[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-147257-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-147257-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","how-to-train-your-dragon-automatic-diffusion-based-rigging-for-characters-with-diverse-topologies","How to Train Your Dragon - Automatic Diffusion-Based Rigging for Characters with Diverse Topologies","","Recent diffusion-based methods produce strong results for animating images of human subjects, yet they largely depend on human-specific pose representations and extensive training with labeled real videos. This work extends diffusion animation to characters with diverse skeletal topologies using only 3–5 example frames paired with skeletal information. A procedural data pipeline samples varied topologies on the fly, and a new skeleton representation enables training across textures and structures. Fine-tuning rapidly adapts to new characters and generalizes to realistic and stylized cartoons, supported by a first 2D video dataset with per-frame keypoint annotations for both humanoid and non-humanoid subjects.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/how-to-train-your-dragon-automatic-diffusion-based-rigging-for-characters-with-diverse-topologies/147257/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/how-to-train-your-dragon-automatic-diffusion-based-rigging-for-characters-with-diverse-topologies/147257.png","ImageObject",300,407,{"name":42,"@type":43},"Aurelia","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-18","2026-08-26",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",9,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does AniDiffusion address in character animation?","Question",{"text":63,"@type":64},"It simplifies 2D rigging by inferring rigging behavior from a small set of example images and their corresponding skeletal structures, avoiding a traditional arduous manual rigging workflow.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How much example data does the method require to rig a new character?",{"text":68,"@type":64},"It uses a small number of example frames, typically 3–5, where the character appears in different poses with corresponding skeletal information.",{"name":70,"@type":61,"acceptedAnswer":71},"How does the method generalize beyond humanoid characters?",{"text":72,"@type":64},"It does not assume a fixed topology or texture category; instead, it adapts to unseen target characters and works on a wide range of objects by using a pose representation and a procedural training strategy with diverse topologies.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},147257,1787760907,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144,"preview_page_count":4},1099514068365,"https://ap-avatar.wpscdn.com/avatar/10000253d8d9f28188e?_k=1776742907772140068","arXiv :2503 . 15586v1 [ cs .GR] 19 Mar 2025  \nHow to Train Your Dragon: Automatic Diffusion-Based Rigging for  \nCharacters with Diverse Topologies  \nZeqi Gu 1,3 Difan Liu2 Timothy Langlois2 Matthew Fisher2 Abe Davis3  \n1Cornell Tech, USA  \n2Adobe Research, USA  \n3 Cornell University, USA  \nFigure 1: Workflow. We propose AniDiffusion, a diffusion-based automatic rigging method that helps cartoon creators re-pose a character without going through the traditional arduous rigging procedure. Our method first provides an annotation user interface and asks the user to provide several keyframes of the character with desired control points annotated. Then we invite the user to input the coordinates of those keypoints to indicate how they want to re-pose the character per-frame. After a simple fine-tuning, our model is able to synthesize the character in the new motion. Unlike most existing works, our method does not assume the topology nor the texture of the character belonging  \nto a certain category such as humanoids, and thus works on a much wider range of objects, such as a fox (blue and green box), a pinwheel (orange box, top row), and a dragon (orange box, bottom row) . Image credits: fox ©Edina Gecse and pinwheel ©Ekaterine Kantaria; the dragon is self-created.  \nAbstract  \nRecent diffusion-based methods have achieved impressive results on animating images of human subjects. However, most of that success has built on human-specific body pose representations and extensive training with labeled real videos. In this work, we extend the ability of such models to animate images of characters with more diverse skeletal topologies. Given a small number (3– 5) of example frames showing the character in different poses with corresponding skeletal information, our model quickly infers a rig for that character that can generate images corresponding to new skeleton poses. We propose a procedural data generation pipeline that efficiently samples training data with diverse topologies on the fly. We use it, along with a novel skeleton representation, to train our model on articulated shapes spanning a large space of textures and topologies. Then during  \nfine-tuning, our model rapidly adapts to unseen target characters and generalizes well to rendering new poses, both for realistic and more stylized cartoon appearances. To better evaluate performance on this novel and challenging task, we create the first 2D video dataset that contains both humanoid and non-humanoid subjects with per-frame keypoint annotations. With extensive experiments, we demonstrate the superior quality of our results.  \nCCS Concepts  \n• Computing methodologies → Animation; Image manipulation;  \n2 of 11 Z. Gu et al. / How to Train Your Dragon  \n1. Introduction  \nAnimation has always been a labor-intensive process. Before the introduction of digital tools, animation was done by drawing each individual frame by hand, which meant an enormous amount of redundant work went into creating long sequences. Modern software helps reduce this redundant work by representing characters with a hierarchy of parts and corresponding transformations, which are collectively called a rig. Each part in the rig represents geometry that can appear in multiple frames, and each transformation represents an interpolatable way that the character can move. These rigs let artists animate a character by specifying the configuration of a corresponding skeletal representation at a sparse set of keyframes. This representation can greatly accelerate the animation workflow. However, the process of creating an animation rig, called rigging, can be complicated and tedious. It involves factoring the character’s geometry into parts and specifying how each part is affected by transformations in the skeletal hierarchy, including the relative rigidity and bending of different parts in response to motion. To further complicate the process, much of a rig’s information typically resides in metadata that the artist mus","cbCainPl70Cu06G1","https://ap.wps.com/l/cbCainPl70Cu06G1","pdf",25577149,11,"English","# Introduction\n## Contributions","[{\"question\":\"What problem does AniDiffusion address in character animation?\",\"answer\":\"It simplifies 2D rigging by inferring rigging behavior from a small set of example images and their corresponding skeletal structures, avoiding a traditional arduous manual rigging workflow.\"},{\"question\":\"How much example data does the method require to rig a new character?\",\"answer\":\"It uses a small number of example frames, typically 3–5, where the character appears in different poses with corresponding skeletal information.\"},{\"question\":\"How does the method generalize beyond humanoid characters?\",\"answer\":\"It does not assume a fixed topology or texture category; instead, it adapts to unseen target characters and works on a wide range of objects by using a pose representation and a procedural training strategy with diverse topologies.\"}]","How to Train Your Dragon - Automatic Diffusion-Based Rigging for Characters with Diverse Topologies | PDF",28]