{"id":49062121,"url":"https://github.com/stability-ai/reli3d","last_synced_at":"2026-04-20T03:00:58.203Z","repository":{"id":351868590,"uuid":"1180819497","full_name":"Stability-AI/ReLi3D","owner":"Stability-AI","description":"Relightable multi-view 3D reconstruction with disentangled illumination and spatially varying PBR materials","archived":false,"fork":false,"pushed_at":"2026-04-16T19:53:51.000Z","size":14930,"stargazers_count":54,"open_issues_count":3,"forks_count":3,"subscribers_count":3,"default_branch":"main","last_synced_at":"2026-04-16T21:34:03.395Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Stability-AI.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE.md","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-03-13T12:54:44.000Z","updated_at":"2026-04-16T19:53:55.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/Stability-AI/ReLi3D","commit_stats":null,"previous_names":["stability-ai/reli3d"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/Stability-AI/ReLi3D","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Stability-AI%2FReLi3D","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Stability-AI%2FReLi3D/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Stability-AI%2FReLi3D/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Stability-AI%2FReLi3D/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Stability-AI","download_url":"https://codeload.github.com/Stability-AI/ReLi3D/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Stability-AI%2FReLi3D/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32031070,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-20T00:18:06.643Z","status":"online","status_checked_at":"2026-04-20T02:00:06.527Z","response_time":94,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-04-20T03:00:57.317Z","updated_at":"2026-04-20T03:00:58.198Z","avatar_url":"https://github.com/Stability-AI.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# ReLi3D\n\n\u003cp align=\"left\"\u003e\n  \u003cstrong\u003e\n    Relightable Large Reconstruction from Multi-View Images\n  \u003c/strong\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"media/reli3d_teaser.jpg\" alt=\"ReLi3D Teaser\" width=\"100%\" /\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003cspan\u003e\u003ca href=\"https://arxiv.org/abs/2603.19753\"\u003ePaper\u003c/a\u003e\u003c/span\u003e\u0026nbsp;\u0026nbsp;\u0026nbsp;\n  \u003cspan\u003e\u003ca href=\"https://reli3d.jdihlmann.com/\"\u003eProject Page\u003c/a\u003e\u003c/span\u003e\u0026nbsp;\u0026nbsp;\u0026nbsp;\n  \u003cspan\u003e\u003ca href=\"https://huggingface.co/StabilityLabs/ReLi3D\"\u003eModel (Hugging Face)\u003c/a\u003e\u003c/span\u003e\u0026nbsp;\u0026nbsp;\u0026nbsp;\n  \u003cspan\u003e\u003ca href=\"https://github.com/Stability-AI/ReLi3D\"\u003eCode\u003c/a\u003e\u003c/span\u003e\n\u003c/p\u003e\n\n## About\n\nReLi3D is an inference-only release for relightable 3D asset reconstruction.\n\nGiven a set of object images and camera poses (`transforms.json` + RGBA frames), ReLi3D predicts a UV-unwrapped textured mesh with material attributes and estimated illumination.\n\nTraining and experiment pipelines are intentionally excluded from this repository.\n\n## Repository Layout\n\n- `demos/reli3d/infer_from_transforms.py`: Main inference entrypoint.\n- `configs/reli3d/inference.yaml`: Runtime system config.\n- `scripts/download_model_from_hf.py`: Explicit model artifact downloader.\n- `src/`: Runtime model components used for inference.\n- `native/`: Native extensions (`uv_unwrapper`, `texture_baker`).\n- `artifacts/model/`: Local model artifacts (`config.yaml`, `reli3d_final.ckpt`).\n\n## Installation\n\n```bash\ngit clone https://github.com/Stability-AI/ReLi3D.git\ncd ReLi3D\n\npython3.10 -m venv .venv\nsource .venv/bin/activate\n\npip install --upgrade pip\npip install -r requirements.txt\npip install ./native/uv_unwrapper ./native/texture_baker\n```\n\n## Model Artifacts\n\nDefault model source: `StabilityLabs/ReLi3D` on Hugging Face.\n\nIf the model repo is private or gated, authenticate first:\n\n```bash\nhuggingface-cli login\ngit config --global credential.helper store\n```\n\nDownload artifacts explicitly:\n\n```bash\npython scripts/download_model_from_hf.py \\\n  --repo-id StabilityLabs/ReLi3D \\\n  --output-dir artifacts/model\n```\n\nSelective download:\n\n```bash\n# Only config\npython scripts/download_model_from_hf.py --skip-checkpoint\n\n# Only checkpoint\npython scripts/download_model_from_hf.py --skip-config\n```\n\n`demos/reli3d/infer_from_transforms.py` also auto-downloads missing default artifacts during inference.\n\nResolution order for checkpoint:\n\n1. `--checkpoint`\n2. `RELI3D_CHECKPOINT`\n3. `artifacts/model/reli3d_final.ckpt`\n\nDefault config path:\n\n- `artifacts/model/config.yaml` (falls back to `artifacts/model/raw.yaml` if present)\n\n## Input Format\n\nEach object under `--input-root` must contain:\n\n- `transforms.json`\n- `rgba/\u003cview\u003e.png`\n\nExample:\n\n```text\ndemo_files/objects/\n  Camera_01/\n    transforms.json\n    rgba/\n      0000.png\n      0010.png\n      0021.png\n      0031.png\n```\n\nExpected `transforms.json` frame keys:\n\n- `file_path`\n- `transform_matrix` (or `camera_transform`)\n- `camera_fov`\n- optional `camera_principal_point`\n\n## Quickstart\n\n```bash\npython demos/reli3d/infer_from_transforms.py \\\n  --input-root demo_files/objects \\\n  --objects Camera_01 \\\n  --output-root outputs \\\n  --num-views 4 \\\n  --texture-size 256 \\\n  --overwrite\n```\n\n## Exact Final-Run Parity\n\nIf you want strict parity with your exact final run setup, pass explicit artifact paths:\n\n```bash\npython demos/reli3d/infer_from_transforms.py \\\n  --input-root demo_files/objects \\\n  --objects Camera_01 \\\n  --config /path/to/raw.yaml \\\n  --checkpoint /path/to/epoch=0-step=80000.ckpt \\\n  --output-root outputs \\\n  --overwrite\n```\n\n## Outputs\n\nPer object output directory:\n\n- `outputs/\u003cobject\u003e/mesh.glb`\n- `outputs/\u003cobject\u003e/illumination.hdr` (if predicted)\n- `outputs/\u003cobject\u003e/run_info.json`\n\n`run_info.json` stores repo-relative or filename-only paths to avoid leaking internal absolute filesystem paths.\n\n## License\n\nThis code and model usage are subject to Stability AI Community License terms.\n\nFor individuals or organizations generating annual revenue of USD 1,000,000 (or local currency equivalent) or more, commercial usage requires an enterprise license from Stability AI.\n\n- License details: https://stability.ai/license\n- Enterprise request: https://stability.ai/enterprise\n\n## Citation\n\n```\n@inproceeding{ dihlmann2026reli3d,\nauthor = {Dihlmann, Jan-Niklas and Boss, Mark and Donne, Simon and Engelhardt, Andreas and Lensch, Hendrik P. A. and Jampani, Varun},\ntitle = {ReLi3D: Relightable Multi-view 3D Reconstruction with Disentangled Illumination},\nbooktitle = {International Conference on Learning Representations (ICLR)},\nyear ={2026}\n}\n```\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstability-ai%2Freli3d","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fstability-ai%2Freli3d","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstability-ai%2Freli3d/lists"}