Infinite Mobility: Scalable High-Fidelity Synthesis of Articulated Objects via Procedural Generation
Getting Started | Contributing | Join Us
This repo is about the procedural generation of articulated objects like those in this scene.
- Initial release of the code
- Make installation more self-contained
- Add usd format support
- Generate interactive environment with infinigen layout solvers.
First, follow installation instruction of Infinigen to setup the basic environment Installation Instructions.
important: we are based on an earlier version of Infinigen.(git 572bfe7)
It could break the pipeline if the newest version is installed
Then, please run setup.py to configure our dependencies.
python setup.py
Next, download our part dataset here Finally, configure dataset path in python code line 30. For example, if part dataset is /datasets/parts
AUXILIARY_PATH = "/datasets/parts"
It should work just fine now!🎊
We provide a script for you to generate as many articulated objects as you like!
python paralled_generate.py <Factory Name> <Number> <MaxProcess>
If you want to generate 100 officechairs, and generate 10 samples at the same time, run
python paralled_generate.py OfficeChairFactory 100 10
Results will be in outputs folder in the form of URDF.
For Factory Name we support now, refer to the form below.
python show.py <path to dir or urdf file>
If you want to visualize one urdf file in ./outputs/***.urdf, run
python show.py ./outputs/***.urdf
We also support visualize all urdfs in one dir.If you want to visualize all urdfs in ./output, run
python show.py ./outputs
We welcome contributions! You can contribute in many ways:
- Contribute code to Infinigen repository - Procedural generators for more categories with interactive parts are still needed, we are happy to update our articulation modifications as Infinigen evolves!
- Contribute more diverse parts with fine geometry - Compared to original infinigen, our work introduces substitution of certain parts with collected meshes. A more diversed parts dataset would boost our performance in many applications!
We are seeking engineers, interns, researchers, and PhD candidates. If you have an interest in 3D content generation, please send your resume to [email protected].
If you are having difficulties running scripts in our repo, please open issues! We are happy to provide supports and have discussions!