{"id":27884316,"url":"https://github.com/skyworkai/mvgamba","last_synced_at":"2025-05-05T06:36:18.531Z","repository":{"id":267423617,"uuid":"821200664","full_name":"SkyworkAI/MVGamba","owner":"SkyworkAI","description":"[NeurIPS2024] MVGamba: Unify 3D Content Generation as State Space Sequence Modeling","archived":false,"fork":false,"pushed_at":"2024-12-10T08:42:33.000Z","size":183,"stargazers_count":4,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"master","last_synced_at":"2024-12-10T09:25:34.914Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/SkyworkAI.png","metadata":{"files":{"readme":"readme.md","changelog":null,"contributing":null,"funding":null,"license":null,"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}},"created_at":"2024-06-28T03:10:02.000Z","updated_at":"2024-12-10T08:42:37.000Z","dependencies_parsed_at":"2024-12-15T23:16:40.012Z","dependency_job_id":null,"html_url":"https://github.com/SkyworkAI/MVGamba","commit_stats":null,"previous_names":["skyworkai/mvgamba"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SkyworkAI%2FMVGamba","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SkyworkAI%2FMVGamba/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SkyworkAI%2FMVGamba/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/SkyworkAI%2FMVGamba/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/SkyworkAI","download_url":"https://codeload.github.com/SkyworkAI/MVGamba/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":252454577,"owners_count":21750493,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","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":"2025-05-05T06:36:18.136Z","updated_at":"2025-05-05T06:36:18.517Z","avatar_url":"https://github.com/SkyworkAI.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n## MVGamba: Unify 3D Content Generation as State Space Sequence Modeling (NeurIPS 2024)\n\nThis is the official implementation of **MVGamba**, a novel **Multi-View Gamba Model** designed for **unified 3D content generation**.\n\nhttps://github.com/user-attachments/assets/f71f0349-0e11-4c31-94a7-b7a57275ccb3\n\n\n### [Arxiv](https://arxiv.org/abs/2406.06367) | [Weights]\n\n\n- [x] Release MVGamba-3DGS training and inference code.\n- [x] Release MVGamba-2DGS training and inference code, please refer to the surfel branch.\n- [ ] Release pretrained checkpoints.\n\n### Install\n\n```bash\n# xformers is required! please refer to https://github.com/facebookresearch/xformers for details.\n# for example, we use torch 2.1.0 + cuda 11.8\npip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu118\npip install causal-conv1d==1.2.0 mamba-ssm\ngit clone --recursive git@github.com:SkyworkAI/MVGamba.git\n# a modified 3D gaussian splatting (+ depth, alpha rendering)\npip install ./submodules/diff-gaussian-rasterization\n# 2D gaussian surfel \npip install ./submodules/diff-surfel-rasterization\n\n# for mesh extraction\npip install git+https://github.com/NVlabs/nvdiffrast\n\n# other dependencies\npip install -r requirements.txt\n```\n\n### Training\n\n**NOTE**: Due to company property restrictions, we are unable to provide the full training data used for training MVGamba. Here, we alternatively follow the practice by @kiui(https://github.com/ashawkey) and provide the ~80K subset of **[Objaverse](https://objaverse.allenai.org/objaverse-1.0)** and in **[objaverse_filter](https://github.com/ashawkey/objaverse_filter)**. please check and modify the [dataset](./core/provider_ikun.py) implementation!\n\n```bash\n# training (use single node training)\naccelerate launch --config_file acc_configs/gpu8.yaml main.py mvgamba --workspace /root/Results/workspace_mvgamba\n\n# training (use multi-node training)\nbash ./scripts/mvgamba_dist.sh\n```\n\n### Inference\n\n For [MVDream](https://github.com/bytedance/MVDream) and [ImageDream](https://github.com/bytedance/ImageDream), we use a [diffusers implementation](https://github.com/ashawkey/mvdream_diffusers). The required model weights are downloaded automatically.\n\n```bash \nbash ./scripts/mvgamba_infer.sh\n```\n### Acknowledgement\n\nThis work is built on many amazing research works and open-source projects, thanks a lot to all the authors for sharing!\n\n- [LGM](https://github.com/3DTopia/LGM)\n- [OpenLRM](https://github.com/3DTopia/OpenLRM)\n- [Gamba](https://github.com/SkyworkAI/Gamba)\n- [gaussian-splatting](https://github.com/graphdeco-inria/gaussian-splatting) and [diff-gaussian-rasterization](https://github.com/graphdeco-inria/diff-gaussian-rasterization)\n- [nvdiffrast](https://github.com/NVlabs/nvdiffrast)\n- [dearpygui](https://github.com/hoffstadt/DearPyGui)\n- [tyro](https://github.com/brentyi/tyro)\n\n\n### Citation\n\n```bibtex\n\n@article{yi2024mvgamba,\n  title={MVGamba: Unify 3D Content Generation as State Space Sequence Modeling},\n  author={Yi, Xuanyu and Wu, Zike and Shen, Qiuhong and Xu, Qingshan and Zhou, Pan and Lim, Joo-Hwee and Yan, Shuicheng and Wang, Xinchao and Zhang, Hanwang},\n  journal={arXiv preprint arXiv:2406.06367},\n  year={2024}\n}\n\n```\nPlease also check our another project for ultra fast single view 3D generation [Gamba](https://arxiv.org/abs/2403.18795). The code and pretrained weights have been released in https://github.com/SkyworkAI/Gamba.\n\n\n```bibtex\n\n@article{shen2024gamba,\n  title={Gamba: Marry gaussian splatting with mamba for single view 3d reconstruction},\n  author={Shen, Qiuhong and Wu, Zike and Yi, Xuanyu and Zhou, Pan and Zhang, Hanwang and Yan, Shuicheng and Wang, Xinchao},\n  journal={arXiv preprint arXiv:2403.18795},\n  year={2024}\n}\n\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fskyworkai%2Fmvgamba","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fskyworkai%2Fmvgamba","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fskyworkai%2Fmvgamba/lists"}