[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72164":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":13,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":9,"pushedAt":9,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":28,"lastSyncTime":29,"discoverSource":30},72164,"InstantMesh","TencentARC\u002FInstantMesh","TencentARC","InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models",null,"Python",4414,491,46,115,0,4,16,12,30.08,"Apache License 2.0",false,"main",true,[],"2026-06-12 02:02:59","\u003Cdiv align=\"center\">\n  \n# InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models\n\n\u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.07191\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FArXiv-2404.07191-brightgreen\">\u003C\u002Fa> \n\u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002FTencentARC\u002FInstantMesh\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Model_Card-Huggingface-orange\">\u003C\u002Fa> \n\u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTencentARC\u002FInstantMesh\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Gradio%20Demo-Huggingface-orange\">\u003C\u002Fa> \u003Cbr>\n\u003Ca href=\"https:\u002F\u002Freplicate.com\u002Fcamenduru\u002Finstantmesh\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDemo-Replicate-blue\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fcamenduru\u002FInstantMesh-jupyter\u002Fblob\u002Fmain\u002FInstantMesh_jupyter.ipynb\">\u003Cimg src=\"https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fjtydhr88\u002FComfyUI-InstantMesh\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDemo-ComfyUI-8A2BE2\">\u003C\u002Fa>\n\n\u003C\u002Fdiv>\n\n---\n\nThis repo is the official implementation of InstantMesh, a feed-forward framework for efficient 3D mesh generation from a single image based on the LRM\u002FInstant3D architecture.\n\nhttps:\u002F\u002Fgithub.com\u002FTencentARC\u002FInstantMesh\u002Fassets\u002F20635237\u002Fdab3511e-e7c6-4c0b-bab7-15772045c47d\n\n# 🚩 Features and Todo List\n- [x] 🔥🔥 Release Zero123++ fine-tuning code. \n- [x] 🔥🔥 Support for running gradio demo on two GPUs to save memory.\n- [x] 🔥🔥 Support for running demo with docker. Please refer to the [docker](docker\u002F) directory.\n- [x] Release inference and training code.\n- [x] Release model weights.\n- [x] Release huggingface gradio demo. Please try it at [demo](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002FTencentARC\u002FInstantMesh) link.\n- [ ] Add support for more multi-view diffusion models.\n\n# ⚙️ Dependencies and Installation\n\nWe recommend using `Python>=3.10`, `PyTorch>=2.1.0`, and `CUDA>=12.1`.\n```bash\nconda create --name instantmesh python=3.10\nconda activate instantmesh\npip install -U pip\n\n# Ensure Ninja is installed\nconda install Ninja\n\n# Install the correct version of CUDA\nconda install cuda -c nvidia\u002Flabel\u002Fcuda-12.1.0\n\n# Install PyTorch and xformers\n# You may need to install another xformers version if you use a different PyTorch version\npip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu121\npip install xformers==0.0.22.post7\n\n# Install other requirements\npip install -r requirements.txt\n```\n\n# 💫 How to Use\n\n## Download the models\n\nWe provide 4 sparse-view reconstruction model variants and a customized Zero123++ UNet for white-background image generation in the [model card](https:\u002F\u002Fhuggingface.co\u002FTencentARC\u002FInstantMesh).\n\nOur inference script will download the models automatically. Alternatively, you can manually download the models and put them under the `ckpts\u002F` directory.\n\nBy default, we use the `instant-mesh-large` reconstruction model variant.\n\n## Start a local gradio demo\n\nTo start a gradio demo in your local machine, simply run:\n```bash\npython app.py\n```\n\nIf you have multiple GPUs in your machine, the demo app will run on two GPUs automatically to save memory. You can also force it to run on a single GPU:\n```bash\nCUDA_VISIBLE_DEVICES=0 python app.py\n```\n\nAlternatively, you can run the demo with docker. Please follow the instructions in the [docker](docker\u002F) directory.\n\n## Running with command line\n\nTo generate 3D meshes from images via command line, simply run:\n```bash\npython run.py configs\u002Finstant-mesh-large.yaml examples\u002Fhatsune_miku.png --save_video\n```\n\nWe use [rembg](https:\u002F\u002Fgithub.com\u002Fdanielgatis\u002Frembg) to segment the foreground object. If the input image already has an alpha mask, please specify the `no_rembg` flag:\n```bash\npython run.py configs\u002Finstant-mesh-large.yaml examples\u002Fhatsune_miku.png --save_video --no_rembg\n```\n\nBy default, our script exports a `.obj` mesh with vertex colors, please specify the `--export_texmap` flag if you hope to export a mesh with a texture map instead (this will cost longer time):\n```bash\npython run.py configs\u002Finstant-mesh-large.yaml examples\u002Fhatsune_miku.png --save_video --export_texmap\n```\n\nPlease use a different `.yaml` config file in the [configs](.\u002Fconfigs) directory if you hope to use other reconstruction model variants. For example, using the `instant-nerf-large` model for generation:\n```bash\npython run.py configs\u002Finstant-nerf-large.yaml examples\u002Fhatsune_miku.png --save_video\n```\n**Note:** When using the `NeRF` model variants for image-to-3D generation, exporting a mesh with texture map by specifying `--export_texmap` may cost long time in the UV unwarping step since the default iso-surface extraction resolution is `256`. You can set a lower iso-surface extraction resolution in the config file.\n\n# 💻 Training\n\nWe provide our training code to facilitate future research. But we cannot provide the training dataset due to its size. Please refer to our [dataloader](src\u002Fdata\u002Fobjaverse.py) for more details.\n\nTo train the sparse-view reconstruction models, please run:\n```bash\n# Training on NeRF representation\npython train.py --base configs\u002Finstant-nerf-large-train.yaml --gpus 0,1,2,3,4,5,6,7 --num_nodes 1\n\n# Training on Mesh representation\npython train.py --base configs\u002Finstant-mesh-large-train.yaml --gpus 0,1,2,3,4,5,6,7 --num_nodes 1\n```\n\nWe also provide our Zero123++ fine-tuning code since it is frequently requested. The running command is:\n```bash\npython train.py --base configs\u002Fzero123plus-finetune.yaml --gpus 0,1,2,3,4,5,6,7 --num_nodes 1\n```\n\n# :books: Citation\n\nIf you find our work useful for your research or applications, please cite using this BibTeX:\n\n```BibTeX\n@article{xu2024instantmesh,\n  title={InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models},\n  author={Xu, Jiale and Cheng, Weihao and Gao, Yiming and Wang, Xintao and Gao, Shenghua and Shan, Ying},\n  journal={arXiv preprint arXiv:2404.07191},\n  year={2024}\n}\n```\n\n# 🤗 Acknowledgements\n\nWe thank the authors of the following projects for their excellent contributions to 3D generative AI!\n\n- [Zero123++](https:\u002F\u002Fgithub.com\u002FSUDO-AI-3D\u002Fzero123plus)\n- [OpenLRM](https:\u002F\u002Fgithub.com\u002F3DTopia\u002FOpenLRM)\n- [FlexiCubes](https:\u002F\u002Fgithub.com\u002Fnv-tlabs\u002FFlexiCubes)\n- [Instant3D](https:\u002F\u002Finstant-3d.github.io\u002F)\n\nThank [@camenduru](https:\u002F\u002Fgithub.com\u002Fcamenduru) for implementing [Replicate Demo](https:\u002F\u002Freplicate.com\u002Fcamenduru\u002Finstantmesh) and [Colab Demo](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fcamenduru\u002FInstantMesh-jupyter\u002Fblob\u002Fmain\u002FInstantMesh_jupyter.ipynb)!  \nThank [@jtydhr88](https:\u002F\u002Fgithub.com\u002Fjtydhr88) for implementing [ComfyUI support](https:\u002F\u002Fgithub.com\u002Fjtydhr88\u002FComfyUI-InstantMesh)!\n","InstantMesh 是一个基于单张图像生成3D网格的高效框架，利用稀疏视图的大规模重建模型。其核心功能包括从单一输入图像快速生成高质量3D网格，并支持多GPU运行以节省内存，同时提供了Gradio演示、Docker部署选项以及HuggingFace上的模型卡。项目采用Python开发，依赖PyTorch等深度学习库，适用于需要快速将2D图像转换为3D模型的应用场景，如虚拟现实内容创建、游戏资产生成或3D打印预览等领域。",2,"2026-06-11 03:40:40","high_star"]