[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-79040":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":19,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":22,"hasPages":22,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":28,"readmeContent":29,"aiSummary":30,"trendingCount":15,"starSnapshotCount":15,"syncStatus":31,"lastSyncTime":32,"discoverSource":33},79040,"PhysX-Omni","physx-omni\u002FPhysX-Omni","physx-omni","PhysX-Omni: Unified Simulation-Ready Physical 3D Generation for Rigid, Deformable, and Articulated Objects","https:\u002F\u002Fphysx-omni.github.io\u002F",null,"Jupyter Notebook",234,10,1,0,3,19,107,12,3.12,"Other",false,"main",[25,26,27],"3d","image-to-3d","physical-modeling","2026-06-12 02:03:49","\u003Cdiv align=\"left\">\n\u003Ch1 align=\"center\">PhysX-Omni: Unified Simulation-Ready Physical 3D Generation\nfor Rigid, Deformable, and Articulated Objects\n\u003C\u002Fh1>\n\u003Cp align=\"center\">\n\u003Ca href='https:\u002F\u002Fphysx-omni.github.io\u002F'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FProject_Page-Website-green?logo=homepage&logoColor=white' alt='Project Page'>\u003C\u002Fa>\n\u003Ca href='https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FPhysX-Omni\u002FPhysXVerse'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Dataset-blue'>\u003C\u002Fa>\n\u003Ca href='https:\u002F\u002Fyoutu.be\u002FZCgj4ffz4yk'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fyoutube\u002Fviews\u002FZCgj4ffz4yk'>\u003C\u002Fa>\n\u003Cdiv>\n\u003Cdiv style=\"width: 100%; text-align: center; margin:auto;\">\n    \u003Cimg style=\"width:100%\" src=\"img\u002Fteaser.png\">\n\u003C\u002Fdiv>\n\n\n\n\n\n## 🏆 News\n\n- We release the code of PhysX-Omni, PhysXVerse, and  PhysX-Bench 🎉\n\n## I. PhysX-Omni\n\n### Installation\n\n1. Clone the repo:\n\n```\ngit clone --recurse-submodules https:\u002F\u002Fgithub.com\u002Fphysx-omni\u002FPhysX-Omni.git\ncd PhysX-Omni \n```\n\n2. Create a new conda environment named `physx-anything` and install the dependencies:\n\n```bash\n. .\u002Fsetup.sh --new-env --basic --xformers --flash-attn --diffoctreerast --spconv --mipgaussian --kaolin --nvdiffrast\n```\n\n**Note**: The detailed usage of `setup.sh` can be found at [TRELLIS](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FTRELLIS)\n\n3. Install the dependencies for Qwen2.5:\n\n```bash\npip install transformers==4.50.0\npip install qwen-vl-utils\npip install 'accelerate>=0.26.0'\n```\n\n**Note**: We release the `requirements.txt` file. You can install all dependencies by running:\n\n```bash\nconda create -n physx-omni python=3.10\nconda activate physx-omni\npip install -r requirements.txt\n```\n\n### Training\n\n1. Download PhysX datasets from [PhysXNet](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FCaoza\u002FPhysX-3D), [PhysX-Mobility](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FCaoza\u002FPhysX-Mobility), and [PhysXVerse](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FPhysX-Omni\u002FPhysXVerse)\n\n2. Run the preprocessing script for PhysXVerse. \n\n   ```python\n   cd dataset\n   python 1voxel_verse.py\n   python 2encode_representation_64_finetune\n   python 3generate_data_new_64_finetune_rle.py\n   ```\n\n   **Note**: Here is a template for you to check the format: [template](https:\u002F\u002Fgithub.com\u002Fziangcao0312\u002FPhysX-Anything\u002Fblob\u002Fmain\u002Fdataset\u002Ftraining_data_template.json).\n\n   **Note**: Preprocess the PhysXNet and PhysX-Mobility follows [PhysX-Anything](https:\u002F\u002Fgithub.com\u002Fziangcao0312\u002FPhysX-Anything)\n\n3. Render the conditioning images (25 images per object) based on your requirements. \n\n   For PhysX-Mobility and PhysXVerse, we use [dataset_toolkits\u002Frender_cond_mobility.py](https:\u002F\u002Fgithub.com\u002Fziangcao0312\u002FPhysX-Anything\u002Ftree\u002Fmain\u002Fdataset_toolits) to generate the conditioning images. \n\n   For PhysXNet, please check [PhysX-3D\u002Fdataset_toolkits\u002Fprecess.sh](https:\u002F\u002Fgithub.com\u002Fziangcao0312\u002FPhysX-3D\u002Fblob\u002Fmain\u002Fdataset_toolkits\u002Fprecess.sh)\n\n4. Set the path in train [configuration](https:\u002F\u002Fgithub.com\u002Fziangcao0312\u002FPhysX-Omni\u002Fblob\u002Fmain\u002Fqwen-vl-finetune\u002Fqwenvl\u002Fdata\u002F__init__.py)\n\n   ```python\n   PHYSXNET = {\n       \"annotation_path\": \"xx\", #json file path\n       \"data_path\": \"xx\",  # conditioning image path\n   }\n   \n   PHYSXMOBILITY = {\n       \"annotation_path\": \"xx\", #json file path\n       \"data_path\": \"xx\",  # conditioning image path\n   }\n   \n   PHYSXVERSE = {\n       \"annotation_path\": \"xx\", #json file path\n       \"data_path\": \"xx\",  # conditioning image path\n   }\n   ```\n\n5. Finetune the model\n\n   ```\n   cd qwen-vl-finetune\n   sbatch scripts\u002Ftrain_physx.sh\n   ```\n\n### Inference\n\n1. Download the pre-train model from [huggingface](https:\u002F\u002Fhuggingface.co\u002FPhysX-Omni\u002FPhysX-Omni).\n\n```bash\npython download.py\n```\n\n2. Run the inference code\n\n```bash\npython 1vlm_demo.py            # vlm inference\n    \npython 2infer_geo.py           # decoder inference\n\npython 3jsongen_update.py      # convert to URDF & XML\n```\n\n## II. PhysX-Bench\n\nThis repository includes the PhysX-Omni benchmark code under [`benchmark\u002F`](benchmark\u002F).\n\nSee [`benchmark\u002FREADME.md`](benchmark\u002FREADME.md) for the benchmark file structure, asset generation pipeline, VLM evaluation commands, denominator validation, and aggregation workflow.\n\nFor environment setup, see [`benchmark\u002FINSTALL.md`](benchmark\u002FINSTALL.md).\n\n## III. PhysXVerse\n\nFor more details about our proposed dataset including dataset structure and annotation, please see this  [PhysXVerse](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FPhysX-Omni\u002FPhysXVerse), [PhysX-Mobility](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FCaoza\u002FPhysX-Mobility) and [PhysXNet](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FCaoza\u002FPhysX-3D).\n\n## IV. Other Tools\n\nWe provide `convert_objects2scene.py`, which converts individual objects into a simulation-ready scene. In addition, we build a simple scene generation pipeline in `applications_scene` based on existing works.\n\n### Acknowledgement\n\nThe data and code is based on [PartNet-mobility](https:\u002F\u002Fsapien.ucsd.edu\u002Fbrowse), [Qwen](https:\u002F\u002Fgithub.com\u002FQwenLM\u002FQwen3-VL), [TRELLIS](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FTRELLIS), [Depth-Anything](https:\u002F\u002Fgithub.com\u002FByteDance-Seed\u002Fdepth-anything-3), [Grounded-Segment-Anything](https:\u002F\u002Fgithub.com\u002FIDEA-Research\u002FGrounded-Segment-Anything) and [CAST](https:\u002F\u002Fgithub.com\u002FFishWoWater\u002FCAST). We would like to express our sincere thanks to the contributors.\n\n## :newspaper_roll: License\n\nDistributed under the S-Lab License. See `LICENSE` for more information.\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Finfo.flagcounter.com\u002FCFxN\">\u003Cimg src=\"https:\u002F\u002Fs01.flagcounter.com\u002Fmap\u002FCFxN\u002Fsize_s\u002Ftxt_000000\u002Fborder_CCCCCC\u002Fpageviews_0\u002Fviewers_0\u002Fflags_0\u002F\" alt=\"Flag Counter\" border=\"0\">\u003C\u002Fa>\n\u003C\u002Fdiv>\n","PhysX-Omni 是一个用于生成刚性、可变形和关节物体的统一物理3D模拟准备模型的项目。它支持从2D图像到3D物理模型的转换，具备处理多种类型物体的能力，并且能够生成适合物理仿真使用的3D数据。该项目基于Jupyter Notebook开发，易于上手和扩展。PhysX-Omni 适用于需要进行3D物理建模与仿真的场景，如机器人设计、虚拟现实环境构建以及游戏开发等领域，为开发者提供了强大的工具来创建高度逼真的动态3D内容。",2,"2026-06-11 03:57:25","CREATED_QUERY"]