Hugging Face Daily Papers · June 10, 2026 · 3 min read

SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

Like Read original ↗

SCAIL-2 provides an end-to-end framework for controlled character animation, eliminating reliance on intermediate pose skeletons and introducing the MotionPair-60K dataset for improved motion transfer performance.</p>\n","updatedAt":"2026-06-10T02:14:57.968Z","author":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","fullname":"taesiri","name":"taesiri","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":314,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8316512703895569},"editors":["taesiri"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.10804","authors":[{"_id":"6a28c857e7d78ea7587e53b5","name":"Wenhao Yan","hidden":false},{"_id":"6a28c857e7d78ea7587e53b6","name":"Fengjia Guo","hidden":false},{"_id":"6a28c857e7d78ea7587e53b7","name":"Zhuoyi Yang","hidden":false},{"_id":"6a28c857e7d78ea7587e53b8","name":"Jie Tang","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6039478ab3ecf716b1a5fd4d/_By1MtbzlZEeJ8U3bHKvW.qt"],"publishedAt":"2026-06-09T00:00:00.000Z","submittedOnDailyAt":"2026-06-10T00:00:00.000Z","title":"SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"Controlled character animation requires transferring motion from a driving sequence to a reference character. Prior works heavily rely on intermediate representations, including pose skeletons to represent motion or masked background to represent environment, which inevitably leads to information loss. To address this, we present SCAIL-2, an framework that bypasses those intermediates and achieves end-to-end character animation. By directly concatenating driving videos to the sequence, the model can obtain all the required visual information from the input video. To address lack of end-to-end data, we unify sub-tasks of character animation with decoupled conditions and then curate a pipeline to synthesize MotionPair-60K, an end-to-end motion transfer dataset containing heterogeneous tasks of character animation. To archive the unification, we utilize in-context mask conditioning and mode-specific RoPE as soft guidance beyond textual instructions and raw visual information. To address synthetic discrepancy in detailed regions, we propose Bias-Aware DPO to construct preference items to mitigate the errors. Extensive experiments demonstrate that our method substantially outperforms existing state-of-the-art approaches in various character animation tasks. A large subset of synthetic data as well as model weights will be released at our project page: https://teal024.github.io/SCAIL-2/.","upvotes":30,"discussionId":"6a28c858e7d78ea7587e53b9","projectPage":"https://teal024.github.io/SCAIL-2/","githubRepo":"https://github.com/zai-org/SCAIL-2","githubRepoAddedBy":"user","ai_summary":"SCAIL-2 enables end-to-end character animation by directly transferring motion from driving videos without intermediate representations, using unified task decomposition and synthetic data generation.","ai_keywords":["end-to-end character animation","motion transfer","driving videos","in-context mask conditioning","mode-specific RoPE","Bias-Aware DPO","MotionPair-60K","decoupled conditions"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":156},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user"},{"_id":"67bbade8a8c89b98ec377944","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67bbade8a8c89b98ec377944/HPtKDo8fnKr4OxpN1Z17D.png","isPro":false,"fullname":"Urodoc Oncall","user":"UDCAI","type":"user"},{"_id":"628f0263f6c5ab7ba0dc7109","avatarUrl":"/avatars/a319c18a37233308bca23fceeb5e6c9d.svg","isPro":false,"fullname":"Thinh Phan","user":"ThinhPhan97","type":"user"},{"_id":"6322e02496d6c8518eb05292","avatarUrl":"/avatars/6b5d317f9ccb0046e40918c7861b4cb5.svg","isPro":false,"fullname":"Zhuoyi Yang","user":"keg-yzy","type":"user"},{"_id":"67cd327432668b04f4555270","avatarUrl":"/avatars/15e2cef976cbe05c4c5858c88dccf4af.svg","isPro":false,"fullname":"Yanling Wang","user":"WYLing","type":"user"},{"_id":"64f5c2f19eaf9d8fb74d2b47","avatarUrl":"/avatars/2623c5fc9bff757f20b66f7626065d52.svg","isPro":false,"fullname":"Zehai He","user":"he-zh22","type":"user"},{"_id":"6438bca4a5e10f6d58694b47","avatarUrl":"/avatars/3aeb25fbc73c5cab1265e13d11adfb76.svg","isPro":false,"fullname":"Xiaotao Gu","user":"xgeric","type":"user"},{"_id":"662f443b0158a8f30d59b8be","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/zhdvRH-NBUbsL3rMwSzQX.png","isPro":false,"fullname":"xuezhao","user":"strongxue","type":"user"},{"_id":"65fd9a138ba1bbf3261e6eac","avatarUrl":"/avatars/abdab315cda21babd1edfe7b394a9e70.svg","isPro":false,"fullname":"wenhao yan","user":"teal024","type":"user"},{"_id":"64f9c64c698370119fb36b6a","avatarUrl":"/avatars/dae735d8c471bcfa5d1a7fc1d921ca13.svg","isPro":false,"fullname":"Shoumobutangshou","user":"SPpump","type":"user"},{"_id":"62d7b131f6e8ba66107af761","avatarUrl":"/avatars/f1c5df47aef69c824fd166722df8f670.svg","isPro":false,"fullname":"Jiazheng Xu","user":"xujz0703","type":"user"},{"_id":"635a1be06805fab09b1059e1","avatarUrl":"/avatars/cc6ccd0b4b6f074f1e920858197f3c2d.svg","isPro":false,"fullname":"tenghui.li","user":"tenpha","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.10804.md"}">

Papers

arxiv:2606.10804

SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning

Published on Jun 9

· Submitted by

taesiri on Jun 10

Upvote

Authors:

Abstract

SCAIL-2 enables end-to-end character animation by directly transferring motion from driving videos without intermediate representations, using unified task decomposition and synthetic data generation.

Generated by Qwen/Qwen2.5-Coder-32B-Instruct

Controlled character animation requires transferring motion from a driving sequence to a reference character. Prior works heavily rely on intermediate representations, including pose skeletons to represent motion or masked background to represent environment, which inevitably leads to information loss. To address this, we present SCAIL-2, an framework that bypasses those intermediates and achieves end-to-end character animation. By directly concatenating driving videos to the sequence, the model can obtain all the required visual information from the input video. To address lack of end-to-end data, we unify sub-tasks of character animation with decoupled conditions and then curate a pipeline to synthesize MotionPair-60K, an end-to-end motion transfer dataset containing heterogeneous tasks of character animation. To archive the unification, we utilize in-context mask conditioning and mode-specific RoPE as soft guidance beyond textual instructions and raw visual information. To address synthetic discrepancy in detailed regions, we propose Bias-Aware DPO to construct preference items to mitigate the errors. Extensive experiments demonstrate that our method substantially outperforms existing state-of-the-art approaches in various character animation tasks. A large subset of synthetic data as well as model weights will be released at our project page: https://teal024.github.io/SCAIL-2/.