m": 0.0057414166823648,
+      "corner_truth_to_fit_mean_max_m": [
+        0.0034466403731989066,
+        0.004110262860931054
+      ],
+      "straight_truth_to_fit_mean_m": 0.003288980337124959,
+      "fit_max_curvature_per_m": 123.24449920654297,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        49.5619010925293,
+        123.24449920654297
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "909fa4c5991e924658e230d891b42027f9454e23297b69e7b0f18ee5600faa9a",
+      "spline/residuals.npz": "0de90046aee43ae7251db5e2a174d68c80e3d4317ea3f3a8fec38c76b47e2046",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "19d1b7bfa714a4afe18a0831dc78c1b42c6d087a91605aaf2ee11fd8ed6af064",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "9021e8d6d6e019fa4adc745ee9b3620ff75852583b523c7918ec6ea9b265ba6a",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455",
+      "spline/reprojection_comparison.json": "b40f5414d66a7047642440527c63a08d74c1b63b7bdfc3f3492451de0821fa8a",
+      "spline/rounds.jsonl": "a1dacca4ea7a924588f3b9713279c4834a7a8809e38b13c8334200e93a87dcd5",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "a2423fa09343e404bf108c2411ed62da67f3dd7477679ebf3d9855e99fe12223"
+    }
+  },
+  "input_copy_derived_output_policy": {
+    "path": "gaussian_data/session_20260921_233144_367389/shape_prior.glb",
+    "copied_from_baseline": false,
+    "reason": "Only the 27 true observation/filter/step inputs were copied. The new colors stage generated this derived GLB.",
+    "final_bytes_equal_new_tube_glb": true
+  },
+  "fit_sha256": "251baa62b8905d4b0ecd2e763e8a1bb85b7142ca69fa8ccf2ae59fb87b1294cf",
+  "baseline_fit_and_manifest_unchanged": true,
+  "approval_status": "No prior is accepted; old approvals remain revoked.",
+  "reused_input_bytes_equal": {
+    "inputs/reference_views.json": true,
+    "inputs/observed_cloud.npz": true,
+    "inputs/depth_offset_diagnostic.json": true,
+    "spline/filter.json": true,
+    "spline/filter_removed.npz": true,
+    "spline/filtered_cloud.npz": true,
+    "steps/inputs.json": true,
+    "steps/diagnose-depth.json": true,
+    "gaussian_data/session_20260921_233144_367389/0.png": true,
+    "gaussian_data/session_20260921_233144_367389/0_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/1.png": true,
+    "gaussian_data/session_20260921_233144_367389/1_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/2.png": true,
+    "gaussian_data/session_20260921_233144_367389/2_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/3.png": true,
+    "gaussian_data/session_20260921_233144_367389/3_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/camera_meta.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/interp_poses.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/mask_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/observation.ply": true
+  },
+  "saved_array_checks": {
+    "spline/filtered_cloud.npz": {
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/residuals.npz": {
+      "distance_m": {
+        "shape": [
+          26339
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/reprojection_residuals.npz": {
+      "pixel_uv": {
+        "shape": [
+          39337,
+          2
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          39337
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "distance_m": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "distance_px": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/geometry.npz": {
+      "vertices": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "faces": {
+        "shape": [
+          3456,
+          3
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "centers": {
+        "shape": [
+          144,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "normals": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/triangle_areas.npz": {
+      "area_m2": {
+        "shape": [
+          3456
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "colors/vertex_colors.npz": {
+      "colors_rgb": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors_u8": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "uint8",
+        "finite": true
+      },
+      "observed_by_view": {
+        "shape": [
+          4,
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      },
+      "filled": {
+        "shape": [
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      }
+    },
+    "hybrid/populations.npz": {
+      "observed_points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "observed_colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "mesh_points": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_colors": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_normals": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      }
+    }
+  },
+  "last_round_matches_saved_final": true,
+  "per_view_residual_recomputed": {
+    "0": {
+      "count": 8480,
+      "median_m": 0.0074629574082791805,
+      "p90_m": 0.011093903612345458,
+      "quantiles_match_saved": true
+    },
+    "1": {
+      "count": 6557,
+      "median_m": 0.0061066653579473495,
+      "p90_m": 0.010635773651301864,
+      "quantiles_match_saved": true
+    },
+    "2": {
+      "count": 6783,
+      "median_m": 0.009259557351469994,
+      "p90_m": 0.012170043773949146,
+      "quantiles_match_saved": true
+    },
+    "3": {
+      "count": 4519,
+      "median_m": 0.009861292317509651,
+      "p90_m": 0.012667492963373663,
+      "quantiles_match_saved": true
+    }
+  },
+  "settings": {
+    "object": {
+      "name": "bike_chain",
+      "representation": "tube",
+      "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+      "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+      "radius_m": 0.00475,
+      "ribbon_width_m": null,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "u_resolution": null,
+      "overlap_rule": "surface_intersection",
+      "twist_penalty": null
+    },
+    "max_rounds": 400,
+    "control_count": 18,
+    "relative_loss_tolerance": 1e-07,
+    "reprojection": {
+      "weight": 8.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1"
+    },
+    "loss_config": {
+      "eval_distances": true,
+      "max_saturation_distance": 0.035,
+      "distance_huber_delta": 0.005,
+      "eval_curvature_cost": true,
+      "num_eval_reg_points": 250,
+      "lambda_curvature_cost": 5e-09,
+      "max_curvature": 10.0,
+      "curvature_huber_delta": 30.0,
+      "max_saturation_curvature": 80.0,
+      "eval_curvilinear_energy": false,
+      "lambda_curvilinear_energy": 0.001,
+      "eval_length_constraint": true,
+      "lambda_length_constraint": 0.1,
+      "arclength_bounds": [
+        0.5830000042915344,
+        0.5849999785423279
+      ]
+    },
+    "solver_config": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 9.294429037254304e-05,
+      "components": {
+        "distance": 2.6436840926180594e-05,
+        "curvature": 1.2765701740136137e-07,
+        "arclength": 2.086917220367468e-06,
+        "mask_reprojection_2d": 6.429287168430164e-05
+      },
+      "grad_norm": 7.992162863956764e-05,
+      "arclength_m": 0.5895682573318481,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 39.227989196777344,
+      "curvature_p50_p90_per_m": [
+        9.342434406280518,
+        21.996375656127924
+      ],
+      "curvature_above_max_fraction": 0.48,
+      "mean_distance_m": 0.007907245111366833,
+      "max_distance_m": 0.01780632697045803,
+      "beyond_2delta_fraction": 0.2766619841299973,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.0074629574082791805,
+          "p90_m": 0.011093903612345458,
+          "max_m": 0.01485209446400404,
+          "beyond_2delta_fraction": 0.1919811320754717,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0061066653579473495,
+          "p90_m": 0.010635773651301864,
+          "max_m": 0.01780632697045803,
+          "beyond_2delta_fraction": 0.11926185755680951,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.009259557351469994,
+          "p90_m": 0.012170043773949146,
+          "max_m": 0.015088371001183987,
+          "beyond_2delta_fraction": 0.4017396432257113,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009861292317509651,
+          "p90_m": 0.012667492963373663,
+          "max_m": 0.01604742929339409,
+          "beyond_2delta_fraction": 0.47621155122814784,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.9733169029086446
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 18,
+      "curvature_weight": 5e-09,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5848049521446228,
+      "mean_centerline_distance_to_truth_m": 0.003304384713212965,
+      "max_centerline_distance_to_truth_m": 0.0057414166823648,
+      "corner_truth_to_fit_mean_max_m": [
+        0.0034466403731989066,
+        0.004110262860931054
+      ],
+      "straight_truth_to_fit_mean_m": 0.003288980337124959,
+      "fit_max_curvature_per_m": 123.24449920654297,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        49.5619010925293,
+        123.24449920654297
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "89d7f78c6867b1e3d72fa071a831b9f81b2381b4177301708673d8f57cde20e9",
+      "spline/residuals.npz": "67bea5a30b5efeec7c7ae6c26d5722c2379f9c31e6715735555a267ecbb9a0e8",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "19d1b7bfa714a4afe18a0831dc78c1b42c6d087a91605aaf2ee11fd8ed6af064",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "9021e8d6d6e019fa4adc745ee9b3620ff75852583b523c7918ec6ea9b265ba6a",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "251baa62b8905d4b0ecd2e763e8a1bb85b7142ca69fa8ccf2ae59fb87b1294cf",
+      "spline/reprojection_comparison.json": "0650062017a1c91b69e6098bc1dffd79788091f14bd5d7cdd21dcda90c3a1c99",
+      "spline/rounds.jsonl": "f92f83fdbeeefd4bebede093b6ed3735c21450979d8d4757a06f615d640634f0",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "22d805b7625267859dcce673641b09158a87d2a042226e7eea5a78597bec2a8f"
+    }
+  },
+  "geometry_checks": {
+    "euler_characteristic": 0,
+    "watertight": true,
+    "components": 1,
+    "winding_consistent": true,
+    "signed_volume_m3": 3.988753950276397e-05,
+    "minimum_outward_normal_dot_m": 0.0045789361996072244,
+    "degenerate_triangles": 0,
+    "self_intersecting": false,
+    "maximum_radius_error_m": 7.36314771567631e-09,
+    "maximum_diameter_error_m": 1.4711541323694788e-08,
+    "bishop_seam_error": 5.960464477539063e-08,
+    "minimum_nonadjacent_centerline_distance_m": 0.006632810951101171,
+    "triangle_area_m2": {
+      "min": 3.3363548276930293e-06,
+      "median": 4.978674734057986e-06,
+      "mean": 5.031806731543179e-06,
+      "max": 6.320388433778611e-06,
+      "density_max_over_min": 1.8943993550437004
+    },
+    "bounds_m": [
+      [
+        -0.03722784295678139,
+        -0.17418868839740753,
+        -0.0011078841052949429
+      ],
+      [
+        0.17201127111911774,
+        -0.0006560203619301319,
+        0.01636137068271637
+      ]
+    ],
+    "radius_m": 0.00475,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "overlap_rule": "surface_intersection",
+    "open3d_reported_intersecting_pairs": 0,
+    "confirmed_intersecting_pairs": [],
+    "passed": true,
+    "orientation_flipped": true
+  },
+  "actual_geometry_counts": {
+    "rings": 144,
+    "vertices": 1728,
+    "triangles": 3456
+  },
+  "recomputed_max_radius_error_m": 7.36314771567631e-09,
+  "vertex_colors": {
+    "rule": "native sRGB/255, cosine weighted over mask/depth/first-hit visible views",
+    "two_sided": false,
+    "depth_tolerance_m": 0.025,
+    "first_hit_tolerance_m": 0.0002,
+    "per_view": [
+      {
+        "view": 0,
+        "serial": "12540311",
+        "observed_vertices": 640,
+        "projected_inside": 1728,
+        "first_hit": 924
+      },
+      {
+        "view": 1,
+        "serial": "12933256",
+        "observed_vertices": 719,
+        "projected_inside": 1728,
+        "first_hit": 899
+      },
+      {
+        "view": 2,
+        "serial": "17736741",
+        "observed_vertices": 643,
+        "projected_inside": 1728,
+        "first_hit": 912
+      },
+      {
+        "view": 3,
+        "serial": "23007103",
+        "observed_vertices": 781,
+        "projected_inside": 1728,
+        "first_hit": 923
+      }
+    ],
+    "observed_vertices": 1149,
+    "filled_vertices": 579,
+    "fill_color_rgb": [
+      0.20322971656294556,
+      0.16635644166967856,
+      0.15421207534277223
+    ],
+    "glb_roundtrip": {
+      "float32_positions_exact": true,
+      "vertex_colors_exact": true,
+      "uv_count": 0,
+      "vertex_reordered_on_open3d_load": true
+    }
+  },
+  "hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "baseline_hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "panels_to_inspect_relative_to_run": [
+    "review/inputs/view_0_12540311.png",
+    "review/inputs/view_1_12933256.png",
+    "review/inputs/view_2_17736741.png",
+    "review/inputs/view_3_23007103.png",
+    "review/spline/centerline_0_with_masks.png",
+    "review/spline/centerline_1_with_masks.png",
+    "review/spline/centerline_2_with_masks.png",
+    "review/spline/centerline_3_with_masks.png",
+    "review/tube/silhouette_0_unmasked.png",
+    "review/tube/silhouette_1_unmasked.png",
+    "review/tube/silhouette_2_unmasked.png",
+    "review/tube/silhouette_3_unmasked.png",
+    "review/tube/tube_scale_views.png",
+    "review/spline/residual_histograms.png",
+    "review/inputs/depth_offset.png",
+    "review/colors/visibility_and_fill.png",
+    "review/hybrid/combined.png",
+    "review/spline/fit_3d.png",
+    "review/spline/curvature_profile.png",
+    "review/spline/round_curves.png",
+    "review/tube/tube_profiles.png"
+  ],
+  "owner_prior_approval": "pending"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames.json
new file mode 100644
index 0000000000000000000000000000000000000000..526382c1ea3627f2f513f4be95c53bf4fccc4879
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames.json
@@ -0,0 +1,30 @@
+{
+  "source": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/orbit.mp4",
+  "video_frame_count": 36,
+  "fps": 12.0,
+  "selected_frame_indices": [
+    0,
+    4,
+    9,
+    13,
+    18,
+    22,
+    26,
+    31,
+    35
+  ],
+  "extracted_frames_relative_to_review": [
+    "orbit_frames/frame_000.png",
+    "orbit_frames/frame_004.png",
+    "orbit_frames/frame_009.png",
+    "orbit_frames/frame_013.png",
+    "orbit_frames/frame_018.png",
+    "orbit_frames/frame_022.png",
+    "orbit_frames/frame_026.png",
+    "orbit_frames/frame_031.png",
+    "orbit_frames/frame_035.png"
+  ],
+  "visual_inspection_pending": false,
+  "visual_inspection_completed": true,
+  "method": "Individual actual image reads of the nine extracted frames."
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_000.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_000.png
new file mode 100644
index 0000000000000000000000000000000000000000..f2808db810012b09a09c9a9112ef19c81cb34003
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_000.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8552555081f4591e55bbe2c1748c43c2a357ceb9a638d13a34fc152a46126465
+size 118183
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_004.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_004.png
new file mode 100644
index 0000000000000000000000000000000000000000..98a242ca4fcf298be451515a47373b0a4dbaa88f
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_004.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7654e31b971b88e2335b6f460f15836fa9cc6b7ff80c87b61ce5df0af300af0e
+size 158159
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_009.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_009.png
new file mode 100644
index 0000000000000000000000000000000000000000..a89b0cb35b863c683c85683a6f3201b5a1d7ac4e
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_009.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc0a95dde70252fb055f7259ff8cecf9bea0f95cf803b0345bc5da17e182e7c3
+size 132916
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_013.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_013.png
new file mode 100644
index 0000000000000000000000000000000000000000..46ae27caca26cd09fcb80b69dadc7f0de2a145f6
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_013.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:511695ba725e00a9dc766822fb79910be01db0d4188f0b77359c00ef58e67007
+size 154301
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_018.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_018.png
new file mode 100644
index 0000000000000000000000000000000000000000..330b7e3c79a11d6e0d0ab7b6d57612781133feb7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_018.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf420997936bbd21befdca46380c27cb1a8eb248740f5ce408cbd43a0622b1f7
+size 131398
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_022.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_022.png
new file mode 100644
index 0000000000000000000000000000000000000000..b2c873b12a305faec798d9da77a4bf9cd7f94ab5
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_022.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a52034227f2d3bfc7de34843de252efc100980975fdbe1ea48b3388c88a135e
+size 156712
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_026.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_026.png
new file mode 100644
index 0000000000000000000000000000000000000000..1f321b4e5a69a3143a395dec8f3ffca000a8a6db
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_026.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40f2e35a4fe0401070b78173f740fc76b07b722a924b12a635e7a62ff9edd4d8
+size 141810
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_031.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_031.png
new file mode 100644
index 0000000000000000000000000000000000000000..a61a1ed537f94f2425c78eb4c3488fbf017b1cbd
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_031.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41d313f3df6468c0b08813daafcaabee1943d84ed745c16909cf3b54584b89c8
+size 154585
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_035.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_035.png
new file mode 100644
index 0000000000000000000000000000000000000000..54863ca76751f8355242494e954aec4c1a5459d2
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/orbit_frames/frame_035.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24b449e16c2f2c4b2a8dfbc0eac1764bee0364857b8be92efcd653db295e7fa8
+size 139680
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/prior_metrics.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/prior_metrics.json
new file mode 100644
index 0000000000000000000000000000000000000000..e6801d7bc9d5751fac4a53d318d91fd8bacee0ff
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_chain_stronger_reprojection_v2/prior_metrics.json
@@ -0,0 +1,491 @@
+{
+  "session_id": "session_20260921_233144_367389",
+  "prior_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2",
+  "owner_prior_approval": "pending",
+  "visual_quality_assessment": "Completed; see AGENT_FINDINGS.md and agent_visual_review.json. Prior acceptance pending; earlier chain acceptances remain revoked.",
+  "object": {
+    "name": "bike_chain",
+    "representation": "tube",
+    "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+    "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+    "radius_m": 0.00475,
+    "ribbon_width_m": null,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "u_resolution": null,
+    "overlap_rule": "surface_intersection",
+    "twist_penalty": null
+  },
+  "initial_grid_sample": 0,
+  "views": [
+    {
+      "serial": "12540311",
+      "view_index": 0,
+      "selected_window_local_index": 1,
+      "converted_frame": 269,
+      "native_container_frame": 270,
+      "official_frame": 269,
+      "depth_frame": 269,
+      "capture_timestamp_ns": 1790047938734565100,
+      "device_timestamp_ns": 1790047938691799126,
+      "initial_grid_offset_ms": -15.595843,
+      "native_count": 406,
+      "initial_observed_depth_points": 8480,
+      "median_3d_residual_m": 0.0074629574082791805,
+      "p90_3d_residual_m": 0.011093903612345458,
+      "visible_centerline_inside_mask_fraction": 0.9933774834437086,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          160,
+          202,
+          253,
+          293,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1602,
+          "hand_A": 3398,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6356,
+          0.7556093596059114,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          0,
+          1177
+        ],
+        "paired_common_min_max": [
+          3195,
+          4564
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/spline/centerline_0_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_0_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_0_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/inputs/view_0_12540311.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/12540311_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/paired_depth.png"
+      }
+    },
+    {
+      "serial": "12933256",
+      "view_index": 1,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938718969257,
+      "device_timestamp_ns": 1790047938670420020,
+      "initial_grid_offset_ms": 0.0,
+      "native_count": 406,
+      "initial_observed_depth_points": 6557,
+      "median_3d_residual_m": 0.0061066653579473495,
+      "p90_3d_residual_m": 0.010635773651301864,
+      "visible_centerline_inside_mask_fraction": 0.9666666666666667,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          202,
+          214,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1867,
+          "hand_A": 3133,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6628,
+          0.9445802955665025,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          1159,
+          1267
+        ],
+        "paired_common_min_max": [
+          1777,
+          4374
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/spline/centerline_1_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_1_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_1_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/inputs/view_1_12933256.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/12933256_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/paired_depth.png"
+      }
+    },
+    {
+      "serial": "17736741",
+      "view_index": 2,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938706187642,
+      "device_timestamp_ns": 1790047938663525231,
+      "initial_grid_offset_ms": 12.781615,
+      "native_count": 406,
+      "initial_observed_depth_points": 6783,
+      "median_3d_residual_m": 0.009259557351469994,
+      "p90_3d_residual_m": 0.012170043773949146,
+      "visible_centerline_inside_mask_fraction": 0.9936305732484076,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          119,
+          121,
+          151,
+          202,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1348,
+          "hand_A": 3652,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8368,
+          0.953579802955665,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          386,
+          897
+        ],
+        "paired_common_min_max": [
+          3762,
+          4527
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/spline/centerline_2_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_2_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_2_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/inputs/view_2_17736741.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/17736741_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/paired_depth.png"
+      }
+    },
+    {
+      "serial": "23007103",
+      "view_index": 3,
+      "selected_window_local_index": 1,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938720158122,
+      "device_timestamp_ns": 1790047938674181398,
+      "initial_grid_offset_ms": -1.188865,
+      "native_count": 407,
+      "initial_observed_depth_points": 4519,
+      "median_3d_residual_m": 0.009861292317509651,
+      "p90_3d_residual_m": 0.012667492963373663,
+      "visible_centerline_inside_mask_fraction": 0.9860335195530726,
+      "track_diagnostics": {
+        "T": 407,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          152,
+          203,
+          217,
+          253,
+          304,
+          355,
+          406
+        ],
+        "source_seed_counts": {
+          "object": 1992,
+          "hand_A": 3008,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8322,
+          0.9410083538083538,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          911,
+          1267
+        ],
+        "paired_common_min_max": [
+          2618,
+          4179
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/spline/centerline_3_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_3_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/silhouette_3_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/inputs/view_3_23007103.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/23007103_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/paired_depth.png"
+      }
+    }
+  ],
+  "recipe": {
+    "depth_source": "waft_stereo_dav2l5",
+    "mask_erosion_px": 2,
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "max_rounds": 400,
+    "reprojection": {
+      "weight": 8.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1"
+    },
+    "solver": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit": {
+    "stage1_rounds": 1,
+    "stage1_stop": "gradient_tolerance",
+    "stage1_final": null,
+    "twist": null,
+    "final": {
+      "loss": 9.294429037254304e-05,
+      "components": {
+        "distance": 2.6436840926180594e-05,
+        "curvature": 1.2765701740136137e-07,
+        "arclength": 2.086917220367468e-06,
+        "mask_reprojection_2d": 6.429287168430164e-05
+      },
+      "grad_norm": 7.992162863956764e-05,
+      "arclength_m": 0.5895682573318481,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 39.227989196777344,
+      "curvature_p50_p90_per_m": [
+        9.342434406280518,
+        21.996375656127924
+      ],
+      "curvature_above_max_fraction": 0.48,
+      "mean_distance_m": 0.007907245111366833,
+      "max_distance_m": 0.01780632697045803,
+      "beyond_2delta_fraction": 0.2766619841299973,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.0074629574082791805,
+          "p90_m": 0.011093903612345458,
+          "max_m": 0.01485209446400404,
+          "beyond_2delta_fraction": 0.1919811320754717,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0061066653579473495,
+          "p90_m": 0.010635773651301864,
+          "max_m": 0.01780632697045803,
+          "beyond_2delta_fraction": 0.11926185755680951,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.009259557351469994,
+          "p90_m": 0.012170043773949146,
+          "max_m": 0.015088371001183987,
+          "beyond_2delta_fraction": 0.4017396432257113,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009861292317509651,
+          "p90_m": 0.012667492963373663,
+          "max_m": 0.01604742929339409,
+          "beyond_2delta_fraction": 0.47621155122814784,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.9733169029086446
+    },
+    "arclength_soft_bounds_m": [
+      0.5830000042915344,
+      0.5849999785423279
+    ],
+    "arclength_outside_soft_bounds_m": 0.004568278789520264
+  },
+  "geometry": {
+    "vertex_count": 1728,
+    "triangle_count": 3456,
+    "checks": {
+      "euler_characteristic": 0,
+      "watertight": true,
+      "components": 1,
+      "winding_consistent": true,
+      "signed_volume_m3": 3.988753950276397e-05,
+      "minimum_outward_normal_dot_m": 0.0045789361996072244,
+      "degenerate_triangles": 0,
+      "self_intersecting": false,
+      "maximum_radius_error_m": 7.36314771567631e-09,
+      "maximum_diameter_error_m": 1.4711541323694788e-08,
+      "bishop_seam_error": 5.960464477539063e-08,
+      "minimum_nonadjacent_centerline_distance_m": 0.006632810951101171,
+      "triangle_area_m2": {
+        "min": 3.3363548276930293e-06,
+        "median": 4.978674734057986e-06,
+        "mean": 5.031806731543179e-06,
+        "max": 6.320388433778611e-06,
+        "density_max_over_min": 1.8943993550437004
+      },
+      "bounds_m": [
+        [
+          -0.03722784295678139,
+          -0.17418868839740753,
+          -0.0011078841052949429
+        ],
+        [
+          0.17201127111911774,
+          -0.0006560203619301319,
+          0.01636137068271637
+        ]
+      ],
+      "radius_m": 0.00475,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "overlap_rule": "surface_intersection",
+      "open3d_reported_intersecting_pairs": 0,
+      "confirmed_intersecting_pairs": [],
+      "passed": true,
+      "orientation_flipped": true
+    }
+  },
+  "appearance_seed_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "vertex_colors": {
+    "observed_vertices": 1149,
+    "filled_vertices": 579
+  },
+  "mechanical_population": {
+    "particle_count": null,
+    "total_unit_mass": null,
+    "status": "not produced by the prior run; no post-prior mechanics are evaluated here"
+  },
+  "source_code": {
+    "commit": "98ebffe13629edeb0a9b29df973bebc5730d7026",
+    "u_pin": "81c718790a37e5e0102eb77af2c6edd34a9db25f"
+  },
+  "media": {
+    "full_rotation": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/orbit.mp4",
+    "geometry_depth_cloud": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/tube/tube_scale_views.png",
+    "depth_offset_diagnostic": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/inputs/depth_offset.png",
+    "residual_histograms": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/spline/residual_histograms.png",
+    "color_visibility_fill": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/colors/visibility_and_fill.png",
+    "hybrid_seed_populations": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/review/hybrid/combined.png",
+    "derived_coverage": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_coverage.png",
+    "derived_30hz_movie": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_30hz_xyz.mp4"
+  },
+  "sources": {
+    "manifest": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/manifest.json",
+    "fit": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/steps/fit.json",
+    "geometry_checks": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/tube/tube_checks.json",
+    "reference_views": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/inputs/reference_views.json",
+    "reprojection": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_stronger_reprojection_v2/spline/reprojection.json",
+    "tracking_summary": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/summary.json",
+    "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json"
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/AGENT_FINDINGS.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/AGENT_FINDINGS.md
new file mode 100644
index 0000000000000000000000000000000000000000..c90c560a3931a4ad408db6b2a1701165a15f63b1
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/AGENT_FINDINGS.md
@@ -0,0 +1,39 @@
+# 233144 — 18-control, lower-curvature initial-prior review
+
+The requested prior is complete and internally consistent. Direct inspection of **21 stable panels and 9 orbit frames** found no material geometry or packaging issue. **Owner prior acceptance remains pending.** This initial-prior assessment does not clear the later sequence for physical fitting: the existing camera-12540311 support gaps, 116 undefined coverage values and isolated derived points remain part of the case evidence.
+
+The actual saved spline has **18 controls** and curvature weight **5e-9**, matching the owner request. Other scientific fit settings match the original prior; the added baseline-run path supports comparison diagnostics. All six preparation stages and the sole packaging/check runner returned zero. Seven initial-prior NPZ families passed finite-array and geometry/color/hybrid correspondence checks. All **27 copied observation/input, filter and step files** remain byte-equal, and the original fit and manifest are preserved. The derived Gaussian GLB was excluded from input copying and its final bytes match the new tube GLB. [Numerical checks](numeric_prior_review.json) · [Actual inspection list](agent_visual_review.json) · [Camera links and saved metrics](PRIOR_REVIEW_c6a_v1.md).
+
+## Comparison with the original prior
+
+| Common diagnostic | Original: 9 controls, 5e-8 | Requested: 18 controls, 5e-9 |
+|---|---:|---:|
+| Mean lifted-point distance to centerline (mm) | 7.369173 | 7.389500 |
+| Mean of camera mean reprojection distances (mm) | 3.698133 | 3.676484 |
+| Fitted length (mm) | 587.726 | 587.702 |
+| Distance above soft upper length bound (mm) | 2.726 | 2.702 |
+| Maximum curvature (/m) | 30.577 | 37.928 |
+
+The mean 3D residual increases slightly while the mean reprojection distance decreases slightly. These small mixed changes do not establish an accuracy improvement. **Total objectives are not directly comparable after changing the curvature weight.** The common diagnostics and centerline-in-mask fractions are retained in [the comparison JSON](baseline_comparison.json).
+
+The fit stopped after one round on gradient tolerance, with gradient norm **9.39435e-05** and a solver limit of 25 iterations per round. Its length is **2.702 mm above** the unchanged soft 585.000 mm upper bound. The 10/m curvature onset and 80/m smooth saturation scale define a soft penalty, not hard curvature bounds. No refit, depth adjustment or scientific correction was performed during this review.
+
+## Initial geometry and visual evidence
+
+The four camera sets show the projected centerline and tube broadly following exposed chain arcs. A substantial section is hidden by the gripping hand/forearm in cameras 12540311 and 17736741; the closed prior infers that arc. The unmasked tube overlay across the hand is not direct observed confirmation. The continuous tube smooths individual links and sharp corners. Visible centerline-in-mask fractions are **99.01%, 96.66%, 98.41%, 98.60%** in camera order 12540311, 12933256, 17736741, 23007103; these are not silhouette IoU.
+
+The 3D, scale and profile panels show one coherent closed loop with gentle vertical rise on one arc. The observations retain depth spread and a gap at the hidden arc. Maximum saved point-to-centerline residual is **17.994 mm**. The read-only object-minus-local-table medians span **8.889–11.941 mm** without correction. Residuals are not ground-truth reconstruction errors, and the profile/3D vertical axes are expanded.
+
+The tube has **144 rings, 1,728 vertices and 3,456 triangles**, unchanged **4.75 mm radius**, step_t 0.125 and 30-degree circumferential sampling. Checks report one watertight component, consistent winding, no degenerate triangles and **zero confirmed surface-intersection pairs**. Recomputed maximum radius error is **7.56e-09 m**, seam error **7.45e-09 m**, and height bounds **0.026–17.292 mm**. The unchanged chain rule uses confirmed surface intersections; the 6.706 mm minimum nonadjacent-centerline distance is retained as a separate diagnostic.
+
+Orbit frames **0, 4, 9, 13, 18, 22, 26, 31 and 35 of 36** were directly inspected. The point/ring visualization shows a coherent loop from distributed azimuths, with no gross crossing, break or seam opening. [Rotation](../prior_chain_m18_lowcurvature_v1/review/tube/orbit.mp4) · [Inspected frames](orbit_frames.json). The single-round convergence line plot appears blank without a marker; numerical fit and rounds.jsonl records preserve its values.
+
+Colors are observed for **1,147 vertices** and median-filled for **581/1,728 (33.62%)**, chiefly along the underside and occluded arc. Color arrays are finite and in range, with exact saved float32-position and vertex-color GLB roundtrip checks. Hybrid initialization contains **26,339 observed + 103,680 mesh-derived = 130,019 seeds**. Observed points/colors exactly match the saved filtered cloud. Keeping step_t unchanged while doubling controls doubles the rings, triangles and mesh-derived seed count; the baseline population was **26,339 + 51,840 = 78,179**. These are appearance seeds, not mechanical particles or mass.
+
+## Preserved later-sequence limits
+
+The [existing reference review](../review_v1/REVIEW.md) records **zero admitted object support in camera 12540311 at native-relative rows 143–238, 240, 243 and 262–326**. Its **116 undefined object-mask coverage values remain NaNs**; they are not zeros or passing coverage. The finite coverage range is 0–0.900 with median 0.479. Magenta hidden predictions over the hand are not admitted measurements.
+
+That review also records isolated derived points below or beside the main loop during middle/lowering stages, including samples 189/216/243/351. Those points, support gaps and validity states remain unchanged. [Retained reference limits](retained_reference_limits.json) preserves their source attribution. This initial-prior inspection does not re-review or clear every later frame, establish repeated-link identity/contact, or authorize physical fitting.
+
+Only this new packet and scoped review scratch were written. The original observations, saved fits, case status and approval records were preserved. Owner prior acceptance and any dependent preparation or training remain separate.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/PRIOR_REVIEW_c6a_v1.md b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/PRIOR_REVIEW_c6a_v1.md
new file mode 100644
index 0000000000000000000000000000000000000000..43582f36c88aefc8abfc18d1c06f07b2b8e2a18a
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/PRIOR_REVIEW_c6a_v1.md
@@ -0,0 +1,45 @@
+> **Current status:** owner prior acceptance revoked/on hold on 2026-10-28. Shared chain fitting settings are under review; downstream preparation is held. Earlier review and approval statements below are historical.
+
+# session_20260921_233144_367389 — prior accepted
+
+Saved C2 **bike_chain** prior evidence (radius 4.75 mm). Agent visual inspection is complete; [findings, comparison and retained sequence limits](AGENT_FINDINGS.md) cover 21 stable panels and 9 orbit frames. The owner accepted this prior on 2026-10-28. Later support gaps and undefined coverage values remain preserved; initial-prior review does not clear later physical fitting.
+
+Initial reference frames are the saved nearest-camera observations of the first 30 Hz sample. The local row, converted frame, native-container frame and capture clock remain distinct in [the metrics JSON](<prior_metrics.json>).
+
+| Camera | Local / converted / native | Centerline / geometry / native input |
+|---|---|---|
+| 12540311 | 1 / 269 / 270 | [centerline](<../prior_chain_m18_lowcurvature_v1/review/spline/centerline_0_with_masks.png>) · [tube](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_0_unmasked.png>) · [geometry + mask](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_0_with_masks.png>) · [RGB/depth](<../prior_chain_m18_lowcurvature_v1/review/inputs/view_0_12540311.png>) |
+| 12933256 | 0 / 268 / 269 | [centerline](<../prior_chain_m18_lowcurvature_v1/review/spline/centerline_1_with_masks.png>) · [tube](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_1_unmasked.png>) · [geometry + mask](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_1_with_masks.png>) · [RGB/depth](<../prior_chain_m18_lowcurvature_v1/review/inputs/view_1_12933256.png>) |
+| 17736741 | 0 / 268 / 269 | [centerline](<../prior_chain_m18_lowcurvature_v1/review/spline/centerline_2_with_masks.png>) · [tube](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_2_unmasked.png>) · [geometry + mask](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_2_with_masks.png>) · [RGB/depth](<../prior_chain_m18_lowcurvature_v1/review/inputs/view_2_17736741.png>) |
+| 23007103 | 1 / 268 / 269 | [centerline](<../prior_chain_m18_lowcurvature_v1/review/spline/centerline_3_with_masks.png>) · [tube](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_3_unmasked.png>) · [geometry + mask](<../prior_chain_m18_lowcurvature_v1/review/tube/silhouette_3_with_masks.png>) · [RGB/depth](<../prior_chain_m18_lowcurvature_v1/review/inputs/view_3_23007103.png>) |
+
+[Full rotation video](<../prior_chain_m18_lowcurvature_v1/review/tube/orbit.mp4>) · [Geometry and depth cloud](<../prior_chain_m18_lowcurvature_v1/review/tube/tube_scale_views.png>) · [Residual histograms](<../prior_chain_m18_lowcurvature_v1/review/spline/residual_histograms.png>) · [Depth offset diagnostic](<../prior_chain_m18_lowcurvature_v1/review/inputs/depth_offset.png>)
+
+Saved fit: **1 stage-1 rounds**, stop `gradient_tolerance`; final length **587.702 mm**, maximum curvature **37.928/m**, final gradient norm **9.39435e-05**. Soft length bounds are 583.000–585.000 mm; distance outside those bounds is 2.702 mm.
+
+Saved geometry checks report `passed=True`: **1728 mesh vertices / 3456 triangles**, 1 component(s), `watertight=True`, `self_intersecting=False`. [Full geometry checks](<../prior_chain_m18_lowcurvature_v1/tube/tube_checks.json>).
+
+The following 3D residuals are lifted-depth point distances to the **centerline**, not measured reconstruction errors. The centerline-in-mask fraction is a separate saved diagnostic; it is not silhouette IoU or a visual quality decision.
+
+| Camera | Median / p90 3D residual (mm) | Visible centerline inside mask | Native depth and visibility evidence |
+|---|---|---|---|
+| 12540311 | 6.872 / 10.085 | 99.01% | [Full native video](<../review_v1/12540311/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12540311/12540311_visibility_certainty.png>) · [Support/rejections](<../review_v1/12540311/support_and_rejections.png>) · [Paired depth](<../review_v1/12540311/paired_depth.png>) |
+| 12933256 | 5.860 / 10.534 | 96.66% | [Full native video](<../review_v1/12933256/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/12933256/12933256_visibility_certainty.png>) · [Support/rejections](<../review_v1/12933256/support_and_rejections.png>) · [Paired depth](<../review_v1/12933256/paired_depth.png>) |
+| 17736741 | 8.450 / 11.143 | 98.41% | [Full native video](<../review_v1/17736741/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/17736741/17736741_visibility_certainty.png>) · [Support/rejections](<../review_v1/17736741/support_and_rejections.png>) · [Paired depth](<../review_v1/17736741/paired_depth.png>) |
+| 23007103 | 9.256 / 12.147 | 98.60% | [Full native video](<../review_v1/23007103/rgb_waft_depth_full_window.mp4>) · [Visibility/certainty](<../review_v1/23007103/23007103_visibility_certainty.png>) · [Support/rejections](<../review_v1/23007103/support_and_rejections.png>) · [Paired depth](<../review_v1/23007103/paired_depth.png>) |
+
+The saved statistical filter used 20 neighbours / 2 sigma and kept **26339 of 27239** eroded points. The run used `waft_stereo_dav2l5`, 2 px erosion, a 400-round budget and reprojection weight **4**.
+
+Appearance initialization contains **26339 observed + 103680 mesh-derived = 130019 seeds**. Vertex colors are observed for 1147 mesh vertices and median-filled for 581. Mesh vertices and appearance seeds are not mechanical particles. No post-prior mechanical population or unit-mass total is supplied by this prior run; trained appearance and physical fitting are separate stages.
+
+[Appearance seed populations](<../prior_chain_m18_lowcurvature_v1/review/hybrid/combined.png>) · [Observed and filled colors](<../prior_chain_m18_lowcurvature_v1/review/colors/visibility_and_fill.png>) · [30 Hz observation coverage](<../review_v1/derived/derived_coverage.png>)
+
+[Full saved 30 Hz observations](<../review_v1/derived/derived_30hz_xyz.mp4>)
+
+[Manifest](<../prior_chain_m18_lowcurvature_v1/manifest.json>) · [Fit settings and metrics](<../prior_chain_m18_lowcurvature_v1/steps/fit.json>) · [Reference observations](<../prior_chain_m18_lowcurvature_v1/inputs/reference_views.json>) · [Tracking/lift summary](<../review_v1/summary.json>)
+
+The completed initial-prior review is recorded in [agent findings](AGENT_FINDINGS.md) and [the exact inspection list](agent_visual_review.json). The [existing reference review](../review_v1/REVIEW.md) retains later support gaps and isolated derived points. Owner prior approval remains pending. No case status, prior acceptance, mechanics approval or trained-appearance approval is changed here.
+
+## Owner prior decision
+
+On 2026-10-28 the owner accepted this exact18-control prior for mechanical inputs and static appearance training. Appearance and physical fitting remain separate.
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/agent_visual_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/agent_visual_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..7f9886980fe6447dbdd524ba6bf1e983504477c5
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/agent_visual_review.json
@@ -0,0 +1,203 @@
+{
+  "scope": "233144 initial M18/lower-curvature saved prior only; later sequence limitations are carried forward from the existing reference review.",
+  "inspection_completed_utc": "2026-10-28T22:18:32.447628+00:00",
+  "stable_panels": [
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_0_12540311.png",
+      "inspected": true,
+      "finding": "Mask follows the exposed chain loop on the table; the gripping hand/forearm is covered by the hand union. No broad red invalid-depth region is visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_1_12933256.png",
+      "inspected": true,
+      "finding": "Mask follows the exposed chain loop on the table; the gripping hand/forearm is covered by the hand union. No broad red invalid-depth region is visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_2_17736741.png",
+      "inspected": true,
+      "finding": "Mask follows the exposed chain loop on the table; the gripping hand/forearm is covered by the hand union. No broad red invalid-depth region is visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_3_23007103.png",
+      "inspected": true,
+      "finding": "Mask follows the exposed chain loop on the table; the gripping hand/forearm is covered by the hand union. No broad red invalid-depth region is visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_0_with_masks.png",
+      "inspected": true,
+      "finding": "Smooth closed projected centerline follows exposed chain arcs. Closure passes through the hand/forearm mask and is inferred in the hidden region. A substantial lower/near arc is hidden in this view."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_1_with_masks.png",
+      "inspected": true,
+      "finding": "Smooth closed projected centerline follows exposed chain arcs. Closure passes through the hand/forearm mask and is inferred in the hidden region."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_2_with_masks.png",
+      "inspected": true,
+      "finding": "Smooth closed projected centerline follows exposed chain arcs. Closure passes through the hand/forearm mask and is inferred in the hidden region. A substantial lower/near arc is hidden in this view."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_3_with_masks.png",
+      "inspected": true,
+      "finding": "Smooth closed projected centerline follows exposed chain arcs. Closure passes through the hand/forearm mask and is inferred in the hidden region."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_0_unmasked.png",
+      "inspected": true,
+      "finding": "Continuous tube broadly follows exposed chain location and width, smoothing link structure and corners. The unmasked overlay draws inferred tube across the hand; that section is not observed confirmation."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_1_unmasked.png",
+      "inspected": true,
+      "finding": "Continuous tube broadly follows exposed chain location and width, smoothing link structure and corners. The unmasked overlay draws inferred tube across the hand; that section is not observed confirmation."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_2_unmasked.png",
+      "inspected": true,
+      "finding": "Continuous tube broadly follows exposed chain location and width, smoothing link structure and corners. The unmasked overlay draws inferred tube across the hand; that section is not observed confirmation."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_3_unmasked.png",
+      "inspected": true,
+      "finding": "Continuous tube broadly follows exposed chain location and width, smoothing link structure and corners. The unmasked overlay draws inferred tube across the hand; that section is not observed confirmation."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/tube_scale_views.png",
+      "inspected": true,
+      "finding": "Closed loop with low height variation. Observations remain spread around and above the tube, with a visible gap at the occluded arc; fixed 9.5 mm diameter retained."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/residual_histograms.png",
+      "inspected": true,
+      "finding": "Moderate residual tails remain in all views, with views 2/3 concentrated near 8\u201310 mm and view 1 extending toward 18 mm. These are lifted-point-to-centerline distances, not ground-truth errors."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/depth_offset.png",
+      "inspected": true,
+      "finding": "Read-only object-minus-local-table medians range 8.889\u201311.941 mm; no correction applied."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/colors/visibility_and_fill.png",
+      "inspected": true,
+      "finding": "1,147 observed-color vertices and 581 median-filled vertices; fill is concentrated along underside and the occluded upper arc."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/hybrid/combined.png",
+      "inspected": true,
+      "finding": "26,339 observed and 103,680 mesh-derived seeds follow the same closed loop. Observed depth spread and the hidden-arc gap remain distinct from the continuous mesh population."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/fit_3d.png",
+      "inspected": true,
+      "finding": "18-control centerline threads the observed loop and bridges the occluded arc. Some gentle vertical rise is visible; z display is expanded."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/curvature_profile.png",
+      "inspected": true,
+      "finding": "Localized smooth peaks, maximum 37.928/m, under the plotted 80/m smooth saturation scale. The 10/m onset is a soft penalty threshold."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/round_curves.png",
+      "inspected": true,
+      "finding": "Single-round line plot appears blank without a marker; saved numeric fit and rounds.jsonl provide the values."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/tube_profiles.png",
+      "inspected": true,
+      "finding": "Closed loop with rounded corners and gentle height rise on one arc. Fixed 9.5 mm diameter; expanded vertical display. No gross crossing, break or twist visible."
+    }
+  ],
+  "stable_panel_count": 21,
+  "orbit_frames": [
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_000.png",
+      "frame_index": 0,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_004.png",
+      "frame_index": 4,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_009.png",
+      "frame_index": 9,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_013.png",
+      "frame_index": 13,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_018.png",
+      "frame_index": 18,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_022.png",
+      "frame_index": 22,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_026.png",
+      "frame_index": 26,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_031.png",
+      "frame_index": 31,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    },
+    {
+      "path": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_review_m18_lowcurvature_v1/orbit_frames/frame_035.png",
+      "frame_index": 35,
+      "inspected": true,
+      "finding": "Coherent single closed loop from this azimuth, rounded corners and gentle vertical variation; no gross crossing, disconnection or seam opening visible."
+    }
+  ],
+  "orbit_frame_count": 9,
+  "orbit_inspection_pending": false,
+  "owner_prior_approval": "pending",
+  "assessment": "No material geometry or packaging issue found in this bounded initial-prior review; later sequence is not cleared for physical fitting.",
+  "retained_reference_limits": {
+    "source": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/REVIEW.md",
+    "source_findings": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/agent_findings.json",
+    "additional_diagnostics": {
+      "12540311_zero_object_native_relative_spans_inclusive": [
+        [
+          143,
+          238
+        ],
+        [
+          240,
+          240
+        ],
+        [
+          243,
+          243
+        ],
+        [
+          262,
+          326
+        ]
+      ],
+      "12540311_coverage_nan_count": 116,
+      "12540311_finite_coverage_min_median_max": [
+        0.0,
+        0.47863454,
+        0.8996283
+      ]
+    },
+    "interpretation": "Later camera12540311 support gaps,116 undefined coverage values and isolated derived points remain; initial-prior review does not clear later sequence for physical fitting."
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/baseline_comparison.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/baseline_comparison.json
new file mode 100644
index 0000000000000000000000000000000000000000..0108862e100579240c9fa2e771ca6766f0246452
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/baseline_comparison.json
@@ -0,0 +1,49 @@
+{
+  "baseline_controls": 9,
+  "new_controls": 18,
+  "baseline_curvature_weight": 5e-08,
+  "new_curvature_weight": 5e-09,
+  "common_metrics": {
+    "mean_point_to_centerline_distance_m": {
+      "baseline": 0.0073691726396086865,
+      "new": 0.007389499891093807
+    },
+    "mean_of_view_mean_reprojection_distances_m": {
+      "baseline": 0.003698133051579657,
+      "new": 0.0036764842167760453
+    },
+    "arclength_m": {
+      "baseline": 0.5877262949943542,
+      "new": 0.5877017974853516
+    },
+    "max_curvature_per_m": {
+      "baseline": 30.576679229736328,
+      "new": 37.92765808105469
+    }
+  },
+  "baseline_visible_centerline_inside_mask": [
+    0.9902597402597403,
+    0.9722222222222222,
+    0.9874213836477987,
+    0.9888268156424581
+  ],
+  "new_visible_centerline_inside_mask": [
+    0.9901315789473685,
+    0.9665738161559888,
+    0.9841269841269842,
+    0.9860335195530726
+  ],
+  "total_objective_comparison": "Not directly comparable because the curvature weight changed.",
+  "baseline_hybrid_counts": {
+    "observed": 26339,
+    "mesh": 51840,
+    "total": 78179,
+    "mesh_to_observed": 1.968184061657618
+  },
+  "new_hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/numeric_prior_review.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/numeric_prior_review.json
new file mode 100644
index 0000000000000000000000000000000000000000..28e10be0ab8e7169b459fe2f6d3522ddb4dc9073
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/numeric_prior_review.json
@@ -0,0 +1,771 @@
+{
+  "scope": "Saved 233144 M18/lower-curvature prior review only; no fitting, sampling or scientific correction executed by this reviewer.",
+  "stage_results": {
+    "fit": {
+      "returncode": 0,
+      "elapsed_s": 184.04458521200286
+    },
+    "tube": {
+      "returncode": 0,
+      "elapsed_s": 25.636197654996067
+    },
+    "colors": {
+      "returncode": 0,
+      "elapsed_s": 32.876061360002495
+    },
+    "hybrid": {
+      "returncode": 0,
+      "elapsed_s": 7.612649925009464
+    },
+    "review": {
+      "returncode": 0,
+      "elapsed_s": 24.963562548000482
+    },
+    "manifest": {
+      "returncode": 0,
+      "elapsed_s": 6.891923165996559
+    }
+  },
+  "saved_fit_identity_unchanged_during_review": true,
+  "actual_settings_match_owner_request_and_other_baseline_settings": true,
+  "baseline_fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 6.0984086303506047e-05,
+      "components": {
+        "distance": 2.398061587882694e-05,
+        "curvature": 1.0113907364939223e-06,
+        "arclength": 7.432801680806733e-07,
+        "mask_reprojection_2d": 3.524879866745323e-05
+      },
+      "grad_norm": 8.26684627099894e-05,
+      "arclength_m": 0.5877262949943542,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 30.576679229736328,
+      "curvature_p50_p90_per_m": [
+        11.315874099731445,
+        22.88189315795898
+      ],
+      "curvature_above_max_fraction": 0.596,
+      "mean_distance_m": 0.0073691726396086865,
+      "max_distance_m": 0.01859072595834732,
+      "beyond_2delta_fraction": 0.19199665894680892,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.006901396904140711,
+          "p90_m": 0.010149775166064502,
+          "max_m": 0.014187222346663475,
+          "beyond_2delta_fraction": 0.11014150943396227,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.005649127997457981,
+          "p90_m": 0.01055502500385046,
+          "max_m": 0.01859072595834732,
+          "beyond_2delta_fraction": 0.11056885770931829,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008469968102872372,
+          "p90_m": 0.011138298735022546,
+          "max_m": 0.013993453234434128,
+          "beyond_2delta_fraction": 0.26020934689665337,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.00907266978174448,
+          "p90_m": 0.012268818728625775,
+          "max_m": 0.016254890710115433,
+          "beyond_2delta_fraction": 0.36136313343660104,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.9808226845736205
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 9,
+      "curvature_weight": 5e-08,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5824263095855713,
+      "mean_centerline_distance_to_truth_m": 0.003666582156588717,
+      "max_centerline_distance_to_truth_m": 0.0065246143314547185,
+      "corner_truth_to_fit_mean_max_m": [
+        0.004143985746667632,
+        0.00652506213845758
+      ],
+      "straight_truth_to_fit_mean_m": 0.0035861119631459973,
+      "fit_max_curvature_per_m": 104.8658676147461,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        44.93003845214844,
+        104.8658676147461
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "6ae821f556bafc3087c9465198c48b7cb71a0900def108c67dd6f77a83bf469c",
+      "spline/residuals.npz": "93a21e8964e40e297d636b0bc6f05a996afdfee6aed84aeef4abac28be2c211a",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "54fe1bc122eeb141e50b44ab2e31363746cadcfdd4e5233a1b5c1b3aa4a45832",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "da22c7756a7715e3b2fbfbd7ac9dd9f6168f1a9b8d66be670636199e1990ce7f",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "8f912477f655e0d4e5a6d5e154765bb2582aea55a53373728078767228420189",
+      "spline/rounds.jsonl": "76ae11e7bab8c081d9a19eaeceaec464720ca0ce6c547579fb5569be2ebf464e",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "da0fbbe4aaa65082603b4e4a92b62db1b135dc95ca20d838db25afa46fef6e99"
+    }
+  },
+  "derived_geometry_check": {
+    "path": "gaussian_data/session_20260921_233144_367389/shape_prior.glb",
+    "excluded_from_initial_copy": true,
+    "final_bytes_equal_new_tube_glb": true
+  },
+  "fit_sha256": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455",
+  "baseline_fit_and_manifest_unchanged": true,
+  "reused_input_bytes_equal": {
+    "inputs/reference_views.json": true,
+    "inputs/observed_cloud.npz": true,
+    "inputs/depth_offset_diagnostic.json": true,
+    "gaussian_data/session_20260921_233144_367389/mask_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/0.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_3.png": true,
+    "gaussian_data/session_20260921_233144_367389/camera_meta.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/2.png": true,
+    "gaussian_data/session_20260921_233144_367389/2_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/mask_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/0_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_0.png": true,
+    "gaussian_data/session_20260921_233144_367389/3_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/interp_poses.pkl": true,
+    "gaussian_data/session_20260921_233144_367389/observation.ply": true,
+    "gaussian_data/session_20260921_233144_367389/1_depth.npy": true,
+    "gaussian_data/session_20260921_233144_367389/mask_1.png": true,
+    "gaussian_data/session_20260921_233144_367389/1.png": true,
+    "gaussian_data/session_20260921_233144_367389/mask_human_2.png": true,
+    "gaussian_data/session_20260921_233144_367389/3.png": true,
+    "spline/filter.json": true,
+    "spline/filter_removed.npz": true,
+    "spline/filtered_cloud.npz": true,
+    "steps/inputs.json": true,
+    "steps/diagnose-depth.json": true
+  },
+  "saved_array_checks": {
+    "spline/filtered_cloud.npz": {
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/residuals.npz": {
+      "distance_m": {
+        "shape": [
+          26339
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          26339
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "pixel_uv": {
+        "shape": [
+          26339,
+          2
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "spline/reprojection_residuals.npz": {
+      "pixel_uv": {
+        "shape": [
+          39337,
+          2
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "view": {
+        "shape": [
+          39337
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "distance_m": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "distance_px": {
+        "shape": [
+          39337
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/geometry.npz": {
+      "vertices": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "faces": {
+        "shape": [
+          3456,
+          3
+        ],
+        "dtype": "int64",
+        "finite": true
+      },
+      "centers": {
+        "shape": [
+          144,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "normals": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "tube/triangle_areas.npz": {
+      "area_m2": {
+        "shape": [
+          3456
+        ],
+        "dtype": "float64",
+        "finite": true
+      }
+    },
+    "colors/vertex_colors.npz": {
+      "colors_rgb": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "colors_u8": {
+        "shape": [
+          1728,
+          3
+        ],
+        "dtype": "uint8",
+        "finite": true
+      },
+      "observed_by_view": {
+        "shape": [
+          4,
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      },
+      "filled": {
+        "shape": [
+          1728
+        ],
+        "dtype": "bool",
+        "finite": true
+      }
+    },
+    "hybrid/populations.npz": {
+      "observed_points": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "observed_colors": {
+        "shape": [
+          26339,
+          3
+        ],
+        "dtype": "float64",
+        "finite": true
+      },
+      "mesh_points": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_colors": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      },
+      "mesh_normals": {
+        "shape": [
+          103680,
+          3
+        ],
+        "dtype": "float32",
+        "finite": true
+      }
+    }
+  },
+  "settings": {
+    "object": {
+      "name": "bike_chain",
+      "representation": "tube",
+      "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+      "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+      "radius_m": 0.00475,
+      "ribbon_width_m": null,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "u_resolution": null,
+      "overlap_rule": "surface_intersection",
+      "twist_penalty": null
+    },
+    "max_rounds": 400,
+    "control_count": 18,
+    "relative_loss_tolerance": 1e-07,
+    "reprojection": {
+      "weight": 4.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+    },
+    "loss_config": {
+      "eval_distances": true,
+      "max_saturation_distance": 0.035,
+      "distance_huber_delta": 0.005,
+      "eval_curvature_cost": true,
+      "num_eval_reg_points": 250,
+      "lambda_curvature_cost": 5e-09,
+      "max_curvature": 10.0,
+      "curvature_huber_delta": 30.0,
+      "max_saturation_curvature": 80.0,
+      "eval_curvilinear_energy": false,
+      "lambda_curvilinear_energy": 0.001,
+      "eval_length_constraint": true,
+      "lambda_length_constraint": 0.1,
+      "arclength_bounds": [
+        0.5830000042915344,
+        0.5849999785423279
+      ]
+    },
+    "solver_config": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit_outputs": {
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "rounds": 1,
+    "stop": "gradient_tolerance",
+    "final": {
+      "loss": 5.9778743889182806e-05,
+      "components": {
+        "distance": 2.406455132586416e-05,
+        "curvature": 1.2257775949819916e-07,
+        "arclength": 7.299825597328891e-07,
+        "mask_reprojection_2d": 3.486163404886611e-05
+      },
+      "grad_norm": 9.39435267355293e-05,
+      "arclength_m": 0.5877017974853516,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 37.92765808105469,
+      "curvature_p50_p90_per_m": [
+        9.860498428344727,
+        22.216008949279782
+      ],
+      "curvature_above_max_fraction": 0.492,
+      "mean_distance_m": 0.007389499891093807,
+      "max_distance_m": 0.017994370311498642,
+      "beyond_2delta_fraction": 0.19343938646114128,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.00687211100012064,
+          "p90_m": 0.010084983985871076,
+          "max_m": 0.013782148249447346,
+          "beyond_2delta_fraction": 0.10613207547169812,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0058602094650268555,
+          "p90_m": 0.010533764958381653,
+          "max_m": 0.017994370311498642,
+          "beyond_2delta_fraction": 0.11087387524782676,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008450270630419254,
+          "p90_m": 0.011142997816205025,
+          "max_m": 0.014050017111003399,
+          "beyond_2delta_fraction": 0.2523956951201533,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009255929850041866,
+          "p90_m": 0.012146585062146187,
+          "max_m": 0.015855133533477783,
+          "beyond_2delta_fraction": 0.38858154458951094,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.981192264140135
+    },
+    "deterministic": true,
+    "partial_view_witness": {
+      "source": "synthetic rounded trapezoid, four retained RP2 camera centres",
+      "control_count": 18,
+      "curvature_weight": 5e-09,
+      "seed": 5,
+      "noise_sigma_m": 0.0015,
+      "points": 288,
+      "tube_radius_m": 0.00475,
+      "truth_arclength_m": 0.5839999999999999,
+      "truth_corner_curvature_per_m": 100.0,
+      "fit_arclength_m": 0.5848049521446228,
+      "mean_centerline_distance_to_truth_m": 0.003304384713212965,
+      "max_centerline_distance_to_truth_m": 0.0057414166823648,
+      "corner_truth_to_fit_mean_max_m": [
+        0.0034466403731989066,
+        0.004110262860931054
+      ],
+      "straight_truth_to_fit_mean_m": 0.003288980337124959,
+      "fit_max_curvature_per_m": 123.24449920654297,
+      "fit_curvature_nearest_corners_p50_max_per_m": [
+        49.5619010925293,
+        123.24449920654297
+      ],
+      "rounds": 1,
+      "stop": "gradient_tolerance"
+    },
+    "files": {
+      "spline/reprojection_residuals.npz": "909fa4c5991e924658e230d891b42027f9454e23297b69e7b0f18ee5600faa9a",
+      "spline/residuals.npz": "0de90046aee43ae7251db5e2a174d68c80e3d4317ea3f3a8fec38c76b47e2046",
+      "spline/filtered_cloud.npz": "68e21697a4a03433fbf255ddaae6b84d0556a0027a3e035ff7f3441a61525847",
+      "spline/spline_init.json": "19d1b7bfa714a4afe18a0831dc78c1b42c6d087a91605aaf2ee11fd8ed6af064",
+      "spline/determinism.json": "773d1a02e2924f0d7df2d5d0011979c4e408678ebfdb9633a8348b9788132683",
+      "spline/filter.json": "a64b3b76aed2c7de63cbc747c2aaf37138acf8455e4731761aee615257c0e333",
+      "spline/witness.json": "9021e8d6d6e019fa4adc745ee9b3620ff75852583b523c7918ec6ea9b265ba6a",
+      "spline/filter_removed.npz": "aa8cd5f2c6f83bf06f2d9c22d21e15414e88630bf276b1bf6f00593f36e8c5c2",
+      "spline/spline_fit.json": "32e3eed4983d3822010585d2c629079ed88c8a7d7d21425745da289c6e6b0455",
+      "spline/reprojection_comparison.json": "b40f5414d66a7047642440527c63a08d74c1b63b7bdfc3f3492451de0821fa8a",
+      "spline/rounds.jsonl": "a1dacca4ea7a924588f3b9713279c4834a7a8809e38b13c8334200e93a87dcd5",
+      "spline/init_ellipse.json": "ca18195cdf76514f76c5e7f0c278c3a0d05584929738ac154d92ae7adf5064a9",
+      "spline/reprojection.json": "a2423fa09343e404bf108c2411ed62da67f3dd7477679ebf3d9855e99fe12223"
+    }
+  },
+  "geometry_checks": {
+    "euler_characteristic": 0,
+    "watertight": true,
+    "components": 1,
+    "winding_consistent": true,
+    "signed_volume_m3": 3.976149783233774e-05,
+    "minimum_outward_normal_dot_m": 0.004579434340664218,
+    "degenerate_triangles": 0,
+    "self_intersecting": false,
+    "maximum_radius_error_m": 7.564100575063737e-09,
+    "maximum_diameter_error_m": 1.5116993679412105e-08,
+    "bishop_seam_error": 7.450580596923828e-09,
+    "minimum_nonadjacent_centerline_distance_m": 0.006705765687998025,
+    "triangle_area_m2": {
+      "min": 3.3956602598738355e-06,
+      "median": 4.991684670915842e-06,
+      "mean": 5.015910096568565e-06,
+      "max": 6.254291218252229e-06,
+      "density_max_over_min": 1.841848341590158
+    },
+    "bounds_m": [
+      [
+        -0.0369919128715992,
+        -0.17409859597682953,
+        2.6456546038389206e-05
+      ],
+      [
+        0.1717047244310379,
+        -0.0010159998200833797,
+        0.017292136326432228
+      ]
+    ],
+    "radius_m": 0.00475,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "overlap_rule": "surface_intersection",
+    "open3d_reported_intersecting_pairs": 0,
+    "confirmed_intersecting_pairs": [],
+    "passed": true,
+    "orientation_flipped": true
+  },
+  "actual_geometry_counts": {
+    "rings": 144,
+    "vertices": 1728,
+    "triangles": 3456
+  },
+  "recomputed_max_radius_error_m": 7.564100575063737e-09,
+  "vertex_colors": {
+    "rule": "native sRGB/255, cosine weighted over mask/depth/first-hit visible views",
+    "two_sided": false,
+    "depth_tolerance_m": 0.025,
+    "first_hit_tolerance_m": 0.0002,
+    "per_view": [
+      {
+        "view": 0,
+        "serial": "12540311",
+        "observed_vertices": 625,
+        "projected_inside": 1728,
+        "first_hit": 925
+      },
+      {
+        "view": 1,
+        "serial": "12933256",
+        "observed_vertices": 678,
+        "projected_inside": 1728,
+        "first_hit": 900
+      },
+      {
+        "view": 2,
+        "serial": "17736741",
+        "observed_vertices": 637,
+        "projected_inside": 1728,
+        "first_hit": 919
+      },
+      {
+        "view": 3,
+        "serial": "23007103",
+        "observed_vertices": 733,
+        "projected_inside": 1728,
+        "first_hit": 914
+      }
+    ],
+    "observed_vertices": 1147,
+    "filled_vertices": 581,
+    "fill_color_rgb": [
+      0.20826440167538152,
+      0.1689816457792374,
+      0.1572933793181106
+    ],
+    "glb_roundtrip": {
+      "float32_positions_exact": true,
+      "vertex_colors_exact": true,
+      "uv_count": 0,
+      "vertex_reordered_on_open3d_load": true
+    }
+  },
+  "hybrid_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "baseline_hybrid_counts": {
+    "observed": 26339,
+    "mesh": 51840,
+    "total": 78179,
+    "mesh_to_observed": 1.968184061657618
+  },
+  "panels_to_inspect_relative_to_run": [
+    "review/inputs/view_0_12540311.png",
+    "review/inputs/view_1_12933256.png",
+    "review/inputs/view_2_17736741.png",
+    "review/inputs/view_3_23007103.png",
+    "review/spline/centerline_0_with_masks.png",
+    "review/spline/centerline_1_with_masks.png",
+    "review/spline/centerline_2_with_masks.png",
+    "review/spline/centerline_3_with_masks.png",
+    "review/tube/silhouette_0_unmasked.png",
+    "review/tube/silhouette_1_unmasked.png",
+    "review/tube/silhouette_2_unmasked.png",
+    "review/tube/silhouette_3_unmasked.png",
+    "review/tube/tube_scale_views.png",
+    "review/spline/residual_histograms.png",
+    "review/inputs/depth_offset.png",
+    "review/colors/visibility_and_fill.png",
+    "review/hybrid/combined.png",
+    "review/spline/fit_3d.png",
+    "review/spline/curvature_profile.png",
+    "review/spline/round_curves.png",
+    "review/tube/tube_profiles.png"
+  ],
+  "owner_prior_approval": "pending"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames.json
new file mode 100644
index 0000000000000000000000000000000000000000..bceccbf1154fd731107a3d1375577b8097773277
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames.json
@@ -0,0 +1,29 @@
+{
+  "source": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/orbit.mp4",
+  "video_frame_count": 36,
+  "fps": 12.0,
+  "selected_frame_indices": [
+    0,
+    4,
+    9,
+    13,
+    18,
+    22,
+    26,
+    31,
+    35
+  ],
+  "extracted_frames_relative_to_review": [
+    "orbit_frames/frame_000.png",
+    "orbit_frames/frame_004.png",
+    "orbit_frames/frame_009.png",
+    "orbit_frames/frame_013.png",
+    "orbit_frames/frame_018.png",
+    "orbit_frames/frame_022.png",
+    "orbit_frames/frame_026.png",
+    "orbit_frames/frame_031.png",
+    "orbit_frames/frame_035.png"
+  ],
+  "visual_inspection_pending": false,
+  "visual_inspection_record": "agent_visual_review.json"
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_000.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_000.png
new file mode 100644
index 0000000000000000000000000000000000000000..691846b5cedd0ffc7459903393eba31fe1ced9c1
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_000.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0e2d1878ed7a8222848000b5754e752d389df021e8e6bc25a688c7d17cb0325
+size 118636
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_004.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_004.png
new file mode 100644
index 0000000000000000000000000000000000000000..f877b4fafcaf4228a6c2cdd1ab56d01a4b7741f8
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_004.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a11ba059ce2d8706b0fd03e16c604eab5ce3bb405ae284ff830a09605d99e3e6
+size 156818
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_009.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_009.png
new file mode 100644
index 0000000000000000000000000000000000000000..6de5a81ba7238ae1f1a0d0034e8ad37f094e1489
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_009.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc8d19608e9ff6697cd9bfedaf4182220618af9952c2c6c3c6a7a923f388ff22
+size 131147
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_013.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_013.png
new file mode 100644
index 0000000000000000000000000000000000000000..4487ff93eec8d99760680cbfb7727da4c721969c
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_013.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:115998e9f180686475b6a223d05518439dae93f97cb4d1040c24a5332a2161fe
+size 153832
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_018.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_018.png
new file mode 100644
index 0000000000000000000000000000000000000000..41c7fa6d429d510682f75349faef1e22cd55b927
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_018.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4af9166ef893d57caf51c15e56062d45253f448c5a47b80c72279883094949d4
+size 131953
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_022.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_022.png
new file mode 100644
index 0000000000000000000000000000000000000000..e1d0a4782849c6e52b7dd95081b2d3583a28e1e7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_022.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d27346893ec51b8d3cb8e20b782ef0be82b03c9f2539273919eaa0c075ab4be
+size 156259
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_026.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_026.png
new file mode 100644
index 0000000000000000000000000000000000000000..d1ced4581b44fa6277c0e35732e21b835ddeb7ea
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_026.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42c86e6c53ef2f0d9c164115d214f7a0a1aea9e52c86b2d56ae8e53810d1176c
+size 140452
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_031.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_031.png
new file mode 100644
index 0000000000000000000000000000000000000000..a5afdde3f10c0473964ecc3e308a86ec337df8f1
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_031.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6eb0fed680ca966a6ae06e455a54f68ea026e9e221d2a3593d88623bc7949376
+size 154664
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_035.png b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_035.png
new file mode 100644
index 0000000000000000000000000000000000000000..dfeec5897f8db50904468503325b7cbc34c25fc7
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/orbit_frames/frame_035.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52b7c4b214ada8f45e5bc983d3a0f6250e99272787903a0521114c5ef7f35719
+size 139847
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/prior_metrics.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/prior_metrics.json
new file mode 100644
index 0000000000000000000000000000000000000000..c111067afc948828dd12a2f238f179fa51ee1414
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/prior_metrics.json
@@ -0,0 +1,491 @@
+{
+  "session_id": "session_20260921_233144_367389",
+  "prior_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1",
+  "owner_prior_approval": "pending",
+  "visual_quality_assessment": "not performed by this packager",
+  "object": {
+    "name": "bike_chain",
+    "representation": "tube",
+    "fitter_source": "scripts/chain_state_estimation.py loss/solver literals",
+    "dimension_source": "owner approximate 9.5 mm diameter (corpus Q2, 2026-10-25)",
+    "radius_m": 0.00475,
+    "ribbon_width_m": null,
+    "step_t": 0.125,
+    "step_angle_deg": 30,
+    "u_resolution": null,
+    "overlap_rule": "surface_intersection",
+    "twist_penalty": null
+  },
+  "initial_grid_sample": 0,
+  "views": [
+    {
+      "serial": "12540311",
+      "view_index": 0,
+      "selected_window_local_index": 1,
+      "converted_frame": 269,
+      "native_container_frame": 270,
+      "official_frame": 269,
+      "depth_frame": 269,
+      "capture_timestamp_ns": 1790047938734565100,
+      "device_timestamp_ns": 1790047938691799126,
+      "initial_grid_offset_ms": -15.595843,
+      "native_count": 406,
+      "initial_observed_depth_points": 8480,
+      "median_3d_residual_m": 0.00687211100012064,
+      "p90_3d_residual_m": 0.010084983985871076,
+      "visible_centerline_inside_mask_fraction": 0.9901315789473685,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          160,
+          202,
+          253,
+          293,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1602,
+          "hand_A": 3398,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6356,
+          0.7556093596059114,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          0,
+          1177
+        ],
+        "paired_common_min_max": [
+          3195,
+          4564
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_0_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_0_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_0_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_0_12540311.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/12540311_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12540311/paired_depth.png"
+      }
+    },
+    {
+      "serial": "12933256",
+      "view_index": 1,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938718969257,
+      "device_timestamp_ns": 1790047938670420020,
+      "initial_grid_offset_ms": 0.0,
+      "native_count": 406,
+      "initial_observed_depth_points": 6557,
+      "median_3d_residual_m": 0.0058602094650268555,
+      "p90_3d_residual_m": 0.010533764958381653,
+      "visible_centerline_inside_mask_fraction": 0.9665738161559888,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          151,
+          202,
+          214,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1867,
+          "hand_A": 3133,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.6628,
+          0.9445802955665025,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          1159,
+          1267
+        ],
+        "paired_common_min_max": [
+          1777,
+          4374
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_1_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_1_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_1_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_1_12933256.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/12933256_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/12933256/paired_depth.png"
+      }
+    },
+    {
+      "serial": "17736741",
+      "view_index": 2,
+      "selected_window_local_index": 0,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938706187642,
+      "device_timestamp_ns": 1790047938663525231,
+      "initial_grid_offset_ms": 12.781615,
+      "native_count": 406,
+      "initial_observed_depth_points": 6783,
+      "median_3d_residual_m": 0.008450270630419254,
+      "p90_3d_residual_m": 0.011142997816205025,
+      "visible_centerline_inside_mask_fraction": 0.9841269841269842,
+      "track_diagnostics": {
+        "T": 406,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          119,
+          121,
+          151,
+          202,
+          253,
+          303,
+          354,
+          405
+        ],
+        "source_seed_counts": {
+          "object": 1348,
+          "hand_A": 3652,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8368,
+          0.953579802955665,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          386,
+          897
+        ],
+        "paired_common_min_max": [
+          3762,
+          4527
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_2_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_2_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_2_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_2_17736741.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/17736741_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/17736741/paired_depth.png"
+      }
+    },
+    {
+      "serial": "23007103",
+      "view_index": 3,
+      "selected_window_local_index": 1,
+      "converted_frame": 268,
+      "native_container_frame": 269,
+      "official_frame": 268,
+      "depth_frame": 268,
+      "capture_timestamp_ns": 1790047938720158122,
+      "device_timestamp_ns": 1790047938674181398,
+      "initial_grid_offset_ms": -1.188865,
+      "native_count": 407,
+      "initial_observed_depth_points": 4519,
+      "median_3d_residual_m": 0.009255929850041866,
+      "p90_3d_residual_m": 0.012146585062146187,
+      "visible_centerline_inside_mask_fraction": 0.9860335195530726,
+      "track_diagnostics": {
+        "T": 407,
+        "queries": 5000,
+        "keyframes": [
+          0,
+          50,
+          101,
+          152,
+          203,
+          217,
+          253,
+          304,
+          355,
+          406
+        ],
+        "source_seed_counts": {
+          "object": 1992,
+          "hand_A": 3008,
+          "hand_B": 0
+        },
+        "tap_visible_fraction_min_mean_max": [
+          0.8322,
+          0.9410083538083538,
+          1.0
+        ],
+        "accepted_object_min_max": [
+          911,
+          1267
+        ],
+        "paired_common_min_max": [
+          2618,
+          4179
+        ]
+      },
+      "links": {
+        "centerline_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/centerline_3_with_masks.png",
+        "geometry_overlay": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_3_unmasked.png",
+        "geometry_with_mask": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/silhouette_3_with_masks.png",
+        "initial_rgb_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/view_3_23007103.png",
+        "native_rgb_depth_video": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/rgb_waft_depth_full_window.mp4",
+        "visibility_certainty": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/23007103_visibility_certainty.png",
+        "support_rejections": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/support_and_rejections.png",
+        "paired_depth": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/23007103/paired_depth.png"
+      }
+    }
+  ],
+  "recipe": {
+    "depth_source": "waft_stereo_dav2l5",
+    "mask_erosion_px": 2,
+    "filter": {
+      "method": "Open3D.remove_statistical_outlier",
+      "nb_neighbors": 20,
+      "std_ratio": 2.0,
+      "input_count": 27239,
+      "kept_count": 26339,
+      "removed_count": 900,
+      "per_view": [
+        {
+          "view": 0,
+          "input": 8645,
+          "kept": 8480,
+          "removed": 165,
+          "live_20_0.05_kept": 6121,
+          "no_filter_kept": 8645
+        },
+        {
+          "view": 1,
+          "input": 6875,
+          "kept": 6557,
+          "removed": 318,
+          "live_20_0.05_kept": 3896,
+          "no_filter_kept": 6875
+        },
+        {
+          "view": 2,
+          "input": 6954,
+          "kept": 6783,
+          "removed": 171,
+          "live_20_0.05_kept": 4963,
+          "no_filter_kept": 6954
+        },
+        {
+          "view": 3,
+          "input": 4765,
+          "kept": 4519,
+          "removed": 246,
+          "live_20_0.05_kept": 1189,
+          "no_filter_kept": 4765
+        }
+      ]
+    },
+    "max_rounds": 400,
+    "reprojection": {
+      "weight": 4.0,
+      "direction": "visible mask pixels to projected centerline",
+      "ribbon_twist_stage": null,
+      "mask": "accepted object outside hand/occluder union",
+      "view_aggregation": "equal view means",
+      "distance": "metric image plane using per-view median visible belt depth and fx/fy",
+      "samples": 360,
+      "huber_delta_m": 0.005,
+      "smooth_cap_m": 0.035,
+      "baseline_run": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_v1"
+    },
+    "solver": {
+      "solver_type": "lbfgs",
+      "linesearch_type": "zoom",
+      "max_linesearch_steps": 10,
+      "grad_norm_tolerance": 0.0001,
+      "max_iter": 25
+    },
+    "ribbon_twist_refinement": null
+  },
+  "fit": {
+    "stage1_rounds": 1,
+    "stage1_stop": "gradient_tolerance",
+    "stage1_final": null,
+    "twist": null,
+    "final": {
+      "loss": 5.9778743889182806e-05,
+      "components": {
+        "distance": 2.406455132586416e-05,
+        "curvature": 1.2257775949819916e-07,
+        "arclength": 7.299825597328891e-07,
+        "mask_reprojection_2d": 3.486163404886611e-05
+      },
+      "grad_norm": 9.39435267355293e-05,
+      "arclength_m": 0.5877017974853516,
+      "arclength_bound_active_within_1mm": true,
+      "max_curvature_per_m": 37.92765808105469,
+      "curvature_p50_p90_per_m": [
+        9.860498428344727,
+        22.216008949279782
+      ],
+      "curvature_above_max_fraction": 0.492,
+      "mean_distance_m": 0.007389499891093807,
+      "max_distance_m": 0.017994370311498642,
+      "beyond_2delta_fraction": 0.19343938646114128,
+      "at_cap_fraction": 0.0,
+      "per_view": [
+        {
+          "view": 0,
+          "count": 8480,
+          "median_m": 0.00687211100012064,
+          "p90_m": 0.010084983985871076,
+          "max_m": 0.013782148249447346,
+          "beyond_2delta_fraction": 0.10613207547169812,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 1,
+          "count": 6557,
+          "median_m": 0.0058602094650268555,
+          "p90_m": 0.010533764958381653,
+          "max_m": 0.017994370311498642,
+          "beyond_2delta_fraction": 0.11087387524782676,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 2,
+          "count": 6783,
+          "median_m": 0.008450270630419254,
+          "p90_m": 0.011142997816205025,
+          "max_m": 0.014050017111003399,
+          "beyond_2delta_fraction": 0.2523956951201533,
+          "at_cap_fraction": 0.0
+        },
+        {
+          "view": 3,
+          "count": 4519,
+          "median_m": 0.009255929850041866,
+          "p90_m": 0.012146585062146187,
+          "max_m": 0.015855133533477783,
+          "beyond_2delta_fraction": 0.38858154458951094,
+          "at_cap_fraction": 0.0
+        }
+      ],
+      "round": 1,
+      "relative_loss_change": 0.981192264140135
+    },
+    "arclength_soft_bounds_m": [
+      0.5830000042915344,
+      0.5849999785423279
+    ],
+    "arclength_outside_soft_bounds_m": 0.0027018189430236816
+  },
+  "geometry": {
+    "vertex_count": 1728,
+    "triangle_count": 3456,
+    "checks": {
+      "euler_characteristic": 0,
+      "watertight": true,
+      "components": 1,
+      "winding_consistent": true,
+      "signed_volume_m3": 3.976149783233774e-05,
+      "minimum_outward_normal_dot_m": 0.004579434340664218,
+      "degenerate_triangles": 0,
+      "self_intersecting": false,
+      "maximum_radius_error_m": 7.564100575063737e-09,
+      "maximum_diameter_error_m": 1.5116993679412105e-08,
+      "bishop_seam_error": 7.450580596923828e-09,
+      "minimum_nonadjacent_centerline_distance_m": 0.006705765687998025,
+      "triangle_area_m2": {
+        "min": 3.3956602598738355e-06,
+        "median": 4.991684670915842e-06,
+        "mean": 5.015910096568565e-06,
+        "max": 6.254291218252229e-06,
+        "density_max_over_min": 1.841848341590158
+      },
+      "bounds_m": [
+        [
+          -0.0369919128715992,
+          -0.17409859597682953,
+          2.6456546038389206e-05
+        ],
+        [
+          0.1717047244310379,
+          -0.0010159998200833797,
+          0.017292136326432228
+        ]
+      ],
+      "radius_m": 0.00475,
+      "step_t": 0.125,
+      "step_angle_deg": 30,
+      "overlap_rule": "surface_intersection",
+      "open3d_reported_intersecting_pairs": 0,
+      "confirmed_intersecting_pairs": [],
+      "passed": true,
+      "orientation_flipped": true
+    }
+  },
+  "appearance_seed_counts": {
+    "observed": 26339,
+    "mesh": 103680,
+    "total": 130019,
+    "mesh_to_observed": 3.936368123315236
+  },
+  "vertex_colors": {
+    "observed_vertices": 1147,
+    "filled_vertices": 581
+  },
+  "mechanical_population": {
+    "particle_count": null,
+    "total_unit_mass": null,
+    "status": "not produced by the prior run; no post-prior mechanics are evaluated here"
+  },
+  "source_code": {
+    "commit": "d1d2f1381470138dfbb3c58d448e53bfed1e0d7e",
+    "u_pin": "81c718790a37e5e0102eb77af2c6edd34a9db25f"
+  },
+  "media": {
+    "full_rotation": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/orbit.mp4",
+    "geometry_depth_cloud": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/tube/tube_scale_views.png",
+    "depth_offset_diagnostic": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/inputs/depth_offset.png",
+    "residual_histograms": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/spline/residual_histograms.png",
+    "color_visibility_fill": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/colors/visibility_and_fill.png",
+    "hybrid_seed_populations": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/review/hybrid/combined.png",
+    "derived_coverage": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_coverage.png",
+    "derived_30hz_movie": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/derived/derived_30hz_xyz.mp4"
+  },
+  "sources": {
+    "manifest": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/manifest.json",
+    "fit": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/steps/fit.json",
+    "geometry_checks": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/tube/tube_checks.json",
+    "reference_views": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/inputs/reference_views.json",
+    "reprojection": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/prior_chain_m18_lowcurvature_v1/spline/reprojection.json",
+    "tracking_summary": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/summary.json",
+    "frames": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/frames_waft/per_camera_inputs.json"
+  }
+}
diff --git a/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/retained_reference_limits.json b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/retained_reference_limits.json
new file mode 100644
index 0000000000000000000000000000000000000000..fe534d556a6039c0e74cc36feed60e2da79ab696
--- /dev/null
+++ b/sessions/session_20260921_233144_367389/preprocessing/chain-stronger-reprojection-v2/native/prior_review_m18_lowcurvature_v1/retained_reference_limits.json
@@ -0,0 +1,31 @@
+{
+  "source": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/REVIEW.md",
+  "source_findings": "/mnt/data2/magna/belt_perception_shared/phystwin_model_corpus_expansion/lab/session_20260921_233144_367389/window_anchor0268-0673/accepted_input_preparation_c6a_v1/review_v1/agent_findings.json",
+  "additional_diagnostics": {
+    "12540311_zero_object_native_relative_spans_inclusive": [
+      [
+        143,
+        238
+      ],
+      [
+        240,
+        240
+      ],
+      [
+        243,
+        243
+      ],
+      [
+        262,
+        326
+      ]
+    ],
+    "12540311_coverage_nan_count": 116,
+    "12540311_finite_coverage_min_median_max": [
+      0.0,
+      0.47863454,
+      0.8996283
+    ]
+  },
+  "interpretation": "Later camera12540311 support gaps,116 undefined coverage values and isolated derived points remain; initial-prior review does not clear later sequence for physical fitting."
+}