ZGen 1.2.5: +78 ~5 -20

.gitattributes

@@ -153,3 +153,11 @@ ZGenData/models/trellis2/shape_enc_1.onnx.data filter=lfs diff=lfs merge=lfs -te
 ZGenData/models/trellis2/shape_enc_2.onnx.data filter=lfs diff=lfs merge=lfs -text
 ZGenData/models/trellis2/shape_enc_3.onnx.data filter=lfs diff=lfs merge=lfs -text
 ZGenData/models/trellis2/shape_enc_4.onnx.data filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/hunyuan21_20261016_224600/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_053901/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_114858/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_130210/model_mesh.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_135421/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_142722/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_144855/model.glb filter=lfs diff=lfs merge=lfs -text
+ZGenData/samples/pixal3d_20261018_151837/model.glb filter=lfs diff=lfs merge=lfs -text

ZGenData/Hunyuan21.exe

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:83bad810cd4dafdeef71cd9debe83cfb89f04edf86a866c74f8f8635736d96aa
-size 1695744
+oid sha256:de6a4337ae1cd257e25aedf33c760b9f8f51afac8e3bb8d7a3fa65e4903928ca
+size 1649664

ZGenData/Pixal3D.exe

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b1a3f61b34c98c289297fa7bf1fd2c88847f1f40b54abd21921561aa3ea97b89
+oid sha256:622498acb411d7587f0b50d8132a4595c4f368cc6100fedecfdfcbf7d1739f7d
 size 2018816

ZGenData/Trellis2.exe

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c981c3df52d355c0d640a20634c3968d19155770e5d02655f8f05250bec4cd6
-size 1968128
+oid sha256:7234ab7034adbead4655aa4fc6dc3a546a88e8cce28ee5dfef839ca5aa8aaf26
+size 1921024

ZGenData/ZGen.exe

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f952a0a344554710840ddcaf9d9ee5f18b125380967ea69a7c823549028f8521
-size 1505280
+oid sha256:f962884c666dc02f0c9de7088b556bfb52b8f0127451094c915c88927c8b3e64
+size 1531904

ZGenData/ZGen_cfg.txt

@@ -1 +1 @@
-ZGenVersion=1.2.2
+ZGenVersion=1.2.5

ZGenData/manifest.json

@@ -1,6 +1,6 @@
 {
- "version": "1.2.2",
- "generated": "2026-10-16 23:57:56",
+ "version": "1.2.5",
+ "generated": "2026-10-18 19:53:34",
  "files": [
   {
    "path": "HDRI.exr",
@@ -9,13 +9,13 @@
   },
   {
    "path": "Hunyuan21.exe",
-   "size": 1695744,
-   "sha256": "83bad810cd4dafdeef71cd9debe83cfb89f04edf86a866c74f8f8635736d96aa"
+   "size": 1649664,
+   "sha256": "de6a4337ae1cd257e25aedf33c760b9f8f51afac8e3bb8d7a3fa65e4903928ca"
   },
   {
    "path": "Pixal3D.exe",
    "size": 2018816,
-   "sha256": "b1a3f61b34c98c289297fa7bf1fd2c88847f1f40b54abd21921561aa3ea97b89"
+   "sha256": "622498acb411d7587f0b50d8132a4595c4f368cc6100fedecfdfcbf7d1739f7d"
   },
   {
    "path": "ThirdPartyLicenses/BEN2_LICENSE.txt",
@@ -154,8 +154,8 @@
   },
   {
    "path": "Trellis2.exe",
-   "size": 1968128,
-   "sha256": "1c981c3df52d355c0d640a20634c3968d19155770e5d02655f8f05250bec4cd6"
+   "size": 1921024,
+   "sha256": "7234ab7034adbead4655aa4fc6dc3a546a88e8cce28ee5dfef839ca5aa8aaf26"
   },
   {
    "path": "ZBrush/ZGenImport.txt",
@@ -164,8 +164,8 @@
   },
   {
    "path": "ZGen.exe",
-   "size": 1505280,
-   "sha256": "f952a0a344554710840ddcaf9d9ee5f18b125380967ea69a7c823549028f8521"
+   "size": 1531904,
+   "sha256": "f962884c666dc02f0c9de7088b556bfb52b8f0127451094c915c88927c8b3e64"
   },
   {
    "path": "ZGenLicense.txt",
@@ -174,8 +174,8 @@
   },
   {
    "path": "ZGen_cfg.txt",
-   "size": 19,
-   "sha256": "ed18aa7294219ac53ca462bca5fd1ae80d7d16d5695254f21c49f3eb475ff1f9"
+   "size": 17,
+   "sha256": "4ef5c98ed506bc0b24b8bf778cad43e6d2e31c1e9d4db32139beeeb7f09658c6"
   },
   {
    "path": "assets/matcap/matcap_bronze_skin.png",
@@ -547,6 +547,11 @@
    "size": 36191232,
    "sha256": "63f202a7a91810cec1659cb30a0eec88181fc2d0d83bd724b7fba348fe6a2451"
   },
+  {
+   "path": "models/pixal3d/rounded_gelu.dll",
+   "size": 38912,
+   "sha256": "450fbcfa77c5bdf02c03d4f4f25043595007178877001d4443301af6a8811f0c"
+  },
   {
    "path": "models/pixal3d/sh1024_flow.onnx",
    "size": 1918535,
@@ -557,6 +562,11 @@
    "size": 2807937024,
    "sha256": "73b26ff420cfa4540d2937050b55a836102cfce7edc733273b515078286e1999"
   },
+  {
+   "path": "models/pixal3d/sh1024_flow_rg.onnx",
+   "size": 1427259,
+   "sha256": "822d560a926cab1409dce9d8679479a81def9c64905f62f48f9fc3de6ee0b441"
+  },
   {
    "path": "models/pixal3d/sh512_flow.onnx",
    "size": 1917836,
@@ -567,6 +577,11 @@
    "size": 2807937024,
    "sha256": "431665a340997dfad752f12c7497eb78eb4bf368b18460d244e48a9c2207afd0"
   },
+  {
+   "path": "models/pixal3d/sh512_flow_rg.onnx",
+   "size": 1426560,
+   "sha256": "08fcd5aa4cd51883bacde1ca5659433c36ddc9fc7f87dfb295bf621eb03aafdf"
+  },
   {
    "path": "models/pixal3d/shape_dec_0.onnx",
    "size": 67105,
@@ -637,6 +652,11 @@
    "size": 2713178112,
    "sha256": "849e14ae6400ca49f284204b0b0230b62e648f27ee1b09b472a5430d9dfea3ed"
   },
+  {
+   "path": "models/pixal3d/ss_flow_rg.onnx",
+   "size": 253571208,
+   "sha256": "39bfb61a206957fe8b8c52738bd7ab56b70f67212f7833c23752754c7108c89c"
+  },
   {
    "path": "models/pixal3d/tex_dec_0.onnx",
    "size": 66350,
@@ -697,6 +717,11 @@
    "size": 2808056832,
    "sha256": "fd96d2781c2d5de07607cd5ae1342151cf34463fcf82a9d22624432b10ca1e17"
   },
+  {
+   "path": "models/pixal3d/tex_flow_rg.onnx",
+   "size": 1425375,
+   "sha256": "c7b2301566272400096af8521206d372dfb5762a0b40f12305bc9d9eeee35059"
+  },
   {
    "path": "models/trellis2/BEN2_Base.onnx",
    "size": 222932053,
@@ -988,54 +1013,39 @@
    "sha256": "02c79b76bd0349323baca425ae0c400f17e95bda964da46bd08f9d76c3a8afe9"
   },
   {
-   "path": "samples/pixal3d_20261013_133323/config.txt",
-   "size": 191,
-   "sha256": "d70f40c688cf612bbc5ed31fe9f624697e611700bb06d10e3525dfc35928f5f7"
+   "path": "samples/hunyuan21_20261016_224600/config.txt",
+   "size": 183,
+   "sha256": "4cd188cfce7935510ff368945502331742b566b8b45c28225c316fd1e98298a4"
   },
   {
-   "path": "samples/pixal3d_20261013_133323/icon.png",
-   "size": 35256,
-   "sha256": "3897fb61f75163996f350c61f20825b45a1cc9c5c43cc0eabf46192b659f71b7"
+   "path": "samples/hunyuan21_20261016_224600/icon.png",
+   "size": 32227,
+   "sha256": "7dda5d5d0fc24061e21ea278e70240ed0f4407b45217ba75477a8d3636043542"
   },
   {
-   "path": "samples/pixal3d_20261013_133323/info.txt",
-   "size": 103,
-   "sha256": "8e64ea953a95b6977e2a256422728567a668dd555fc669acf4b99303aad27172"
+   "path": "samples/hunyuan21_20261016_224600/info.txt",
+   "size": 132,
+   "sha256": "24c67372c8f9c50fe81c5a0995757f83a400e18b34365e2e82114867a62243e3"
   },
   {
-   "path": "samples/pixal3d_20261013_133323/input.png",
-   "size": 2347981,
-   "sha256": "c1dc8f0d697823046ae590d94cb1d1fe607d6f0129afd7e50f2994995d7f4e54"
-  },
-  {
-   "path": "samples/pixal3d_20261013_133323/model.glb",
-   "size": 83745560,
-   "sha256": "cbcfa05e0b9bc8d8561c306f3687be3855cd413d0eb169a45a3ea9807453b14d"
-  },
-  {
-   "path": "samples/pixal3d_20261013_155732/config.txt",
-   "size": 240,
-   "sha256": "3ccbc3630f2610b484a36497b06e1073a923f96e252b1cfb9b15d9ff46dd17c6"
-  },
-  {
-   "path": "samples/pixal3d_20261013_155732/icon.png",
-   "size": 31540,
-   "sha256": "65faaf94f618764ab910e4004f5127fc7e240a8c3f15f0b62dae312e0f2f79b8"
+   "path": "samples/hunyuan21_20261016_224600/input.png",
+   "size": 2198681,
+   "sha256": "8f8e272e189237648801f6200e4d34bd9c9f2a4ae90f32ea9f58b85270938c41"
   },
   {
-   "path": "samples/pixal3d_20261013_155732/info.txt",
-   "size": 105,
-   "sha256": "2b9bfcab0e0050befd9e466c3def129499688d1eed479c8fea3ad21650a34d62"
+   "path": "samples/hunyuan21_20261016_224600/log.txt",
+   "size": 8346,
+   "sha256": "e3ea7fa220d9c505d9b63e43ef02f035a34f34783bb79486f0450b1511007f62"
   },
   {
-   "path": "samples/pixal3d_20261013_155732/input.png",
-   "size": 2536975,
-   "sha256": "9a8ce901d827aaa70dbc702d142c4ecb6f6deac7f24220d2201a64295541a143"
+   "path": "samples/hunyuan21_20261016_224600/model.glb",
+   "size": 60409524,
+   "sha256": "abc35579f70ee217222e53dfe9fa59c23feda2c6a9e5c7eb51ac5018e04d91df"
   },
   {
-   "path": "samples/pixal3d_20261013_155732/model.glb",
-   "size": 141780184,
-   "sha256": "ead347c0892f18082b5411d516338e0c81a81a8377accee1ce57f9cc8f1ded1e"
+   "path": "samples/hunyuan21_20261016_224600/run.txt",
+   "size": 679,
+   "sha256": "7226b3aa41e94fb6197832ca3ea35032af11fb2dbcb8bfd27cfab06f416979c1"
   },
   {
    "path": "samples/pixal3d_20261014_125637/config.txt",
@@ -1102,31 +1112,6 @@
    "size": 61387608,
    "sha256": "6c4a28526ec7a1db8f28175be71ef473a26db12e2e0d76201e887c353ad47e7e"
   },
-  {
-   "path": "samples/pixal3d_20261014_155321/config.txt",
-   "size": 240,
-   "sha256": "57ccb4260ca8fcc53799f8d0c6489d14c8c8a1a1a4c1b632989507384349249c"
-  },
-  {
-   "path": "samples/pixal3d_20261014_155321/icon.png",
-   "size": 35842,
-   "sha256": "0fbb3f7fcf68c9916a13d1172aba0086e1ab6012db78e6c011771253439b415f"
-  },
-  {
-   "path": "samples/pixal3d_20261014_155321/info.txt",
-   "size": 102,
-   "sha256": "7562c0c0f8647f87b9adbb58d668ab062a01db2843a0c9c7bdfc7e103ace50a3"
-  },
-  {
-   "path": "samples/pixal3d_20261014_155321/input.png",
-   "size": 2868318,
-   "sha256": "dd8598c665e609749014b7cfccacee9c7333ea29210e50612da8272fe53e66c5"
-  },
-  {
-   "path": "samples/pixal3d_20261014_155321/model.glb",
-   "size": 75552264,
-   "sha256": "c8596aba5f66478009f0d3cd55c53d87039637f6adc96e67314aed886f6fc8be"
-  },
   {
    "path": "samples/pixal3d_20261014_215500/config.txt",
    "size": 242,
@@ -1153,29 +1138,334 @@
    "sha256": "4569016d125628e228b91f80ba0f49399e133d30804f61b5a79c82c33ce56281"
   },
   {
-   "path": "samples/trellis2_20261013_100355/config.txt",
-   "size": 94,
-   "sha256": "b8f9d4a591056d5b9572f235ce5fbfe0129bdf88ba59b3533192cab87d23e0f4"
+   "path": "samples/pixal3d_20261018_053901/config.txt",
+   "size": 328,
+   "sha256": "475a51417cc8ab22ae4104311e85d8c77bcd762ea858b8be6a6f5903738937a2"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/icon.png",
+   "size": 38413,
+   "sha256": "bea500d8014e3951509eaa01f1c58e455b0c081c9eed17084f73fc660ac288ce"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/info.txt",
+   "size": 102,
+   "sha256": "f26069b8fc509a72dca8493d14953173f7cfea392e75803f2c3cd88384fcf833"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/input.png",
+   "size": 2868318,
+   "sha256": "dd8598c665e609749014b7cfccacee9c7333ea29210e50612da8272fe53e66c5"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/log.txt",
+   "size": 9698,
+   "sha256": "1c9184412ffe47b5e00e4f7dfada1acf0f8b071306938606eaad222916053b21"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/model.glb",
+   "size": 81416896,
+   "sha256": "a2e30802ffe94bc0a1b03a57b46dab42ba57af015175881163b1331eff032eba"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/run.txt",
+   "size": 671,
+   "sha256": "7ed1b1b050cc5c9c9151b6ea88bcd3bc1a3eeeb7c8c5ed22c4c02bd3d179fdc4"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/state/latent.bin",
+   "size": 5644960,
+   "sha256": "94ad176a4f4289eb40dd6e2a18a6a76cfc7fe6edfd102995cc65f3afcc4ca571"
+  },
+  {
+   "path": "samples/pixal3d_20261018_053901/state/mesh.bin",
+   "size": 31501960,
+   "sha256": "18fcb0476e36a7f20a22df935bd1dc400094f3b132cc65efb59a1b7b840e24fd"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/config.txt",
+   "size": 410,
+   "sha256": "308feec19059d6ad30a1dda8f9949df53d536deb4912c65a2879113c92f9cad5"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/icon.png",
+   "size": 45617,
+   "sha256": "4428ce4ed68ebed48a4b3be6237ffc9cd26de00eb0d762d8f6fbb78d33696101"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/info.txt",
+   "size": 105,
+   "sha256": "be6e9ec33984b04cbac54a711e14f2df31e9f4312af4d8cee0b57d5eb13e36d5"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/input.png",
+   "size": 2709338,
+   "sha256": "7d44ea1b48d1b64f6c0f90a6da391b700c863644609ca5989ac6bf1a2f183602"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/log.txt",
+   "size": 8974,
+   "sha256": "4ae45156dd422f0328cc75421609d65e071664fad950d6b7fb238d7ce5630748"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/model.glb",
+   "size": 81774628,
+   "sha256": "f2463e2371054a51969b80fd33489beefb019d5b0f3bdcc27ba07a61ed512c28"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/run.txt",
+   "size": 674,
+   "sha256": "ee0567ddab630caca872314ef2c779cc086b70a5966c0d302b1a504d9be006c1"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/state/latent.bin",
+   "size": 5844040,
+   "sha256": "2ed2cc06a489c47af690ce3b64dfef503cc46beb4d1a59ad9e5c9cdd46d85b5e"
+  },
+  {
+   "path": "samples/pixal3d_20261018_114858/state/mesh.bin",
+   "size": 36763804,
+   "sha256": "b830006d45a6447d8ac2aa0dbbf464d0473d5e23ed6465aafeef8fd56997e0c3"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/config.txt",
+   "size": 410,
+   "sha256": "8f7d9733077c6c805dba3c836a5391cffba698575898d92af35b2c0c17b1a058"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/config_texture.txt",
+   "size": 423,
+   "sha256": "107a8dc1176a67cf4cc60424b146a0292a26e0ecd0330e48608c6e1857a80892"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/icon.png",
+   "size": 37879,
+   "sha256": "a3ec23315ddac50d423e36f1c4f2a022aa760f435562aac1be8e3ab31234d32e"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/info.txt",
+   "size": 103,
+   "sha256": "1feb56036b796c9724a2452547ec576d8268ecf91f210afb6ac582e0e294b76a"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/input.png",
+   "size": 2539817,
+   "sha256": "fd320aaa1bffe069015cc3c384a1496b7593c77648edb20c047dfdd09bf9d8ce"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/log.txt",
+   "size": 8816,
+   "sha256": "d28c1a297d55ef10c19c36899cfe1ad8fd301cf970c30af22d10e8579611878d"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/log_texture.txt",
+   "size": 1108,
+   "sha256": "b45867f5490437af8f3c5081e4073d436af58bb3c193f2a53e66b52aedd825da"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/model_mesh.glb",
+   "size": 89199544,
+   "sha256": "33ed4ab159d6733fe3378340f1a52017a891fbda39f574cfd3148b61a57e0fbb"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/run.txt",
+   "size": 692,
+   "sha256": "ab810ee3a56843fcec3a453e0f90f0db2d887dd0956aa05b353eb8012a274db5"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/run_texture.txt",
+   "size": 633,
+   "sha256": "fc72169510928eab2b13bbb63757c41748c860264a390e3f5edb57f8d1a1a08b"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/state/latent.bin",
+   "size": 3926040,
+   "sha256": "f5824028ce5b6b5083da1a603e73117ee2818932b325d05cb3abf69daf046c4c"
+  },
+  {
+   "path": "samples/pixal3d_20261018_130210/state/mesh.bin",
+   "size": 36831196,
+   "sha256": "3ce3772e2ef2ccdfae0615739c0584323f6c983e15c783478eee35ce7fafc187"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/config.txt",
+   "size": 410,
+   "sha256": "8424d40717351ffc66de2b177eec481bea71cc017b36ea2d76044a17c9d0e481"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/icon.png",
+   "size": 42713,
+   "sha256": "ac3b6391383e085c6dbdb2b53f314da6a4dd1aa421f93bb6d059ea2a51d61cae"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/info.txt",
+   "size": 104,
+   "sha256": "3ef699a3d8a7fbb523b01327558fb3b6fc6dff4d60e935c6bf071f188e439a6a"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/input.png",
+   "size": 2810080,
+   "sha256": "e126bbbde28a3e404a1add2be44bf254c3de2cef1198db234ee713b5369995c6"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/log.txt",
+   "size": 8966,
+   "sha256": "490a0c66b87c77c0589a735cff46d7514596b0dc3aca935c3d4f6c57f3517224"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/model.glb",
+   "size": 83021832,
+   "sha256": "2cf42721d44d3d12c1f421078ed078b16720991eff1c7619bd17c2cd32fac26e"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/run.txt",
+   "size": 673,
+   "sha256": "d844754ce84e6c750e708675b2c3f1a7acd824854b7d605632e71840ed08ae06"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/state/latent.bin",
+   "size": 6317940,
+   "sha256": "b6d5bfbdf99db4d3c7c434853b5ba44cfd9cd53e4f6c800713a5cb2218643ec4"
+  },
+  {
+   "path": "samples/pixal3d_20261018_135421/state/mesh.bin",
+   "size": 37075096,
+   "sha256": "447b80bd64daf8885c059a0500d1be4d8e9ee9fafa32c0a3eab05db2a1b97942"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/config.txt",
+   "size": 410,
+   "sha256": "ec48f18fe3c06a164f5a7c0cf62ac5ce418cead9bc299f7dd10af3852c2c09c0"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/icon.png",
+   "size": 36903,
+   "sha256": "e6eb8211222df2da92c6e2dd86cd932edc2769e803822475d78b676c725e8382"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/info.txt",
+   "size": 102,
+   "sha256": "df9c88af3bcb8afd9233d6dc94ae92ecefd9a8ee0bac8db7a80d3da7ef88467f"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/input.png",
+   "size": 2532006,
+   "sha256": "c3c44ae21d535a5c62caca434bc14159100aa1668261a6f24fd26f35724ad690"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/log.txt",
+   "size": 8701,
+   "sha256": "2b3bf5009d6fb9b5f64721d788cd6106d62575ac08100c672bba3f9bc3bd1b2c"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/model.glb",
+   "size": 87563476,
+   "sha256": "570af783c9488df9895b82cf318249578c0425d7dfd3dee39f1a455e23cec35c"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/run.txt",
+   "size": 691,
+   "sha256": "0c3eb8e96b044cc738bb530aa42473be882121884f34b03ea1e6b800c7e9a73e"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/state/latent.bin",
+   "size": 4389160,
+   "sha256": "7f3a84120edc9e052f786d69bfb3eed7f3cc3cbba4566c8d033f595e65c007cc"
+  },
+  {
+   "path": "samples/pixal3d_20261018_142722/state/mesh.bin",
+   "size": 34990408,
+   "sha256": "3d2bc7876e86aee17dec5d271184fc1ba72587ca2bcad29f9f71d77c57f82f35"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/config.txt",
+   "size": 410,
+   "sha256": "09605635258bb97eaa514f5b7d856ccfa2467976b14e346bd29ead04b12198e4"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/icon.png",
+   "size": 35856,
+   "sha256": "e7f18b4619924d5d4abc0e49837e028dab925ebcfb43bce6bdb70a635c75ac64"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/info.txt",
+   "size": 102,
+   "sha256": "8a4effca31b40397ebd2bf6ab52801665d19f888b0552d7f681d0389ebd86137"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/input.png",
+   "size": 2592504,
+   "sha256": "1238dbc906021c52a2add4b3830ad02c467ed2e9c41f8da6abfc256ef68ec73d"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/log.txt",
+   "size": 8686,
+   "sha256": "f95309fc72b6e9f2986d512891f2f6fb87c8db42e9c727bd181d80f3985f6a04"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/model.glb",
+   "size": 74318180,
+   "sha256": "8b056c71f7a6864b84c34d46664984b750329b87deb1fa0871cfd31bd4d788a5"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/run.txt",
+   "size": 666,
+   "sha256": "d5b8364d57be8c6db32c61664fdfcdee58174ed0411f41985811f87fb12e6151"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/state/latent.bin",
+   "size": 4071920,
+   "sha256": "85fa3a6d447733210d78843de5346779af5ad3b2d0a56a2d4ca9d61a098cda5a"
+  },
+  {
+   "path": "samples/pixal3d_20261018_144855/state/mesh.bin",
+   "size": 33283432,
+   "sha256": "89696c46b9a30df9435cd7de6fb77eca5180c359b35a642371fd6997c5db9511"
+  },
+  {
+   "path": "samples/pixal3d_20261018_151837/config.txt",
+   "size": 410,
+   "sha256": "e6f2acbb1ffd39dd2de6464761f096dd4833c2fa84b97bac81b17f6130fa678c"
+  },
+  {
+   "path": "samples/pixal3d_20261018_151837/icon.png",
+   "size": 39598,
+   "sha256": "e0f218bf2e95a98a41c842d839818b8a51ad62b182b267d472b6973f4cdecb57"
+  },
+  {
+   "path": "samples/pixal3d_20261018_151837/info.txt",
+   "size": 103,
+   "sha256": "0d5edc354ab0a5278895baf403b91ff5ce054a6586e3538397d95903114f3964"
+  },
+  {
+   "path": "samples/pixal3d_20261018_151837/input.png",
+   "size": 2616719,
+   "sha256": "849cf8b802581d572b68a2ab75b9ace4a87d7203a321d767216798ba82797047"
+  },
+  {
+   "path": "samples/pixal3d_20261018_151837/log.txt",
+   "size": 8991,
+   "sha256": "1f2c6726af2e87a7d23156d22628ba93f4ae60103767f5ddd8ffea3a57eecfa4"
   },
   {
-   "path": "samples/trellis2_20261013_100355/icon.png",
-   "size": 26431,
-   "sha256": "0b0486880feeaefa295e4e18791d289bbbdf4b2f947cd0f1ed3bfd7a837cf60c"
+   "path": "samples/pixal3d_20261018_151837/model.glb",
+   "size": 94417928,
+   "sha256": "2b905b9e157cc790e021c6a6a098295002d9a6535fa00e670bfa9a8a409879ec"
   },
   {
-   "path": "samples/trellis2_20261013_100355/info.txt",
-   "size": 84,
-   "sha256": "f1203d24cefcfc66623a1c98c250e4c32f08a57df099e3443dd92b8243ca66e8"
+   "path": "samples/pixal3d_20261018_151837/run.txt",
+   "size": 672,
+   "sha256": "995e63f9685018270a658d131df4e13b43f1cdc09cf7653243320a8b1e6cda4f"
   },
   {
-   "path": "samples/trellis2_20261013_100355/input.png",
-   "size": 4528424,
-   "sha256": "da388036a0eccb2fe45ffded0bb5ecf33cbb22f4a3490acd5bfc73db703a6bde"
+   "path": "samples/pixal3d_20261018_151837/state/latent.bin",
+   "size": 3829580,
+   "sha256": "6b253034ecb928ae1b48c042f15ea625c4ccef45a6eeac3246906e2758b7b3bc"
   },
   {
-   "path": "samples/trellis2_20261013_100355/model.glb",
-   "size": 27314332,
-   "sha256": "6edf8efd2c9df99f5683442b9a687ff320afd3cbd2b04dfba178902b906f257b"
+   "path": "samples/pixal3d_20261018_151837/state/mesh.bin",
+   "size": 37746676,
+   "sha256": "da0c6e26b0464b13c029db27fb5c8db3c5412fd5dc7b660388aabef25df4fc06"
   },
   {
    "path": "samples/trellis2_20261013_123456/config.txt",

ZGenData/models/pixal3d/sh1024_flow_rg.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:822d560a926cab1409dce9d8679479a81def9c64905f62f48f9fc3de6ee0b441
+size 1427259

ZGenData/models/pixal3d/sh512_flow_rg.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08fcd5aa4cd51883bacde1ca5659433c36ddc9fc7f87dfb295bf621eb03aafdf
+size 1426560

ZGenData/models/pixal3d/ss_flow_rg.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39bfb61a206957fe8b8c52738bd7ab56b70f67212f7833c23752754c7108c89c
+size 253571208

ZGenData/models/pixal3d/tex_flow_rg.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7b2301566272400096af8521206d372dfb5762a0b40f12305bc9d9eeee35059
+size 1425375

ZGenData/samples/{pixal3d_20261013_133323 → hunyuan21_20261016_224600}/config.txt

@@ -1,14 +1,13 @@
-steps=12
-seed=42
-mesh_res=1024
-mix=0
+steps=24
+seed=1234
+mesh_res=256
 texture=1
 texture_size=4096
-decimate=600000
-decimate_1024=600000
+decimate=2500000
 tessimate=1
 tess_size_e5=12000
-camera_mode=auto
 solid_fill=0
 dynamic=0
-app_version=1.0.4
+texture_engine=trellis
+tex_mode=2
+app_version=1.2.2

ZGenData/samples/hunyuan21_20261016_224600/info.txt

@@ -0,0 +1,4 @@
+model=Hunyuan3D-2.1
+seconds=345
+status=D:\GitHub\ZGen\app\ZGenData\output\hunyuan21_20261016_224600\model.glb
+app_version=1.2.2

ZGenData/samples/hunyuan21_20261016_224600/log.txt

@@ -0,0 +1,73 @@
+license: Licensed: 27 days left
+Hunyuan3D-2.1 native C++ engine; ORT 1.23.2
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+[1/5] Background removal and DINO image features
+      background removed (BEN2)
+[2/5] DiT flow sampling (seed 1234)
+      CFG as two B=1 calls (8 GB layout)
+      DiT step 1/24
+      DiT step 4/24
+      DiT step 8/24
+      DiT step 12/24
+      DiT step 16/24
+      DiT step 20/24
+      DiT step 24/24
+[3/5] VAE feature decode
+[4/5] Dense geometry decode
+[5/5] Marching cubes and mesh processing
+[+] TRELLIS.2 texture on the Hunyuan mesh
+---- TRELLIS.2 texture (trellis_texture) ----
+[OK] Signed license is valid.
+license: Licensed: 27 days left
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+Texture mesh: D:\GitHub\ZGen\app\ZGenData\hunyuan_final.obj
+Image:  D:\GitHub\ZGen\input_Demo\Single\fwds.png
+Output: D:\GitHub\ZGen\app\ZGenData\trellis_texture\textured.glb
+seed=1234 tex_steps=12
+[1/5] mesh 628397 vertices, 1256790 faces
+      auto mask: background removed
+[2/5] dino 1024 8.6s
+      dual grid 3883278 voxels -> 970254 / 242665 / 60426 / 14971 tokens
+[3-4/5] dual grid + shape encoder 17.9s (14971 tokens)
+      tex_flow_1024 step 1/12 (14971 tokens)
+      tex_flow_1024 step 2/12 (14971 tokens)
+      tex_flow_1024 step 3/12 (14971 tokens)
+      tex_flow_1024 step 4/12 (14971 tokens)
+      tex_flow_1024 step 5/12 (14971 tokens)
+      tex_flow_1024 step 6/12 (14971 tokens)
+      tex_flow_1024 step 7/12 (14971 tokens)
+      tex_flow_1024 step 8/12 (14971 tokens)
+      tex_flow_1024 step 9/12 (14971 tokens)
+      tex_flow_1024 step 10/12 (14971 tokens)
+      tex_flow_1024 step 11/12 (14971 tokens)
+      tex_flow_1024 step 12/12 (14971 tokens)
+      colour match to the photo (Lab): L 42.8 -> 67.5, a 21.8 -> 17.5, b 11.4 -> 15.9 (spread x 1.13 1.40 1.61)
+      texture flow + decoder 110.6s
+      UV unwrap: 328 clusters / 655 charts (planar 31, ortho 285, lscm 203, piecewise 136), atlas 4328x4360, 674273 vertices
+[5/5] UV 4096x4096 + bake 14.6s
+SUCCESS OK vertices=674273 faces=1256790 seconds=151.6
+2026-10-16 22:49:30.8443912 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:30.8451632 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:49:32.1905436 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:32.1911042 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:49:34.2576489 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:34.2582187 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:49:36.5030464 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:36.5036199 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:49:39.5387409 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:39.5393194 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:49:42.1417272 [W:onnxruntime:, transformer_memcpy.cc:111 onnxruntime::MemcpyTransformer::ApplyImpl] 120 Memcpy nodes are added to the graph main_graph for CUDAExecutionProvider. It might have negative impact on performance (including unable to run CUDA graph). Set session_options.log_severity_level=1 to see the detail logs before this message.
+2026-10-16 22:49:42.2121626 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:49:42.2127377 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:51:08.9335067 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:51:08.9340763 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:51:14.4267291 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:51:14.4273535 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:51:22.1301799 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:51:22.1307475 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:51:25.3878210 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:51:25.3886075 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+2026-10-16 22:51:28.5791493 [W:onnxruntime:, session_state.cc:1316 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
+2026-10-16 22:51:28.5797058 [W:onnxruntime:, session_state.cc:1318 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.
+----
+SUCCESS D:\GitHub\ZGen\app\ZGenData\output\hunyuan21_20261016_224600\model.glb

ZGenData/samples/{pixal3d_20261013_133323 → hunyuan21_20261016_224600}/model.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cbcfa05e0b9bc8d8561c306f3687be3855cd413d0eb169a45a3ea9807453b14d
-size 83745560
+oid sha256:abc35579f70ee217222e53dfe9fa59c23feda2c6a9e5c7eb51ac5018e04d91df
+size 60409524

ZGenData/samples/hunyuan21_20261016_224600/run.txt

@@ -0,0 +1,13 @@
+ZGen run log
+app version: 1.2.2
+time: 2026-10-16 22:51:46
+model: Hunyuan3D-2.1
+outcome: finished
+status: D:\GitHub\ZGen\app\ZGenData\output\hunyuan21_20261016_224600\model.glb
+seconds: 345
+exit code 0x00000000
+engine exe: D:\GitHub\ZGen\app\ZGenData\Hunyuan21.exe
+models: D:\GitHub\ZGen\app\ZGenData\models\hunyuan21
+working dir: D:\GitHub\ZGen\app\ZGenData
+command: "D:\GitHub\ZGen\app\ZGenData\Hunyuan21.exe" "D:\GitHub\ZGen\input_Demo\Single\fwds.png" --out "D:\GitHub\ZGen\app\ZGenData\output\hunyuan21_20261016_224600\model.glb" --models "D:\GitHub\ZGen\app\ZGenData\models\hunyuan21" --work "D:\GitHub\ZGen\app\ZGenData" --signal --no-window
+engine exe size: 1713664 bytes

ZGenData/samples/pixal3d_20261013_133323/info.txt

@@ -1,4 +0,0 @@
-model=Pixal3D
-seconds=707
-status=OK vertices=1048365 faces=2096766 seconds=691.3
-app_version=1.0.4

ZGenData/samples/pixal3d_20261013_155732/info.txt

@@ -1,4 +0,0 @@
-model=Pixal3D
-seconds=2620
-status=OK vertices=1664479 faces=3329254 seconds=2606.5
-app_version=1.0.4

ZGenData/samples/pixal3d_20261013_155732/model.glb

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ead347c0892f18082b5411d516338e0c81a81a8377accee1ce57f9cc8f1ded1e
-size 141780184

ZGenData/samples/pixal3d_20261014_155321/info.txt

@@ -1,4 +0,0 @@
-model=Pixal3D
-seconds=804
-status=OK vertices=849982 faces=1700152 seconds=786.3
-app_version=1.0.6

ZGenData/samples/{pixal3d_20261013_155732 → pixal3d_20261018_053901}/config.txt

@@ -9,10 +9,12 @@ decimate_1024=600000
 tessimate=1
 tess_size_e5=12000
 camera_mode=auto
-solid_fill=0
+solid_fill=1
 dynamic=0
 polypaint=0
 tex_project=0
 tex_blend=0
 tex_seam=0
-app_version=1.0.4
+speed=hybrid
+app_version=1.2.2
+state_dir=D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_053901\state

ZGenData/samples/pixal3d_20261018_053901/info.txt

@@ -0,0 +1,4 @@
+model=Pixal3D
+seconds=942
+status=OK vertices=874890 faces=1750272 seconds=916.7
+app_version=1.2.2

ZGenData/samples/pixal3d_20261018_053901/log.txt

@@ -0,0 +1,157 @@
+==== Pixal3D engine 0.1 (built Oct 17 2026 22:18:29) run 2026-10-18 05:39:03 ====
+app: ZGen 1.2.2
+system: Windows 10.0 build 19045, 16 CPU threads, RAM 31.9 GB (17.0 GB free)
+gpu: NVIDIA GeForce RTX 2070 SUPER, VRAM 7.8 GB (budget 7.1 GB), driver 32.0.16.1047
+onnxruntime 1.23.2, provider CUDA, exe D:\GitHub\ZGen\app\ZGenData
+input front: D:\GitHub\ZGen\input_Demo\Single\d7fea7e5-7bef-4d08-b298-b6a3903f7f2c.png (1254x1254, 3 ch, 2801 KB)
+options: mesh_res 1536 steps ss/shape/tex 12/12/12 seed 42 texture 1 tex_size 4096 tex_project 0 tessimate 1 tess_size 0.12000 decimate 600000/600000 auto_mask 1
+camera: mode auto fov 20.000 deg distance 3.1192 frame 0.550 crop 1 azimuth 0.0 (auto: final fov/distance below)
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+[OK] Signed license is valid.
+license: Licensed: 26 days left
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+speed: Fast shape + Optimized texture | flow tiles 6000 tokens (overlap 4, one pass while t >= 0.00), decoder slabs above 100 M voxel-channels, fused flow models (*_rg.onnx: RoundedGelu, RmsRope)
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,11 1013x1013 -> -50,-39 1112x1112
+      camera auto: MoGe-2 fov 36.13 deg (15.2s)
+Output: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_053901\model.glb
+seed=42 steps=12/12 provider=CUDA
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,11 1013x1013 -> -50,-39 1112x1112
+      camera front: auto fov 36.126358 deg, distance 1.68639635, crop 1
+[1/6] 1 views (front) 3.1s  [VRAM peak 3576 MiB, now 98 MiB]
+[2/6] dino 512 + ss conditioning 9.5s  [VRAM peak 1514 MiB, now 98 MiB]
+      ss_flow step 1/12 (4096 tokens)
+      ss_flow step 2/12 (4096 tokens)
+      ss_flow step 3/12 (4096 tokens)
+      ss_flow step 4/12 (4096 tokens)
+      ss_flow step 5/12 (4096 tokens)
+      ss_flow step 6/12 (4096 tokens)
+      ss_flow step 7/12 (4096 tokens)
+      ss_flow step 8/12 (4096 tokens)
+      ss_flow step 9/12 (4096 tokens)
+      ss_flow step 10/12 (4096 tokens)
+      ss_flow step 11/12 (4096 tokens)
+      ss_flow step 12/12 (4096 tokens)
+[3/6] sparse structure 45.7s  tokens=4233  [VRAM peak 5156 MiB, now 98 MiB]
+      sh512_flow step 1/12 (4233 tokens)
+      sh512_flow step 2/12 (4233 tokens)
+      sh512_flow step 3/12 (4233 tokens)
+      sh512_flow step 4/12 (4233 tokens)
+      sh512_flow step 5/12 (4233 tokens)
+      sh512_flow step 6/12 (4233 tokens)
+      sh512_flow step 7/12 (4233 tokens)
+      sh512_flow step 8/12 (4233 tokens)
+      sh512_flow step 9/12 (4233 tokens)
+      sh512_flow step 10/12 (4233 tokens)
+      sh512_flow step 11/12 (4233 tokens)
+      sh512_flow step 12/12 (4233 tokens)
+[4/6] shape 512 flow 59.2s  [VRAM peak 5138 MiB, now 102 MiB]
+      level 1: 17957 voxels (64^3)
+      level 2: 74199 voxels (128^3)
+      level 3: 305840 voxels (256^3)
+      level 4: 1262707 voxels (512^3)
+      HR tokens on 96^3: 40321  [VRAM peak 2180 MiB, now 102 MiB]
+[5/6] dino 1024 + NAF + projection 21.7s  [VRAM peak 3702 MiB, now 102 MiB]
+      sh1024_flow: 40321 tokens on 16 tiles (3427..4580 tokens incl. 0 context, overlap 4 cells)
+      sh1024_flow step 1/12 (40321 tokens)
+      sh1024_flow step 2/12 (40321 tokens)
+      sh1024_flow step 3/12 (40321 tokens)
+      sh1024_flow step 4/12 (40321 tokens)
+      sh1024_flow step 5/12 (40321 tokens)
+      sh1024_flow step 6/12 (40321 tokens)
+      sh1024_flow step 7/12 (40321 tokens)
+      sh1024_flow step 8/12 (40321 tokens)
+      sh1024_flow step 9/12 (40321 tokens)
+      sh1024_flow step 10/12 (40321 tokens)
+      sh1024_flow step 11/12 (40321 tokens)
+      sh1024_flow step 12/12 (40321 tokens)
+      shape 1024 flow 313.4s  [VRAM peak 5278 MiB, now 102 MiB]
+      level 1: 169163 voxels (192^3)
+      level 2: 701161 voxels (384^3)
+      level 2 decoder on 2 slabs (halo 11, 179.5 M voxel-channels) 7.7s  [VRAM peak 8220 MiB, now 172 MiB]
+      level 3: 2901564 voxels (768^3)
+      level 3 decoder on 4 slabs (halo 7, 371.4 M voxel-channels) 26.0s  [VRAM peak 8750 MiB, now 130 MiB]
+      level 4: 12054515 voxels (1536^3)
+      level 4 decoder on 8 slabs (halo 2, 771.5 M voxel-channels) 8.1s  [VRAM peak 4884 MiB, now 126 MiB]
+      shape decoder 53.3s  (12054515 voxels at 1536^3)  [VRAM peak 126 MiB, now 102 MiB]
+      mesh: building on a worker thread while the texture flow runs
+      visibility orientation: flipped 1766252 faces
+      tex_flow step 1/12 (40321 tokens)
+      tex_flow step 2/12 (40321 tokens)
+      tex_flow step 3/12 (40321 tokens)
+      solid fill: winding number on 2055888 faces (GPU 11.3s), 432277 cap bricks, 748050 bricks with per-point winding, 170308012 grid points filled 49.1s
+      remesh (dual contouring 1536^3, band 1.0): 12900232 cells -> 12900232 vertices, 25828412 faces (dropped 0 inner) 116.9s
+      removed 632 triangles (inner shell / debris)
+      filled 1 small holes + 0 non-manifold gaps (+34 faces, PolyGroup 2)
+      tex_flow step 4/12 (40321 tokens)
+      tessimate size 0.12000 -> edge 0.00241, about 1740257 faces
+      pre-decimated 25827814 -> 3480454 faces 15.8s
+o-[Validate    ] Removed 9 duplicated vertices
+                 Detected non-manifold vertices
+                    (fixed by generating 19465 new vertices)
+                 (FP64) nb_v:1752763 nb_e:0 nb_f:3476722 nb_b:35836 tri:1 dim:3
+                 Attributes on vertices: point[3]
+      tessimate: repaired in 1.6s -> 3476722 faces
+      tessimate: 7207 border loop(s) before remeshing (diagonal 1.6426)
+      tessimate: closed 3565 loop(s) (0 large, 18 bridges), kept 0 open, 3642 not closable 0.5s
+      tessimate: remeshing (Geogram, all cores)...
+o-[Remesh      ] Start...
+o-[RVD         ] | Computing initial sampling on surface, using dimension=3
+      tex_flow step 5/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 5s
+o-[Lloyd       ] Elapsed time: 4.27s
+      tessimate: remeshing (Geogram, all cores)... 10s
+      tessimate: remeshing (Geogram, all cores)... 15s
+o-[Newton      ] Elapsed time: 11.892s
+o-[Remesh      ] | Computing RVD...
+      tessimate: remeshing (Geogram, all cores)... 20s
+      tex_flow step 6/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 25s
+      tessimate: remeshing (Geogram, all cores)... 30s
+      tessimate: remeshing (Geogram, all cores)... 35s
+      tessimate: remeshing (Geogram, all cores)... 40s
+                 Elapsed: 44.31s
+      tessimate: remeshing (Geogram, all cores)... 45s
+      tex_flow step 7/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 50s
+      tessimate: remeshing (Geogram, all cores)... 55s
+      tessimate: remeshing (Geogram, all cores)... 60s
+      tessimate: remeshing (Geogram, all cores)... 65s
+      tessimate: remeshing (Geogram, all cores)... 70s
+      tex_flow step 8/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 75s
+      tessimate: remeshing (Geogram, all cores)... 80s
+      tessimate: remeshing (Geogram, all cores)... 85s
+      tessimate: remeshing (Geogram, all cores)... 90s
+      tessimate: remeshing (Geogram, all cores)... 95s
+      tex_flow step 9/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 100s
+      tessimate: remeshing (Geogram, all cores)... 105s
+      tessimate: remeshing (Geogram, all cores)... 110s
+      tessimate: remeshing (Geogram, all cores)... 115s
+      tex_flow step 10/12 (40321 tokens)
+      tessimate: remeshing (Geogram, all cores)... 120s
+      tessimate: curvature fill on 511745 vertices in 703 patch(es) (largest 446884, 0 fell back to membrane) 80.6s
+      tessimate: remeshed in 44.3s -> 1749328 faces (5134 loose triangles dropped)
+      tessimate: 3480454 -> 1749328 faces (edge 0.00241): removed 5134 debris parts, closed 3565 holes (3642 failed), 187 border loop(s) left | repair 1.6s holes 1.0s remesh 44.3s
+      [topo] final check                faces 1749328  boundary edges 944  non-manifold 0  components 1 (largest 1749328)
+      [topo] final check                open loops 187 (edges: 32 18 14 12 10 10 ...)  small holes besides the largest loop: 186
+      final fill: closed 187 holes + 0 gaps (+944 faces)
+      [topo] after final fill           faces 1750272  boundary edges 0  non-manifold 0  components 1 (largest 1750272)
+      state: mesh saved (874890 vertices, 1750272 faces) -> D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_053901\state
+      tex_flow step 11/12 (40321 tokens)
+      tex_flow step 12/12 (40321 tokens)
+      texture flow 308.5s  [VRAM peak 6252 MiB, now 102 MiB]
+      level 2 decoder on 2 slabs (halo 11, 179.5 M voxel-channels) 7.4s  [VRAM peak 8218 MiB, now 170 MiB]
+      level 3 decoder on 4 slabs (halo 7, 371.4 M voxel-channels) 21.0s  [VRAM peak 8748 MiB, now 128 MiB]
+      level 4 decoder on 8 slabs (halo 2, 771.5 M voxel-channels) 9.1s  [VRAM peak 4882 MiB, now 124 MiB]
+[6/6] texture flow + decoder 365.0s  [VRAM peak 124 MiB, now 100 MiB]
+      mesh (built alongside the texture flow) ready after 0.0s more
+      UV unwrap: 642 clusters / 12115 charts (planar 173, ortho 1434, lscm 1354, piecewise 9154), atlas 4164x4188, 1019615 vertices 24.6s
+      baked 4096x4096 texture 3.5s
+      texture lookups: 55.8% on the voxels, 16.8% nearest voxel, 27.4% missed (filled from neighbours)  WARNING: many misses, the texture may show blotches
+      GLB: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_053901\model.glb
+[5/5] mesh 30.1s
+SUCCESS OK vertices=874890 faces=1750272 seconds=916.7
+finished in 940.4s, peak VRAM (this process): 8750 MiB

ZGenData/samples/{pixal3d_20261014_155321 → pixal3d_20261018_053901}/model.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8596aba5f66478009f0d3cd55c53d87039637f6adc96e67314aed886f6fc8be
-size 75552264
+oid sha256:a2e30802ffe94bc0a1b03a57b46dab42ba57af015175881163b1331eff032eba
+size 81416896

ZGenData/samples/pixal3d_20261018_053901/run.txt

@@ -0,0 +1,13 @@
+ZGen run log
+app version: 1.2.2
+time: 2026-10-18 05:54:45
+model: Pixal3D
+outcome: finished
+status: OK vertices=874890 faces=1750272 seconds=916.7
+seconds: 942
+exit code 0x00000000
+engine exe: D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+working dir: D:\GitHub\ZGen\app\ZGenData
+command: "D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe" "D:\GitHub\ZGen\input_Demo\Single\d7fea7e5-7bef-4d08-b298-b6a3903f7f2c.png" --out "D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_053901\model.glb" --models "D:\GitHub\ZGen\app\ZGenData\models\pixal3d" --work "D:\GitHub\ZGen\app\ZGenData" --signal --no-window
+engine exe size: 2064384 bytes

ZGenData/samples/pixal3d_20261018_053901/state/latent.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94ad176a4f4289eb40dd6e2a18a6a76cfc7fe6edfd102995cc65f3afcc4ca571
+size 5644960

ZGenData/samples/pixal3d_20261018_053901/state/mesh.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18fcb0476e36a7f20a22df935bd1dc400094f3b132cc65efb59a1b7b840e24fd
+size 31501960

ZGenData/samples/{pixal3d_20261014_155321 → pixal3d_20261018_114858}/config.txt

@@ -1,13 +1,13 @@
 steps=12
 seed=42
-mesh_res=1024
+mesh_res=1536
 mix=0
 texture=1
 texture_size=4096
 decimate=600000
 decimate_1024=600000
 tessimate=1
-tess_size_e5=12000
+tess_size_e5=10000
 camera_mode=auto
 solid_fill=1
 dynamic=0
@@ -15,4 +15,11 @@ polypaint=0
 tex_project=0
 tex_blend=0
 tex_seam=0
-app_version=1.0.6
+speed=hybrid
+mesh_parallel=1
+flow_tile=10000
+flow_tile_ov=8
+flow_tile_ctx=0
+flow_tile_global=0
+app_version=1.2.2
+state_dir=D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_114858\state

ZGenData/samples/pixal3d_20261018_114858/info.txt

@@ -0,0 +1,4 @@
+model=Pixal3D
+seconds=1328
+status=OK vertices=1021211 faces=2042438 seconds=1310.5
+app_version=1.2.2

ZGenData/samples/pixal3d_20261018_114858/log.txt

@@ -0,0 +1,145 @@
+==== Pixal3D engine 0.1 (built Oct 18 2026 11:21:49) run 2026-10-18 11:48:58 ====
+app: ZGen 1.2.2
+system: Windows 10.0 build 19045, 16 CPU threads, RAM 31.9 GB (14.9 GB free)
+gpu: NVIDIA GeForce RTX 2070 SUPER, VRAM 7.8 GB (budget 7.1 GB), driver 32.0.16.1047
+onnxruntime 1.23.2, provider CUDA, exe D:\GitHub\ZGen\app\ZGenData
+input front: D:\GitHub\ZGen\input_Demo\Single\2e7ed9b4-e3a8-45a8-89b1-d402c5b700c0.png (1254x1254, 3 ch, 2645 KB)
+options: mesh_res 1536 steps ss/shape/tex 12/12/12 seed 42 texture 1 tex_size 4096 tex_project 0 tessimate 1 tess_size 0.10000 decimate 600000/600000 auto_mask 1
+camera: mode auto fov 20.000 deg distance 3.1192 frame 0.550 crop 1 azimuth 0.0 (auto: final fov/distance below)
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+[OK] Signed license is valid.
+license: Licensed: 26 days left
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+speed: Fast shape + Optimized texture | flow tiles 10000 tokens (overlap 8, one pass while t >= 0.00), decoder slabs above 100 M voxel-channels, fused flow models (*_rg.onnx: RoundedGelu, RmsRope)
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,8 1024x1016 -> -50,-46 1124x1124
+      camera auto: MoGe-2 fov 28.88 deg (12.9s)
+Output: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_114858\model.glb
+seed=42 steps=12/12 provider=CUDA
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,8 1024x1016 -> -50,-46 1124x1124
+      camera front: auto fov 28.880912 deg, distance 2.13584655, crop 1
+[1/6] 1 views (front) 2.8s  [VRAM peak 3576 MiB, now 98 MiB]
+[2/6] dino 512 + ss conditioning 9.9s  [VRAM peak 1514 MiB, now 98 MiB]
+      ss_flow step 1/12 (4096 tokens)
+      ss_flow step 2/12 (4096 tokens)
+      ss_flow step 3/12 (4096 tokens)
+      ss_flow step 4/12 (4096 tokens)
+      ss_flow step 5/12 (4096 tokens)
+      ss_flow step 6/12 (4096 tokens)
+      ss_flow step 7/12 (4096 tokens)
+      ss_flow step 8/12 (4096 tokens)
+      ss_flow step 9/12 (4096 tokens)
+      ss_flow step 10/12 (4096 tokens)
+      ss_flow step 11/12 (4096 tokens)
+      ss_flow step 12/12 (4096 tokens)
+[3/6] sparse structure 45.3s  tokens=4571  [VRAM peak 5156 MiB, now 98 MiB]
+      sh512_flow step 1/12 (4571 tokens)
+      sh512_flow step 2/12 (4571 tokens)
+      sh512_flow step 3/12 (4571 tokens)
+      sh512_flow step 4/12 (4571 tokens)
+      sh512_flow step 5/12 (4571 tokens)
+      sh512_flow step 6/12 (4571 tokens)
+      sh512_flow step 7/12 (4571 tokens)
+      sh512_flow step 8/12 (4571 tokens)
+      sh512_flow step 9/12 (4571 tokens)
+      sh512_flow step 10/12 (4571 tokens)
+      sh512_flow step 11/12 (4571 tokens)
+      sh512_flow step 12/12 (4571 tokens)
+[4/6] shape 512 flow 59.9s  [VRAM peak 5274 MiB, now 102 MiB]
+      level 1: 18676 voxels (64^3)
+      level 2: 75998 voxels (128^3)
+      level 3: 308622 voxels (256^3)
+      level 4: 1247963 voxels (512^3)
+      HR tokens on 96^3: 41743  [VRAM peak 2198 MiB, now 102 MiB]
+[5/6] dino 1024 + NAF + projection 21.7s  [VRAM peak 3702 MiB, now 102 MiB]
+      sh1024_flow: 41743 tokens on 16 tiles (4987..7756 tokens incl. 0 context, overlap 8 cells)
+      sh1024_flow step 1/12 (41743 tokens)
+      sh1024_flow step 2/12 (41743 tokens)
+      sh1024_flow step 3/12 (41743 tokens)
+      sh1024_flow step 4/12 (41743 tokens)
+      sh1024_flow step 5/12 (41743 tokens)
+      sh1024_flow step 6/12 (41743 tokens)
+      sh1024_flow step 7/12 (41743 tokens)
+      sh1024_flow step 8/12 (41743 tokens)
+      sh1024_flow step 9/12 (41743 tokens)
+      sh1024_flow step 10/12 (41743 tokens)
+      sh1024_flow step 11/12 (41743 tokens)
+      sh1024_flow step 12/12 (41743 tokens)
+      shape 1024 flow 544.4s  [VRAM peak 6894 MiB, now 102 MiB]
+      level 1: 169952 voxels (192^3)
+      level 2: 688773 voxels (384^3)
+      level 2 decoder on 2 slabs (halo 11, 176.3 M voxel-channels) 31.7s  [VRAM peak 8038 MiB, now 172 MiB]
+      level 3: 2784957 voxels (768^3)
+      level 3 decoder on 4 slabs (halo 7, 356.5 M voxel-channels) 118.6s  [VRAM peak 8272 MiB, now 130 MiB]
+      level 4: 11279166 voxels (1536^3)
+      level 4 decoder on 8 slabs (halo 2, 721.9 M voxel-channels) 8.7s  [VRAM peak 4598 MiB, now 126 MiB]
+      shape decoder 171.1s  (11279166 voxels at 1536^3)  [VRAM peak 126 MiB, now 102 MiB]
+      mesh: building on a worker thread while the texture flow runs
+      visibility orientation: flipped 517787 faces
+      tex_flow step 1/12 (41743 tokens)
+      tex_flow step 2/12 (41743 tokens)
+      solid fill: winding number on 370423 faces (GPU 7.3s), 357582 cap bricks, 681613 bricks with per-point winding, 164340420 grid points filled 37.4s
+      remesh (dual contouring 1536^3, band 1.0): 11483778 cells -> 11483778 vertices, 22972312 faces (dropped 0 inner) 98.1s
+      removed 208 triangles (inner shell / debris)
+      filled 0 small holes + 0 non-manifold gaps (+0 faces, PolyGroup 2)
+      tessimate size 0.10000 -> edge 0.00198, about 2226036 faces
+      WARNING: 2226036 faces is too dense for a 4096x4096 texture atlas: tessimate to 2097152 faces instead (raise the poly size or the texture size)
+      tex_flow step 3/12 (41743 tokens)
+      pre-decimated 22972104 -> 4194304 faces 14.3s
+o-[Validate    ] Removed 9 duplicated vertices
+                 Detected non-manifold vertices
+                    (fixed by generating 6151 new vertices)
+                 (FP64) nb_v:2102014 nb_e:0 nb_f:4192713 nb_b:12331 tri:1 dim:3
+                 Attributes on vertices: point[3]
+      tessimate: repaired in 2.0s -> 4192713 faces
+      tessimate: 2419 border loop(s) before remeshing (diagonal 1.6202)
+      tessimate: closed 1547 loop(s) (0 large, 10 bridges), kept 0 open, 872 not closable 0.6s
+      tessimate: remeshing (Geogram, all cores)...
+o-[Remesh      ] Start...
+o-[RVD         ] | Computing initial sampling on surface, using dimension=3
+      tessimate: remeshing (Geogram, all cores)... 5s
+o-[Lloyd       ] Elapsed time: 5.851s
+      tex_flow step 4/12 (41743 tokens)
+      tessimate: remeshing (Geogram, all cores)... 10s
+      tessimate: remeshing (Geogram, all cores)... 15s
+      tessimate: remeshing (Geogram, all cores)... 20s
+o-[Newton      ] Elapsed time: 15.09s
+o-[Remesh      ] | Computing RVD...
+      tessimate: remeshing (Geogram, all cores)... 25s
+      tessimate: remeshing (Geogram, all cores)... 30s
+      tex_flow step 5/12 (41743 tokens)
+      tessimate: remeshing (Geogram, all cores)... 35s
+      tessimate: remeshing (Geogram, all cores)... 40s
+      tessimate: remeshing (Geogram, all cores)... 45s
+      tessimate: remeshing (Geogram, all cores)... 50s
+      tessimate: remeshing (Geogram, all cores)... 55s
+                 Elapsed: 55.413s
+      tessimate: patch smoothing off
+      tessimate: remeshed in 55.4s -> 2042136 faces (55011 loose triangles dropped)
+      tessimate: 4194304 -> 2042136 faces (edge 0.00000): removed 55011 debris parts, closed 1547 holes (872 failed), 58 border loop(s) left | repair 2.0s holes 1.2s remesh 55.4s
+      [topo] final check                faces 2042136  boundary edges 302  non-manifold 0  components 1 (largest 2042136)
+      [topo] final check                open loops 58 (edges: 24 14 10 9 8 8 ...)  small holes besides the largest loop: 57
+      final fill: closed 58 holes + 0 gaps (+302 faces)
+      [topo] after final fill           faces 2042438  boundary edges 0  non-manifold 0  components 1 (largest 2042438)
+      state: mesh saved (1021211 vertices, 2042438 faces) -> D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_114858\state
+      tex_flow step 6/12 (41743 tokens)
+      tex_flow step 7/12 (41743 tokens)
+      tex_flow step 8/12 (41743 tokens)
+      tex_flow step 9/12 (41743 tokens)
+      tex_flow step 10/12 (41743 tokens)
+      tex_flow step 11/12 (41743 tokens)
+      tex_flow step 12/12 (41743 tokens)
+      texture flow 345.1s  [VRAM peak 6318 MiB, now 102 MiB]
+      level 2 decoder on 2 slabs (halo 11, 176.3 M voxel-channels) 13.9s  [VRAM peak 8038 MiB, now 172 MiB]
+      level 3 decoder on 4 slabs (halo 7, 356.5 M voxel-channels) 22.6s  [VRAM peak 8272 MiB, now 130 MiB]
+      level 4 decoder on 8 slabs (halo 2, 721.9 M voxel-channels) 8.6s  [VRAM peak 4598 MiB, now 126 MiB]
+[6/6] texture flow + decoder 409.7s  [VRAM peak 126 MiB, now 102 MiB]
+      mesh (built alongside the texture flow) ready after 0.0s more
+      UV unwrap: 568 clusters / 3383 charts (planar 13, ortho 491, lscm 548, piecewise 2331), atlas 4324x4304, 1111541 vertices 25.7s
+      baked 4096x4096 texture 2.2s
+      texture lookups: 88.0% on the voxels, 11.6% nearest voxel, 0.4% missed (filled from neighbours)
+      GLB: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_114858\model.glb
+[5/5] mesh 29.8s
+SUCCESS OK vertices=1021211 faces=2042438 seconds=1310.5
+finished in 1328.0s, peak VRAM (this process): 8272 MiB

ZGenData/samples/{trellis2_20261013_100355 → pixal3d_20261018_114858}/model.glb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6edf8efd2c9df99f5683442b9a687ff320afd3cbd2b04dfba178902b906f257b
-size 27314332
+oid sha256:f2463e2371054a51969b80fd33489beefb019d5b0f3bdcc27ba07a61ed512c28
+size 81774628

ZGenData/samples/pixal3d_20261018_114858/run.txt

@@ -0,0 +1,13 @@
+ZGen run log
+app version: 1.2.2
+time: 2026-10-18 12:11:09
+model: Pixal3D
+outcome: finished
+status: OK vertices=1021211 faces=2042438 seconds=1310.5
+seconds: 1328
+exit code 0x00000000
+engine exe: D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+working dir: D:\GitHub\ZGen\app\ZGenData
+command: "D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe" "D:\GitHub\ZGen\input_Demo\Single\2e7ed9b4-e3a8-45a8-89b1-d402c5b700c0.png" --out "D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_114858\model.glb" --models "D:\GitHub\ZGen\app\ZGenData\models\pixal3d" --work "D:\GitHub\ZGen\app\ZGenData" --signal --no-window
+engine exe size: 2066944 bytes

ZGenData/samples/pixal3d_20261018_114858/state/latent.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ed2cc06a489c47af690ce3b64dfef503cc46beb4d1a59ad9e5c9cdd46d85b5e
+size 5844040

ZGenData/samples/pixal3d_20261018_114858/state/mesh.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b830006d45a6447d8ac2aa0dbbf464d0473d5e23ed6465aafeef8fd56997e0c3
+size 36763804

ZGenData/samples/pixal3d_20261018_130210/config.txt

@@ -0,0 +1,25 @@
+steps=12
+seed=42
+mesh_res=1536
+mix=0
+texture=1
+texture_size=4096
+decimate=600000
+decimate_1024=600000
+tessimate=1
+tess_size_e5=10000
+camera_mode=auto
+solid_fill=1
+dynamic=0
+polypaint=0
+tex_project=0
+tex_blend=0
+tex_seam=0
+speed=hybrid
+mesh_parallel=1
+flow_tile=10000
+flow_tile_ov=8
+flow_tile_ctx=0
+flow_tile_global=0
+app_version=1.2.4
+state_dir=D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\state

ZGenData/samples/pixal3d_20261018_130210/config_texture.txt

@@ -0,0 +1,26 @@
+steps=12
+seed=42
+mesh_res=1536
+mix=0
+texture=1
+texture_size=4096
+decimate=600000
+decimate_1024=600000
+tessimate=1
+tess_size_e5=10000
+camera_mode=auto
+solid_fill=1
+dynamic=0
+polypaint=0
+tex_project=0
+tex_blend=0
+tex_seam=0
+speed=hybrid
+mesh_parallel=1
+flow_tile=10000
+flow_tile_ov=8
+flow_tile_ctx=0
+flow_tile_global=0
+app_version=1.2.4
+texture=1
+texture_from=D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\state

ZGenData/samples/pixal3d_20261018_130210/info.txt

@@ -0,0 +1,4 @@
+model=Pixal3D
+seconds=614
+status=OK vertices=1023079 faces=2046186 seconds=603.8
+app_version=1.2.4

ZGenData/samples/pixal3d_20261018_130210/log.txt

@@ -0,0 +1,143 @@
+==== Pixal3D engine 0.1 (built Oct 18 2026 12:59:42) run 2026-10-18 13:02:10 ====
+app: ZGen 1.2.4
+system: Windows 10.0 build 19045, 16 CPU threads, RAM 31.9 GB (15.6 GB free)
+gpu: NVIDIA GeForce RTX 2070 SUPER, VRAM 7.8 GB (budget 7.1 GB), driver 32.0.16.1047
+onnxruntime 1.23.2, provider CUDA, exe D:\GitHub\ZGen\app\ZGenData
+input front: C:\Users\User\AppData\Local\Temp\ZGen\views\e510ea80-9cc5-4667-84bd-8f79e325bdb0_nomarker.png (1254x1254, 3 ch, 2480 KB)
+options: mesh_res 1536 steps ss/shape/tex 12/12/12 seed 42 texture 1 tex_size 4096 tex_project 0 tessimate 1 tess_size 0.10000 decimate 600000/600000 auto_mask 1
+camera: mode auto fov 20.000 deg distance 3.1192 frame 0.550 crop 1 azimuth 0.0 (auto: final fov/distance below)
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+[OK] Signed license is valid.
+license: Licensed: 26 days left
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+speed: Fast shape + Optimized texture | flow tiles 10000 tokens (overlap 8, one pass while t >= 0.00), decoder slabs above 100 M voxel-channels, fused flow models (*_rg.onnx: RoundedGelu, RmsRope)
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,65 1024x959 -> -50,-18 1124x1124
+      camera auto: MoGe-2 fov 34.55 deg (7.7s)
+Output: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\model.glb
+seed=42 steps=12/12 provider=CUDA
+      front: background removed (BEN2_Base.onnx)
+      crop: alpha bbox 0,65 1024x959 -> -50,-18 1124x1124
+      camera front: auto fov 34.553939 deg, distance 1.76834987, crop 1
+[1/6] 1 views (front) 2.9s  [VRAM peak 3576 MiB, now 98 MiB]
+[2/6] dino 512 + ss conditioning 9.7s  [VRAM peak 1514 MiB, now 98 MiB]
+      ss_flow step 1/12 (4096 tokens)
+      ss_flow step 2/12 (4096 tokens)
+      ss_flow step 3/12 (4096 tokens)
+      ss_flow step 4/12 (4096 tokens)
+      ss_flow step 5/12 (4096 tokens)
+      ss_flow step 6/12 (4096 tokens)
+      ss_flow step 7/12 (4096 tokens)
+      ss_flow step 8/12 (4096 tokens)
+      ss_flow step 9/12 (4096 tokens)
+      ss_flow step 10/12 (4096 tokens)
+      ss_flow step 11/12 (4096 tokens)
+      ss_flow step 12/12 (4096 tokens)
+[3/6] sparse structure 43.0s  tokens=3094  [VRAM peak 4956 MiB, now 98 MiB]
+      sh512_flow step 1/12 (3094 tokens)
+      sh512_flow step 2/12 (3094 tokens)
+      sh512_flow step 3/12 (3094 tokens)
+      sh512_flow step 4/12 (3094 tokens)
+      sh512_flow step 5/12 (3094 tokens)
+      sh512_flow step 6/12 (3094 tokens)
+      sh512_flow step 7/12 (3094 tokens)
+      sh512_flow step 8/12 (3094 tokens)
+      sh512_flow step 9/12 (3094 tokens)
+      sh512_flow step 10/12 (3094 tokens)
+      sh512_flow step 11/12 (3094 tokens)
+      sh512_flow step 12/12 (3094 tokens)
+[4/6] shape 512 flow 44.9s  [VRAM peak 3614 MiB, now 102 MiB]
+      level 1: 12552 voxels (64^3)
+      level 2: 51032 voxels (128^3)
+      level 3: 206984 voxels (256^3)
+      level 4: 835964 voxels (512^3)
+      HR tokens on 96^3: 28043  [VRAM peak 1506 MiB, now 102 MiB]
+[5/6] dino 1024 + NAF + projection 21.5s  [VRAM peak 3702 MiB, now 102 MiB]
+      sh1024_flow: 28043 tokens on 8 tiles (5983..7100 tokens incl. 0 context, overlap 8 cells)
+      sh1024_flow step 1/12 (28043 tokens)
+      sh1024_flow step 2/12 (28043 tokens)
+      sh1024_flow step 3/12 (28043 tokens)
+      sh1024_flow step 4/12 (28043 tokens)
+      sh1024_flow step 5/12 (28043 tokens)
+      sh1024_flow step 6/12 (28043 tokens)
+      sh1024_flow step 7/12 (28043 tokens)
+      sh1024_flow step 8/12 (28043 tokens)
+      sh1024_flow step 9/12 (28043 tokens)
+      sh1024_flow step 10/12 (28043 tokens)
+      sh1024_flow step 11/12 (28043 tokens)
+      sh1024_flow step 12/12 (28043 tokens)
+      shape 1024 flow 213.0s  [VRAM peak 4448 MiB, now 102 MiB]
+      level 1: 113792 voxels (192^3)
+      level 2: 459406 voxels (384^3)
+      level 2 decoder on 2 slabs (halo 11, 117.6 M voxel-channels) 4.3s  [VRAM peak 5412 MiB, now 172 MiB]
+      level 3: 1849790 voxels (768^3)
+      level 3 decoder on 3 slabs (halo 7, 236.8 M voxel-channels) 7.0s  [VRAM peak 7304 MiB, now 130 MiB]
+      level 4: 7448070 voxels (1536^3)
+      level 4 decoder on 5 slabs (halo 2, 476.7 M voxel-channels) 5.2s  [VRAM peak 3854 MiB, now 126 MiB]
+      shape decoder 24.4s  (7448070 voxels at 1536^3)  [VRAM peak 126 MiB, now 102 MiB]
+      mesh: building on a worker thread while the texture flow runs
+      visibility orientation: flipped 148650 faces
+      tex_flow step 1/12 (28043 tokens)
+      tex_flow step 2/12 (28043 tokens)
+      solid fill: winding number on 299861 faces (GPU 4.2s), 321434 cap bricks, 540753 bricks with per-point winding, 125172238 grid points filled 24.9s
+      tex_flow step 3/12 (28043 tokens)
+      remesh (dual contouring 1536^3, band 1.0): 9876460 cells -> 9876460 vertices, 19759028 faces (dropped 0 inner) 63.7s
+      removed 12 triangles (inner shell / debris)
+      filled 0 small holes + 0 non-manifold gaps (+0 faces, PolyGroup 2)
+      tessimate size 0.10000 -> edge 0.00189, about 2061711 faces
+      tex_flow step 4/12 (28043 tokens)
+      pre-decimated 19759016 -> 4123420 faces 10.9s
+o-[Validate    ] Removed 1 duplicated vertices
+                 Detected non-manifold vertices
+                    (fixed by generating 5203 new vertices)
+                 (FP64) nb_v:2065399 nb_e:0 nb_f:4122424 nb_b:10004 tri:1 dim:3
+                 Attributes on vertices: point[3]
+      tessimate: repaired in 1.8s -> 4122424 faces
+      tessimate: 2034 border loop(s) before remeshing (diagonal 1.5462)
+      tessimate: closed 962 loop(s) (0 large, 7 bridges), kept 0 open, 1072 not closable 0.6s
+      tex_flow step 5/12 (28043 tokens)
+      tessimate: remeshing (Geogram, all cores)...
+o-[Remesh      ] Start...
+o-[RVD         ] | Computing initial sampling on surface, using dimension=3
+      tessimate: remeshing (Geogram, all cores)... 5s
+o-[Lloyd       ] Elapsed time: 5.241s
+      tessimate: remeshing (Geogram, all cores)... 10s
+      tex_flow step 6/12 (28043 tokens)
+      tessimate: remeshing (Geogram, all cores)... 15s
+      tessimate: remeshing (Geogram, all cores)... 20s
+o-[Newton      ] Elapsed time: 14.716s
+o-[Remesh      ] | Computing RVD...
+      tex_flow step 7/12 (28043 tokens)
+      tessimate: remeshing (Geogram, all cores)... 25s
+      tessimate: remeshing (Geogram, all cores)... 30s
+      tex_flow step 8/12 (28043 tokens)
+      tessimate: remeshing (Geogram, all cores)... 35s
+      tessimate: remeshing (Geogram, all cores)... 40s
+      tex_flow step 9/12 (28043 tokens)
+      tessimate: remeshing (Geogram, all cores)... 45s
+                 Elapsed: 49.3s
+      tessimate: patch smoothing off
+      tessimate: remeshed in 49.3s -> 2045839 faces (13172 loose triangles dropped)
+      tessimate: 4123420 -> 2045839 faces (edge 0.00189): removed 13172 debris parts, closed 962 holes (1072 failed), 57 border loop(s) left | repair 1.8s holes 1.1s remesh 49.3s
+      normals were inverted: flipped all faces
+      [topo] final check                faces 2045839  boundary edges 347  non-manifold 0  components 1 (largest 2045839)
+      [topo] final check                open loops 57 (edges: 32 26 18 14 14 10 ...)  small holes besides the largest loop: 56
+      final fill: closed 57 holes + 0 gaps (+347 faces)
+      [topo] after final fill           faces 2046186  boundary edges 0  non-manifold 0  components 1 (largest 2046186)
+      state: mesh saved (1023079 vertices, 2046186 faces) -> D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\state
+      tex_flow step 10/12 (28043 tokens)
+      tex_flow step 11/12 (28043 tokens)
+      tex_flow step 12/12 (28043 tokens)
+      texture flow 165.3s  [VRAM peak 5218 MiB, now 102 MiB]
+      level 2 decoder on 2 slabs (halo 11, 117.6 M voxel-channels) 4.1s  [VRAM peak 5412 MiB, now 172 MiB]
+      level 3 decoder on 3 slabs (halo 7, 236.8 M voxel-channels) 6.8s  [VRAM peak 7304 MiB, now 130 MiB]
+      level 4 decoder on 5 slabs (halo 2, 476.7 M voxel-channels) 4.9s  [VRAM peak 4816 MiB, now 126 MiB]
+[6/6] texture flow + decoder 199.0s  [VRAM peak 126 MiB, now 102 MiB]
+      mesh (built alongside the texture flow) ready after 0.0s more
+      UV unwrap: 569 clusters / 4262 charts (planar 5, ortho 607, lscm 537, piecewise 3113), atlas 4360x4352, 1116107 vertices 25.5s
+      baked 4096x4096 texture 3.0s
+      texture lookups: 65.4% on the voxels, 5.2% nearest voxel, 29.5% missed (filled from neighbours)  WARNING: many misses, the texture may show blotches
+      GLB: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\model.glb
+[5/5] mesh 30.4s
+SUCCESS OK vertices=1023079 faces=2046186 seconds=603.8
+finished in 614.7s, peak VRAM (this process): 7304 MiB

ZGenData/samples/pixal3d_20261018_130210/log_texture.txt

@@ -0,0 +1,13 @@
+==== Pixal3D engine 0.1 (built Oct 18 2026 12:59:42) run 2026-10-18 13:30:30 ====
+app: ZGen 1.2.4
+system: Windows 10.0 build 19045, 16 CPU threads, RAM 31.9 GB (16.2 GB free)
+gpu: NVIDIA GeForce RTX 2070 SUPER, VRAM 7.8 GB (budget 7.1 GB), driver 32.0.16.1047
+onnxruntime 1.23.2, provider CUDA, exe D:\GitHub\ZGen\app\ZGenData
+input front: D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\input.png (1254x1254, 3 ch, 2480 KB)
+options: mesh_res 1536 steps ss/shape/tex 12/12/12 seed 42 texture 1 tex_size 4096 tex_project 0 tessimate 1 tess_size 0.10000 decimate 600000/600000 auto_mask 1
+camera: mode auto fov 20.000 deg distance 3.1192 frame 0.550 crop 1 azimuth 0.0 (auto: final fov/distance below)
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+[OK] Signed license is valid.
+license: Licensed: 26 days left
+perf: VRAM 7.8 GB -> CFG pair in one call up to 8000 tokens (8 GB layout)
+speed: Fast shape + Optimized texture | flow tiles 10000 tokens (overlap 8, one pass while t >= 0.00), decoder slabs above 100 M voxel-channels, fused flow models (*_rg.onnx: RoundedGelu, RmsRope)

ZGenData/samples/pixal3d_20261018_130210/model_mesh.glb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33ed4ab159d6733fe3378340f1a52017a891fbda39f574cfd3148b61a57e0fbb
+size 89199544

ZGenData/samples/pixal3d_20261018_130210/run.txt

@@ -0,0 +1,13 @@
+ZGen run log
+app version: 1.2.4
+time: 2026-10-18 13:12:26
+model: Pixal3D
+outcome: finished
+status: OK vertices=1023079 faces=2046186 seconds=603.8
+seconds: 614
+exit code 0x00000000
+engine exe: D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe
+models: D:\GitHub\ZGen\app\ZGenData\models\pixal3d
+working dir: D:\GitHub\ZGen\app\ZGenData
+command: "D:\GitHub\ZGen\app\ZGenData\Pixal3D.exe" "C:\Users\User\AppData\Local\Temp\ZGen\views\e510ea80-9cc5-4667-84bd-8f79e325bdb0_nomarker.png" --out "D:\GitHub\ZGen\app\ZGenData\output\pixal3d_20261018_130210\model.glb" --models "D:\GitHub\ZGen\app\ZGenData\models\pixal3d" --work "D:\GitHub\ZGen\app\ZGenData" --signal --no-window
+engine exe size: 2066944 bytes