Add files using upload-large-folder tool

.gitattributes

@@ -25,3 +25,6 @@
 *.md text
 *.txt text
 *.py text eol=lf
+dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_explode_pause_fit_back.mp4 filter=lfs diff=lfs merge=lfs -text
+dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_spiral.mp4 filter=lfs diff=lfs merge=lfs -text
+dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_explode_pause_fit_back_then_spiral.mp4 filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -1,13 +1,19 @@
-Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
+SuperFit Primitive Assembly Release License
 
 Copyright (c) 2026 Aditya Ganeshan and the SuperFit authors.
 
-The materials in this repository (the "Dataset") — including the primitive
-assembly artifacts, configuration files, evaluation summaries, manifests, and
-accompanying documentation — are released for non-commercial research use only.
-By accessing or using the Dataset, you agree to the terms of the Creative
-Commons Attribution-NonCommercial 4.0 International Public License
-("CC BY-NC 4.0"), the full text of which is available at:
+This release uses separate licenses for dataset materials and release helper
+code.
+
+1. Dataset Materials and Documentation: CC BY-NC 4.0
+----------------------------------------------------
+
+The dataset materials in this repository (the "Dataset") - including primitive
+assembly artifacts, configuration files, evaluation summaries, manifests,
+metadata, provenance notes, and accompanying documentation - are released for
+non-commercial research use only under the Creative Commons
+Attribution-NonCommercial 4.0 International Public License ("CC BY-NC 4.0").
+The full text is available at:
 
     https://creativecommons.org/licenses/by-nc/4.0/legalcode
 
@@ -16,8 +22,8 @@ Summary of permissions and conditions
 
 You are free to:
 
-  - Share — copy and redistribute the Dataset in any medium or format.
-  - Adapt — remix, transform, and build upon the Dataset.
+  - Share - copy and redistribute the Dataset in any medium or format.
+  - Adapt - remix, transform, and build upon the Dataset.
 
 Under the following terms:
 
@@ -44,11 +50,11 @@ or in part, requires prior written permission from the authors.
 Third-party assets
 ------------------
 
-The Dataset is *derived* from publicly available 3D shape benchmarks
-(Toys4K and PartObjaverse / Objaverse). The terms above govern the derived
-artifacts authored by the SuperFit team. The original 3D meshes are NOT
-redistributed in this repository and remain subject to their own licenses and
-access requirements:
+The Dataset is derived from public 3D shape benchmarks (Toys4K and
+PartObjaverse / Objaverse). The terms above govern the derived artifacts
+authored by the SuperFit team. The original 3D meshes are NOT redistributed in
+this repository and remain subject to their own licenses and access
+requirements:
 
   - Toys4K: <https://rehg.org/publication/dataset2/> (request via the
     authors' form).
@@ -58,8 +64,8 @@ access requirements:
 Users must independently comply with those upstream terms when obtaining or
 using the source meshes.
 
-Disclaimer
-----------
+Disclaimer for Dataset Materials
+--------------------------------
 
 THE DATASET IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
@@ -68,3 +74,46 @@ SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES, OR
 OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT, OR OTHERWISE,
 ARISING FROM, OUT OF, OR IN CONNECTION WITH THE DATASET OR THE USE OR OTHER
 DEALINGS IN THE DATASET.
+
+
+2. Release Helper Code: MIT License
+-----------------------------------
+
+The small release helper scripts in this repository are licensed under the MIT
+License. This includes:
+
+  - load_release.py
+  - examples/*.py
+  - scripts/*.py
+
+These helpers are provided for dataset discovery, loading examples, manifest
+maintenance, and validation. They do not include or relicense the SuperFit
+codebase.
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of the release helper code (the "Software"), to deal in the Software without
+restriction, including without limitation the
+rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+sell copies of the Software, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+
+3. SuperFit Codebase
+--------------------
+
+The SuperFit codebase itself is not part of this dataset license grant.
+SuperFit is distributed separately under its own license; see:
+
+    https://github.com/BardOfCodes/superfit

README.md

@@ -37,7 +37,9 @@ configs:
 
 Pre-computed **primitive assemblies** produced by [SuperFit](https://github.com/BardOfCodes/superfit) (CVPR 2026) on two public 3D shape benchmarks. Each instance stores the fitted primitive parameters, optimization statistics, and optional per-instance evaluation metrics as serialized Python pickles alongside the hyperparameter `config.json` used for fitting.
 
-> **Intended use:** This release is provided **for non-commercial research use only** under the [CC BY-NC 4.0](LICENSE) license. Source 3D meshes are **not** redistributed here; you must obtain [Toys4K](https://rehg.org/publication/dataset2/) and [PartObjaverse](https://partobjaverse.github.io/) / [Objaverse](https://objaverse.allenai.org/) under their respective terms before comparing to ground-truth geometry. Commercial use of this Dataset, in whole or in part, requires prior written permission from the authors.
+The manifest files can be inspected with standard-library Python only. Loading primitive-assembly pickles, recovering expressions, or generating meshes requires the [SuperFit codebase](https://github.com/BardOfCodes/superfit) and its runtime dependencies.
+
+> **Intended use:** The dataset materials in this release are provided **for non-commercial research use only** under [CC BY-NC 4.0](LICENSE). The small release helper scripts are MIT-licensed; see [LICENSE](LICENSE) for scope. Source 3D meshes are **not** redistributed here; you must obtain [Toys4K](https://rehg.org/publication/dataset2/) and [PartObjaverse](https://partobjaverse.github.io/) / [Objaverse](https://objaverse.allenai.org/) under their respective terms before comparing to ground-truth geometry. Commercial use of this Dataset, in whole or in part, requires prior written permission from the authors.
 
 ## Dataset structure
 
@@ -45,7 +47,7 @@ Pre-computed **primitive assemblies** produced by [SuperFit](https://github.com/
 sf_release/
 ├── manifest.jsonl          # one row per (subset, method, object)
 ├── metadata.json           # counts and schema summary
-├── load_release.py         # stdlib manifest loader
+├── load_release.py         # stdlib manifest index + pickle helper
 ├── scripts/                # build, sanitize, validate (maintainers)
 ├── examples/               # usage examples
 └── dataset/                # primitive-assembly artifacts
@@ -93,9 +95,9 @@ Method-level aggregates (where present):
 
 ### Pickle schema
 
-`primitive_assembly.pkl` is a **nested Python `dict`** produced by SuperFit's statistics manager. Keys are dot-separated paths (e.g. `evaluation.iou@128`, `input_mesh_file`, timing scopes). Values are floats, strings, NumPy arrays, nested dicts, or serialized expression metadata. **Loading requires Python 3.10+** and typically **PyTorch** / **NumPy** because tensors may be embedded.
+`primitive_assembly.pkl` is a **flat Python `dict`** produced by SuperFit's statistics manager. Keys are dot-separated paths (e.g. `iter_0.pruned_program`, `evaluation.iou@128`, `input_mesh_file`, timing scopes). Values are floats, strings, NumPy arrays, PyTorch tensors, nested dicts, or serialized SuperFit / GeoLIPI expression metadata. **Loading primitive assemblies requires Python 3.10+ and usually the SuperFit runtime**, including PyTorch / NumPy and the expression classes used by the saved programs.
 
-See the [SuperFit repository](https://github.com/BardOfCodes/superfit) for code that consumes these artifacts.
+See the [SuperFit repository](https://github.com/BardOfCodes/superfit) for the supported runtime and code that consumes these artifacts.
 
 ### Pickle safety
 
@@ -114,7 +116,7 @@ cd superfit-primitive-assemblies
 git lfs pull
 ```
 
-### Manifest loader (stdlib only)
+### Manifest discovery (stdlib only)
 
 ```python
 from pathlib import Path
@@ -125,7 +127,23 @@ print(load_metadata(root)["total_instances"])  # 6700
 
 index = ReleaseIndex(root)
 row = index.get("toys4k", "superfrustum", "airplane_002")
-assembly = index.load(row)  # pickle — trusted source only
+print(row["primitive_assembly_path"])
+```
+
+### Loading artifacts (requires SuperFit runtime)
+
+Primitive assemblies are Python pickles. Unpickling them can require SuperFit,
+PyTorch, NumPy, and expression dependencies from the
+[SuperFit repository](https://github.com/BardOfCodes/superfit):
+
+```python
+from pathlib import Path
+from load_release import ReleaseIndex
+
+root = Path(".")
+index = ReleaseIndex(root)
+row = index.get("toys4k", "superfrustum", "airplane_002")
+assembly = index.load(row)  # trusted pickle; requires SuperFit/PyTorch stack
 print(type(assembly), list(assembly.keys())[:5])
 ```
 
@@ -136,6 +154,55 @@ python examples/load_artifact.py --source toys4k --method cuboid --object-id air
 python examples/load_artifact.py --source partobjaverse --method superfrustum --artifact eval
 ```
 
+### Export a mesh with SuperFit
+
+The released files store primitive expressions, not pre-meshed OBJ/GLB files.
+To generate a mesh, install or add [SuperFit](https://github.com/BardOfCodes/superfit)
+to your Python environment, load a saved expression, evaluate it as an SDF, and
+use SuperFit's mesh extraction helper:
+
+```python
+from pathlib import Path
+
+from geolipi.torch_compute import Sketcher, recursive_evaluate
+from load_release import ReleaseIndex
+from superfit.symbolic.utils import fetch_singular_expr_eval
+from superfit.utils.io import get_best_expr
+from superfit.utils.mesh_sdf import sdf_to_mesh
+
+root = Path(".")
+index = ReleaseIndex(root)
+row = index.get("toys4k", "superfrustum", "airplane_002")
+info = index.load(row)  # trusted pickle; requires SuperFit runtime
+
+expr = get_best_expr(info, prog_type="pruned_program")
+expr = fetch_singular_expr_eval(
+    expr.tensor(device="cuda"),
+    temperature=10000.0,
+    relaxed_eval=True,
+    remove_marker=True,
+    device="cuda",
+)
+
+sketcher = Sketcher(resolution=128, n_dims=3, device="cuda")
+sdf = recursive_evaluate(expr.tensor(device="cuda"), sketcher)
+mesh = sdf_to_mesh(sdf, sketcher)
+mesh.export("airplane_002.obj")
+```
+
+The same flow is available as:
+
+```bash
+python examples/export_mesh_with_superfit.py \
+  --source toys4k \
+  --method superfrustum \
+  --object-id airplane_002 \
+  --output airplane_002.obj
+```
+
+SuperFit's current mesh extraction path uses PyTorch/Kaolin/FlexiCubes and is
+normally run on CUDA.
+
 ### Rebuild manifest / validate before upload
 
 Maintainers can regenerate indexes and run checks from the repo root:
@@ -152,20 +219,22 @@ python scripts/validate_release.py
 - **Not a Parquet/Arrow dataset:** the Hugging Face Dataset Viewer will not render pickle contents; use `manifest.jsonl` for discovery.
 - **No source meshes:** only fitted assemblies and configs are provided.
 - **Incomplete eval coverage:** `dataset/toys4k/superfrustum` has per-instance eval pickles on a subset of instances; see `has_primitive_assembly_eval` in the manifest.
-- **Environment coupling:** unpickling may require the SuperFit stack even though the manifest loader itself is stdlib-only.
+- **Environment coupling:** manifest discovery is stdlib-only, but unpickling primitive assemblies, recovering expressions, and exporting meshes require the SuperFit stack.
 
 ## Licensing and provenance
 
 | Component | License / terms |
 |-----------|-----------------|
-| **This release** (assemblies, configs, manifests) | [CC BY-NC 4.0](LICENSE) — non-commercial research use only |
+| **Dataset materials** (assemblies, configs, eval summaries, manifests, metadata, docs) | [CC BY-NC 4.0](LICENSE) - non-commercial research use only |
+| **Release helper code** (`load_release.py`, `examples/*.py`, `scripts/*.py`) | [MIT](LICENSE) |
+| **SuperFit codebase** | Distributed separately under its own license; see [SuperFit](https://github.com/BardOfCodes/superfit) |
 | **Toys4K** source meshes | Request access via the [authors' form](https://forms.gle/w7Zf82umwaKxr9L7A); follow their terms |
 | **PartObjaverse** (`dataset/partobjaverse/`) | Derived from [SAMPart3D](https://arxiv.org/abs/2411.07184) / Objaverse assets; cite Yang et al. (2024) below |
 | **Objaverse** source meshes | Per-object licenses (CC-BY, CC-BY-NC, etc.); see [Objaverse](https://objaverse.allenai.org/) |
 
 We recommend publishing this Hugging Face repo as **gated** until you confirm redistribution of derived fits is compatible with your Toys4K and Objaverse agreements.
 
-Details: [PROVENANCE.md](PROVENANCE.md).
+This dataset release is **not** licensed under the Adobe Research License; that license applies to the separate SuperFit codebase. Details on source datasets and release posture: [PROVENANCE.md](PROVENANCE.md).
 
 ## Citation
 

dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_explode_pause_fit_back.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9debe81b7edc0dbf2fb501525a11ffbc53adb7ff28eb24dd339fb4ffacea6e64
+size 382364

dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_explode_pause_fit_back_then_spiral.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:298fd098531b0d7f1ceb6a6f3cd236fb981088fc9e1780b56b87a56a949abf6a
+size 1390167

dataset/partobjaverse/superfrustum/b7fce3003d95479a9226f7cb639df893/primitive_assembly_spiral.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74467e826acfa63b07024f2a6a64126fdc40620760fa497397b06a993e42402c
+size 1035731

examples/export_mesh_with_superfit.py

@@ -0,0 +1,132 @@
+#!/usr/bin/env python3
+"""Example: export a reconstructed mesh from a released primitive assembly.
+
+This example requires the SuperFit runtime:
+
+    https://github.com/BardOfCodes/superfit
+
+Run from the dataset repository root after installing SuperFit and its runtime
+dependencies:
+
+    python examples/export_mesh_with_superfit.py \
+        --source toys4k \
+        --method superfrustum \
+        --object-id airplane_002 \
+        --output airplane_002.obj
+
+The mesh path follows SuperFit's own export flow: load the released stats dict,
+recover the saved primitive expression, evaluate it on a 3D grid, then convert
+the SDF to a mesh with SuperFit's ``sdf_to_mesh`` helper.
+"""
+
+from __future__ import annotations
+
+import argparse
+import pickle
+import sys
+from pathlib import Path
+from typing import Any
+
+# Allow `import load_release` when executed as `python examples/export_mesh_with_superfit.py`
+_REPO_ROOT = Path(__file__).resolve().parent.parent
+if str(_REPO_ROOT) not in sys.path:
+    sys.path.insert(0, str(_REPO_ROOT))
+
+from load_release import ReleaseIndex  # noqa: E402
+
+
+def _import_superfit() -> tuple[Any, Any, Any, Any, Any]:
+    try:
+        from geolipi.torch_compute import Sketcher, recursive_evaluate
+        from superfit.symbolic.utils import fetch_singular_expr_eval
+        from superfit.utils.io import get_best_expr
+        from superfit.utils.mesh_sdf import sdf_to_mesh
+    except ImportError as exc:
+        raise SystemExit(
+            "This example requires SuperFit and its runtime dependencies. "
+            "Install or add SuperFit to PYTHONPATH; see "
+            "https://github.com/BardOfCodes/superfit"
+        ) from exc
+    return Sketcher, recursive_evaluate, fetch_singular_expr_eval, get_best_expr, sdf_to_mesh
+
+
+def load_expression(
+    assembly_path: Path,
+    *,
+    program: str,
+    iter_idx: int | None,
+    temperature: float,
+    device: str,
+) -> Any:
+    _, _, fetch_singular_expr_eval, get_best_expr, _ = _import_superfit()
+    with assembly_path.open("rb") as fh:
+        info = pickle.load(fh)
+
+    expr = get_best_expr(info, iter_idx=iter_idx, prog_type=program)
+    return fetch_singular_expr_eval(
+        expr.tensor(device=device),
+        temperature=temperature,
+        relaxed_eval=True,
+        remove_marker=True,
+        device=device,
+    )
+
+
+def export_mesh(expr: Any, output_path: Path, *, resolution: int, device: str) -> None:
+    Sketcher, recursive_evaluate, _, _, sdf_to_mesh = _import_superfit()
+    sketcher = Sketcher(resolution=resolution, n_dims=3, device=device)
+    sdf = recursive_evaluate(expr.tensor(device=device), sketcher)
+    mesh = sdf_to_mesh(sdf, sketcher)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    mesh.export(output_path)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--root", type=Path, default=_REPO_ROOT, help="Dataset repo root.")
+    parser.add_argument("--source", default="toys4k", choices=["toys4k", "partobjaverse"])
+    parser.add_argument("--method", default="superfrustum")
+    parser.add_argument("--object-id", required=True, help="Instance id to export.")
+    parser.add_argument(
+        "--program",
+        default="pruned_program",
+        help="Saved program key to export, e.g. pruned_program or best_program.",
+    )
+    parser.add_argument(
+        "--iter-idx",
+        type=int,
+        default=None,
+        help="Optimization iteration to read. Default: final iteration recorded in the artifact.",
+    )
+    parser.add_argument("--resolution", type=int, default=128, help="SDF grid resolution.")
+    parser.add_argument(
+        "--temperature",
+        type=float,
+        default=10000.0,
+        help="Temperature used when resolving stochastic primitive expressions.",
+    )
+    parser.add_argument(
+        "--device",
+        default="cuda",
+        help="Torch device for expression evaluation. SuperFit mesh export typically requires CUDA.",
+    )
+    parser.add_argument("--output", type=Path, required=True, help="Output mesh path, e.g. out.obj.")
+    args = parser.parse_args()
+
+    index = ReleaseIndex(args.root)
+    row = index.get(args.source, args.method, args.object_id)
+    assembly_path = index.artifact_path(row, "primitive_assembly")
+
+    expr = load_expression(
+        assembly_path,
+        program=args.program,
+        iter_idx=args.iter_idx,
+        temperature=args.temperature,
+        device=args.device,
+    )
+    export_mesh(expr, args.output, resolution=args.resolution, device=args.device)
+    print(f"Exported {args.output}")
+
+
+if __name__ == "__main__":
+    main()

examples/load_artifact.py

@@ -7,7 +7,11 @@ Run from the dataset repository root::
     python examples/load_artifact.py --source toys4k --method superfrustum --object-id airplane_002
     python examples/load_artifact.py --artifact eval
 
-Requires only the Python standard library plus ``load_release.py`` at the repo root.
+Manifest discovery requires only the Python standard library plus
+``load_release.py`` at the repo root. Actually unpickling primitive assemblies
+may require SuperFit, PyTorch, NumPy, and related runtime dependencies. See:
+
+    https://github.com/BardOfCodes/superfit
 """
 
 from __future__ import annotations

load_release.py

@@ -1,17 +1,20 @@
 """Lightweight loader for the SuperFit primitive-assembly release.
 
 Reads ``manifest.jsonl`` at the repository root and provides helpers to filter
-rows and load pickle artifacts. No dependency on PyTorch or the SuperFit package.
+rows and resolve artifact paths. Manifest discovery has no dependency on
+PyTorch or the SuperFit package. Loading pickle artifacts may require SuperFit,
+PyTorch, NumPy, and related runtime dependencies because serialized tensors and
+primitive-expression objects can be embedded in the pickle payloads.
 
 Example::
 
     from pathlib import Path
-    from load_release import ReleaseIndex, load_pickle
+    from load_release import ReleaseIndex
 
     root = Path(".")  # clone root of the Hugging Face dataset repo
     index = ReleaseIndex(root)
     rows = index.filter(source_dataset="toys4k", method="superfrustum", category="airplane")
-    stats = load_pickle(root / rows[0]["primitive_assembly_path"])
+    print(index.artifact_path(rows[0], "primitive_assembly"))
 """
 
 from __future__ import annotations
@@ -121,7 +124,10 @@ class ReleaseIndex:
 
 
 def load_pickle(path: str | Path) -> Any:
-    """Load a pickle file. Only unpickle artifacts you trust."""
+    """Load a pickle file. Only unpickle artifacts you trust.
+
+    Primitive-assembly pickles may require the SuperFit runtime to be installed.
+    """
     with Path(path).open("rb") as fh:
         return pickle.load(fh)