[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120122-en":3,"doc-seo-120122-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},120122,8796095461564,"Liam","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",6,"Technology","Implementing a Machine Learning Deformer for CG Crowds - Our Journey","CG crowd characters are widely used in TV shows and commercials, yet they face a critical deformations bottleneck: most crowd pipelines only ingest limited rig representations such as skinning, blend shapes, and geometry caches, which prevents high-quality close-up facial performances. To overcome this, Golaem pursued a neural-network approach to compress and approximate rig deformations. Starting from early experiments and a minimum viable product, the team navigated multiple pitfalls and dead ends, aiming to share concrete lessons for production-ready implementation and feedback.","Implementing a Machine Learning Deformer for CG Crowds: Our  \nJourney  \nBastien Arcelin∗ Golaem Chantepie, France [bastien.arcelin@golaem.com](bastien.arcelin@golaem.com)  \nSebastien Maraux∗ Golaem Chantepie, France [sebastien.maraux@golaem.com](sebastien.maraux@golaem.com)  \nNicolas Chaverou  \nGolaem Nouméa, New Caledonia [nicolas.chaverou@golaem.com](nicolas.chaverou@golaem.com)  \narXiv :2406 .09783v1 [ cs .GR] 14 Jun 2024  \nFigure 1: Examples of facial deformations using our Machine Learning Deformer in Golaem Crowd © Stim Studio  \nABSTRACT  \nCG crowds have become increasingly popular this last decade in the VFX and animation industry: formerly reserved to only a few high end studios and blockbusters, they are now widely used in TV shows or commercials. Yet, there is still one major limitation: in order to be ingested properly in crowd software, studio rigs have to comply with specific prerequisites, especially in terms of deformations. Usually only skinning, blend shapes and geometry caches are supported preventing close-up shots with facial performances on crowd characters. We envisioned two approaches to tackle this: either reverse engineer the hundreds of deformer nodes available in the major DCCs/plugins and incorporate them in our crowd package, or surf the machine learning wave to compress the deformations of a rig using a neural network architecture. Considering we could not commit 5+ man/years of development into this problem, and that we were excited to dip our toes in the machine learning pool, we went for the latter.  \nFrom our first tests to a minimum viable product, we went through hopes and disappointments: we hit multiple pitfalls, took false shortcuts and dead ends before reaching our destination. With  \n∗ Both authors contributed equally to this research.  \nPermission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission [and/or a fee. Request permissions from permissions@acm.org](and/or a fee. Request permissions from permissions@acm.org).  \nDigiPro ’24, July 27, 2024, Denver, CO, USA  \n© 2024 Copyright held by the owner/author(s) . Publication rights licensed to ACM. ACM ISBN 979-8-4007-0690-5/24/07  \n[https://doi.org/10.1145/3665320.3670994](https://doi.org/10.1145/3665320.3670994)  \nthis paper, we hope to provide a valuable feedback by sharing the lessons we learnt from this experience.  \nCCS CONCEPTS  \n• Computing methodologies → Animation; Neural networks.  \nKEYWORDS  \nAnimation, Neural Networks, Rigging, Crowds  \nACM Reference Format:  \nBastien Arcelin, Sebastien Maraux, and Nicolas Chaverou. 2024. Implementing a Machine Learning Deformer for CG Crowds: Our Journey. In The Digital Production Symposium (DigiPro ’24), July 27, 2024, Denver, CO, USA. ACM, New York, NY, USA, 7 pages. [https://doi.org/10.1145/3665320.3670994](https://doi.org/10.1145/3665320.3670994)  \n1 CONTEXT  \nFounded in 2009 and based in France, Golaem is an innovative software company that grew out of the INRIA research center. It develops technologies for crowd simulation and character layout for the VFX and animation industry.  \n2 WHY DO WE NEED A ML DEFORMER FOR CG CROWDS?  \nWith the democratization ofCG crowds, studios are now expecting crowd characters to blend seamlessly with keyframed hero characters, especially in terms of facial performances. However, they always hit the same limitation: only a subset of deformers are supported in their crowd package-usually skinning, blend shapes and geometry caches. Even if they sample every single rig deformation as a blend shape, thos","cbCaiujoB7fT7xlt","https://ap.wps.com/l/cbCaiujoB7fT7xlt","pdf",5999331,1,7,"English","en",105,"# Context\n# Why do we need a ML deformer for CG crowds?\n# Our first ML deformer\n## The Promise","[{\"question\":\"Why are CG crowd character deformations currently limited in production pipelines?\",\"answer\":\"Crowd software typically supports only a subset of deformers, mainly skinning, blend shapes, and geometry caches. This prevents the same fidelity as hero rigs and makes facial close-ups difficult to achieve.\"},{\"question\":\"What solution did the team choose to address the deformer limitations?\",\"answer\":\"They opted for a machine learning approach that approximates and compresses rig deformations using neural networks, instead of reverse-engineering and integrating all deformer nodes from major DCCs.\"},{\"question\":\"What are the two deformations components in the referenced ML method, and why is that helpful?\",\"answer\":\"The method decomposes global deformation into a linear part and a nonlinear part. The nonlinear component is approximated with a DNN because it is the most computationally expensive during animation.\"}]","Implementing a Machine Learning Deformer for CG Crowds - Our Journey | PDF",1785728314,18,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"implementing-a-machine-learning-deformer-for-cg-crowds-our-journey","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/technology/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/implementing-a-machine-learning-deformer-for-cg-crowds-our-journey/120122/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04","2026-08-03",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why are CG crowd character deformations currently limited in production pipelines?","Question",{"text":76,"@type":77},"Crowd software typically supports only a subset of deformers, mainly skinning, blend shapes, and geometry caches. This prevents the same fidelity as hero rigs and makes facial close-ups difficult to achieve.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"What solution did the team choose to address the deformer limitations?",{"text":81,"@type":77},"They opted for a machine learning approach that approximates and compresses rig deformations using neural networks, instead of reverse-engineering and integrating all deformer nodes from major DCCs.",{"name":83,"@type":74,"acceptedAnswer":84},"What are the two deformations components in the referenced ML method, and why is that helpful?",{"text":85,"@type":77},"The method decomposes global deformation into a linear part and a nonlinear part. The nonlinear component is approximated with a DNN because it is the most computationally expensive during animation.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,114,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":112,"slug":113},50,"technology",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":115,"show_sort_weight":116,"slug":117},"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":46,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":107,"slug":138},19,"General","general"]