Paper: <a href=\"https://arxiv.org/abs/2608.14783\" rel=\"nofollow\">https://arxiv.org/abs/2608.14783</a><br>Github: <a href=\"https://github.com/InternRobotics/MeshCoder\" rel=\"nofollow\">https://github.com/InternRobotics/MeshCoder</a><br>Project Page: <a href=\"https://expmaster.github.io/megaparts_webpage/\" rel=\"nofollow\">https://expmaster.github.io/megaparts_webpage/</a></p>\n","updatedAt":"2026-08-18T02:00:59.887Z","author":{"_id":"63f2ec797ddf724fbcc75aee","avatarUrl":"/avatars/e93432ad11da703d46fe5e594d69f8c0.svg","fullname":"Zhaoyang Lyu","name":"ZhaoyangLyu","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":7,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.47819140553474426},"editors":["ZhaoyangLyu"],"editorAvatarUrls":["/avatars/e93432ad11da703d46fe5e594d69f8c0.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.14783","authors":[{"_id":"6a83bb3e675db694db8cd44c","name":"Manwen Liao","hidden":false},{"_id":"6a83bb3e675db694db8cd44d","name":"Xinyu Lian","hidden":false},{"_id":"6a83bb3e675db694db8cd44e","name":"Jian Mao","hidden":false},{"_id":"6a83bb3e675db694db8cd44f","name":"Kaixu Chen","hidden":false},{"_id":"6a83bb3e675db694db8cd450","name":"Li Luo","hidden":false},{"_id":"6a83bb3e675db694db8cd451","name":"Jinghao Yan","hidden":false},{"_id":"6a83bb3e675db694db8cd452","name":"Wanshui Gan","hidden":false},{"_id":"6a83bb3e675db694db8cd453","name":"Qiao Yu","hidden":false},{"_id":"6a83bb3e675db694db8cd454","name":"Weitian Zhang","hidden":false},{"_id":"6a83bb3e675db694db8cd455","name":"Chunhua Shen","hidden":false},{"_id":"6a83bb3e675db694db8cd456","name":"Guang Chen","hidden":false},{"_id":"6a83bb3e675db694db8cd457","name":"Bo Dai","hidden":false},{"_id":"6a83bb3e675db694db8cd458","name":"Xudong Xu","hidden":false},{"_id":"6a83bb3e675db694db8cd459","name":"Zhaoyang Lyu","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/63f2ec797ddf724fbcc75aee/3fgKAki-IWjFAoVqR6cym.mp4"],"publishedAt":"2026-08-14T00:00:00.000Z","submittedOnDailyAt":"2026-08-18T00:00:00.000Z","title":"MegaParts: Scaling Part-Aware 3D Object Generation to 300 Parts via Token-Efficient Autoregressive Modeling","submittedOnDailyBy":{"_id":"63f2ec797ddf724fbcc75aee","avatarUrl":"/avatars/e93432ad11da703d46fe5e594d69f8c0.svg","isPro":false,"fullname":"Zhaoyang Lyu","user":"ZhaoyangLyu","type":"user","name":"ZhaoyangLyu"},"summary":"Part-aware 3D object generation is essential for graphics applications such as controllable modeling, editing, and articulation, where objects are represented as coherent assemblies of semantic parts. However, existing part-aware generation methods, do not scale well to highly complex objects. As the number of parts increases, generating detailed geometry becomes prohibitively expensive in token length and memory. We introduce MegaParts, a scalable autoregressive 3D generation framework to address this challenge by combining structured sequence modeling with a token-efficient vector-quantized shape tokenizer. Our tokenizer learns discrete latent representations for part-level geometry by minimizing token usage subject to high-fidelity reconstruction, enabling adaptive-length tokenization based on geometric complexity. On top of this compact representation, we train a large language model to generate object bounding boxes, part bounding boxes, and part shape tokens within a unified structured sequence. Combined with efficient long-context training strategy, our token-efficient formulation scales to objects with up to 300 parts and sequence lengths up to 256k tokens. This substantially extends the scale of part-aware 3D generation while preserving compositional structure and enabling fine-grained part-level control. Our method achieves higher mesh quality than baseline autoregressive and diffusion models, showing that compressed discrete part tokens improve not only scalability but also the achievable fidelity of generated geometry. These results suggest that LLM native token-efficient autoregressive modeling is a compelling alternative to diffusion for large-scale part-aware 3D generation. The project page is available at https://expmaster.github.io/megaparts_webpage.","upvotes":13,"discussionId":"6a83bb3e675db694db8cd45a","projectPage":"https://expmaster.github.io/megaparts_webpage/","githubRepo":"https://github.com/InternRobotics/MeshCoder","githubRepoAddedBy":"user","ai_summary":"MegaParts scales part-aware 3D generation via token-efficient vector-quantized part tokens and structured autoregressive sequence modeling with long-context training.","ai_keywords":["autoregressive 3D generation","part-aware generation","vector-quantized shape tokenizer","discrete latent representations","adaptive-length tokenization","structured sequence modeling","long-context training","large language model","compositional structure","part-level control"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":504,"organization":{"_id":"6a4fb75a1c66dbf208e7ddb6","name":"Shanghai-AI-Laboratory","fullname":"Shanghai AI Laboratory","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65cd955637be1841d0b75397/Rao_Kq6NMtTVfSqLUIR4k.webp"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"63f2ec797ddf724fbcc75aee","avatarUrl":"/avatars/e93432ad11da703d46fe5e594d69f8c0.svg","isPro":false,"fullname":"Zhaoyang Lyu","user":"ZhaoyangLyu","type":"user"},{"_id":"6775159d8cde42405c93a560","avatarUrl":"/avatars/d859ad2c10f330b2e02d0f8e56d78760.svg","isPro":false,"fullname":"yinoqifu00","user":"yinoqifu00","type":"user"},{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user"},{"_id":"64c3333b1920bd982088fa8a","avatarUrl":"/avatars/7644abda3d778fa29c1b1d3f8ed5e27c.svg","isPro":false,"fullname":"Jian Mao","user":"SteveJMao","type":"user"},{"_id":"66b1dce83158b1fadffb0d9d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/RbWKIxCjawVaSidwZLZA1.png","isPro":false,"fullname":"Kaixu Chen","user":"kaixuChen","type":"user"},{"_id":"66120ce7db728ea967a38e8e","avatarUrl":"/avatars/c3b0411b13d4e30b9c2fc4f82db60c48.svg","isPro":false,"fullname":"Gan","user":"Wanshui","type":"user"},{"_id":"6a021773b9392f44be7afd3c","avatarUrl":"/avatars/41f4faf73687a960ba126a7d62fa4a4b.svg","isPro":false,"fullname":"yinoqifu000","user":"yinoqifu-00000000","type":"user"},{"_id":"66432376b2604b818043d176","avatarUrl":"/avatars/9f50f7766e8208c8b567790ebfc1b28f.svg","isPro":false,"fullname":"Qiao Yu","user":"YuQiao0303","type":"user"},{"_id":"652e39fea3da41257d6ec253","avatarUrl":"/avatars/fc00bd8bf3704afb269b45b5bf7fd1b9.svg","isPro":false,"fullname":"Li Luo","user":"ReyLock","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"6949ba7ce42374aa7cd7d512","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/gc1cY23AM4rCiaVsFODKO.png","isPro":true,"fullname":"Wiktor Matuszek","user":"WiktorMatuszek","type":"user"},{"_id":"691fe05ed9ea9e63971b2304","avatarUrl":"/avatars/bf138de9f1aaeae802441f346a3f34ac.svg","isPro":false,"fullname":"YxW","user":"Sarsa123","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6a4fb75a1c66dbf208e7ddb6","name":"Shanghai-AI-Laboratory","fullname":"Shanghai AI Laboratory","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65cd955637be1841d0b75397/Rao_Kq6NMtTVfSqLUIR4k.webp"},"query":{}}">

Cite arxiv.org/abs/2608.14783 in a model README.md to link it from this page.

Cite arxiv.org/abs/2608.14783 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2608.14783 in a Space README.md to link it from this page.