[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-156009-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-156009-en":131},{"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","sketchfacegs-real-time-sketch-driven-face-editing-and-generation-with-gaussian-splatting","SketchFaceGS - Real-Time Sketch-Driven Face Editing and Generation with Gaussian Splatting","","3D Gaussian representations enable photorealistic real-time rendering for digital head modeling, yet intuitive interactive creation and editing of 3D Gaussian head models remains difficult. 2D sketches are sparse and depth-ambiguous, lacking high-frequency appearance cues needed to infer dense, geometrically consistent 3D Gaussian structures under real-time constraints. SketchFaceGS presents a feed-forward coarse-to-fine framework that generates and edits photorealistic 3D Gaussian head models from 2D sketches, using Transformer-based UV prediction and UV Mask Fusion for real-time, free-viewpoint editing.",{"@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/sketchfacegs-real-time-sketch-driven-face-editing-and-generation-with-gaussian-splatting/156009/",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/sketchfacegs-real-time-sketch-driven-face-editing-and-generation-with-gaussian-splatting/156009.png","ImageObject",300,407,{"name":42,"@type":43},"Eden","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-27","2026-08-28",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",12,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What challenge does SketchFaceGS address in 3D Gaussian head editing?","Question",{"text":63,"@type":64},"SketchFaceGS targets the difficulty of intuitively creating and interactively editing 3D Gaussian head models from sketches in real time, where strokes are sparse and depth-ambiguous and do not directly provide dense 3D structure cues.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does SketchFaceGS generate a photorealistic 3D face from a 2D sketch?",{"text":68,"@type":64},"It uses a feed-forward coarse-to-fine pipeline: a Transformer-based UV feature-prediction module reconstructs a coarse geometrically consistent UV map, then a 3D UV enhancement module refines high-frequency photorealistic details.",{"name":70,"@type":61,"acceptedAnswer":71},"How does SketchFaceGS support interactive editing?",{"text":72,"@type":64},"It introduces UV Mask Fusion: user draw/erase strokes are converted to 2D masks, back-projected to identify contributing 3D Gaussians, mapped to a canonical UV space to create UV masks, then fused layer-by-layer in the generator for seamless real-time edits.","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},156009,1787942628,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"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":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":140,"language":141,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":12,"update_tm":80,"read_time":145,"preview_page_count":4},1374391974468,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","This CVPR paper is the Open Access version, provided by the Computer Vision Foundation.  \nExcept for this watermark, it is identical to the accepted version; the final published version of the proceedings is available on IEEE Xplore.  \nSketchFaceGS: Real-Time Sketch-Driven Face Editing and Generation with  \nGaussian Splatting  \nBo Li 1,2 Jiahao Kang2 Yubo Ma2 Feng-Lin Liu3,4 Bin Liu2 Fang-Lue Zhang5 Lin Gao3,4 *  \n1 Shandong Technology and Business University  \n2Nanchang Hangkong University  \n3Institute of Computing Technology, Chinese Academy of Sciences  \n4University of Chinese Academy of Sciences  \n5 University of New South Wales  \n[bolimath@gmail.com](bolimath@gmail.com) , [2307070100013@stu.nchu.edu.cn](2307070100013@stu.nchu.edu.cn) , [2304081200005@stu.nchu.edu.cn](2304081200005@stu.nchu.edu.cn) ,  \n[liufenglin21s@ict.ac.cn](liufenglin21s@ict.ac.cn) , [nyliubin@nchu.edu.cn](nyliubin@nchu.edu.cn) , [z.fanglue@gmail.com](z.fanglue@gmail.com) , [gaolin@ict.ac.cn](gaolin@ict.ac.cn)  \nView Change   \nView Change   \nOriginal 3D Face Sequential Editing Operations Realistic Editing Results  \nFigure 1 . SketchFaceGS presents a real-time, sketch-driven editing framework tailored for 3D Gaussian Splatting (3DGS) facial heads. Starting with an original 3D face (a), sequential editing operations (b) are applied to targeted local regions. Our approach delivers photorealistic editing results (c) while preserving the 3D consistency.  \nAbstract  \n3D Gaussian representations have emerged as a powerful paradigm for digital head modeling, achieving photorealistic quality with real-time rendering. However, intuitive and interactive creation or editing of 3D Gaussian head models remains challenging. Although 2D sketches provide an ideal interaction modality for fast, intuitive conceptual design, they are sparse, depth-ambiguous, and lack high-frequency appearance cues, making it difficult to infer dense, geometrically consistent 3D Gaussian structures from strokes—especially under real-time constraints. To address these challenges, we propose SketchFaceGS, the first  \n* Corresponding author.  \nsketch-driven framework for real-time generation and editing of photorealistic 3D Gaussian head models from 2D sketches. Our method uses a feed-forward, coarse-to-fine architecture. A Transformer-based UV feature-prediction module first reconstructs a coarse but geometrically consistent UV feature map from the input sketch, and then a 3D UV feature enhancement module refines it with highfrequency, photorealistic detail to produce a high-fidelity 3D head. For editing, we introduce a UV Mask Fusion technique combined with a layer-by-layer feature-fusion strategy, enabling precise, real-time, free-viewpoint modifications. Extensive experiments show that SketchFaceGS outperforms existing methods in both generation fidelity and editing flexibility, producing high-quality, editable 3D  \nheads from sketches in a single forward pass.  \n1. Introduction  \nAccurate 3D facial modeling remains a challenging problem in computer graphics and computer vision, underpinning applications from cinematic characters to immersive avatars. Prior mesh-based pipelines [1, 46] typically rely on labor-intensive workflows to produce photorealistic faces that capture both geometry and appearance, making them impractical for non-experts and costly to achieve photoreal quality. Recent NeRF-based approaches [2, 19, 42, 47] leverage 3D GANs for controllable face synthesis; however, their practical adoption is often limited by costly rendering and 3D inconsistencies introduced by super-resolution modules. Although 3D Gaussian splatting [32] addresses real-time rendering, existing GAN-based Gaussian head generation methods [26, 33] and text-driven face-editing approaches [50, 51, 54] lack fine-grained, convenient control for detailed facial edits.  \nWe propose an intuitive, sketch-driven approach for 3D face editing that leverages 3D Gaussian Splatting. To overcome the reliance on time-co","cbCaihkPyjtjFl2w","https://ap.wps.com/l/cbCaihkPyjtjFl2w","pdf",25775489,11,"English","# Abstract\n# 1. Introduction\n## Problem and limitations of existing methods\n## Proposed sketch-driven framework (coarse-to-fine)","[{\"question\":\"What challenge does SketchFaceGS address in 3D Gaussian head editing?\",\"answer\":\"SketchFaceGS targets the difficulty of intuitively creating and interactively editing 3D Gaussian head models from sketches in real time, where strokes are sparse and depth-ambiguous and do not directly provide dense 3D structure cues.\"},{\"question\":\"How does SketchFaceGS generate a photorealistic 3D face from a 2D sketch?\",\"answer\":\"It uses a feed-forward coarse-to-fine pipeline: a Transformer-based UV feature-prediction module reconstructs a coarse geometrically consistent UV map, then a 3D UV enhancement module refines high-frequency photorealistic details.\"},{\"question\":\"How does SketchFaceGS support interactive editing?\",\"answer\":\"It introduces UV Mask Fusion: user draw/erase strokes are converted to 2D masks, back-projected to identify contributing 3D Gaussians, mapped to a canonical UV space to create UV masks, then fused layer-by-layer in the generator for seamless real-time edits.\"}]","SketchFaceGS - Real-Time Sketch-Driven Face Editing and Generation with Gaussian Splatting | PDF",28]