[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83927-en":3,"doc-seo-83927-105":29,"detail-sidebar-cat-0-en-105":90},{"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":4,"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":13,"seo_description":14,"update_tm":27,"read_time":28},83927,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","GUSH3R Everyone Everywhere All at Once as Gaussians","Reconstructing dynamic human-scene environments from monocular videos requires jointly modeling scene geometry, camera motion, and non-rigid human dynamics while supporting photorealistic rendering. GUSH3R introduces a feed-forward, online framework that reconstructs dynamic humans (“everyone”) and static scenes (“everywhere”) in one forward pass. The method outputs geometrically consistent 3D Gaussian Splatting primitives, enabling novel view synthesis. Experiments on monocular human-scene datasets show competitive visual quality and substantially improved inference efficiency versus optimization-based approaches.","arXiv :2607 .05243v 1 [ cs .CV] 6 Jul 2026  \nGUSH3R: Everyone Everywhere All at Once as  \nGaussians  \nKeito Abe, Kaede Shiohara Takashi Otonari, Toshihiko Yamasaki  \nThe University of Tokyo  \n{abe, shiohara, otonari, [yamasaki}@cvm.t.u-tokyo.ac.jp](yamasaki}@cvm.t.u-tokyo.ac.jp)[ ](yamasaki}@cvm.t.u-tokyo.ac.jp)Project page: [https://abkeito.github.io/gush3r-page/](https://abkeito.github.io/gush3r-page/)  \nFigure 1: GUSH3R (Gaussian-Unified Scene Human 3D Reconstruction) takes a monocular video as input and produces dynamic human-scene representations using 3D Gaussians.  \nAbstract  \nReconstructing dynamic human-scene environments from monocular videos is a challenging problem that requires jointly modeling scene geometry, camera motion, and non-rigid human dynamics while enabling photorealistic rendering. Recent feed-forward methods can efficiently predict geometry, but they are often limited to non-photorealistic representations such as point clouds and meshes, or they fail to handle non-rigid objects, particularly dynamic humans. To fill this gap, we present GUSH3R (Gaussian-Unified Scene Human 3D Reconstruction), a feed-forward framework for online dynamic human-scene reconstruction. From a monocular human-scene video, our method reconstructs dynamic humans (everyone) and static scenes (everywhere) in a single forward pass (all at once) as 3D Gaussian Splatting (3DGS) primitives (as gaussians), which are geometrically consistent and capable of novel view synthesis. Experiments on monocular human-scene datasets demonstrate that our approach achieves competitive novel view synthesis quality while significantly improving inference efficiency compared to optimization-based methods.  \n∗Project lead.  \nPreprint.  \n1 Introduction  \nReconstructing dynamic human-scene environments from monocular videos is an important problem in computer vision, with applications in novel view synthesis [23, 32], virtual and augmented reality [68], and digital human modeling [20] . Given only a monocular video, the goal of dynamic human-scene reconstruction is to jointly recover scene geometry, camera motion, and dynamic humans, while enabling photorealistic novel view synthesis.  \nExisting 3D/4D reconstruction approaches can be broadly categorized into optimization-based and feed-forward methods. Optimization-based methods, including Neural Radiance Fields (NeRF) [41] and 3D Gaussian Splatting (3DGS) [26]-based approaches, optimize 3D/4D scene representations for each scene [41, 26, 63, 40, 46] . While these methods achieve high reconstruction quality, the optimization process is costly, making them impractical for fast or real-time inference [11, 65, 72] . Moreover, in the 4D reconstruction setting, most methods require multi-view videos [63, 40, 17] or additional sensors such as LiDAR or depth [36], limiting their applicability in real-world scenarios. In contrast, feed-forward methods predict 3D geometry directly from images in a single forward pass, enabling fast inference [61, 27, 58, 23, 33, 60, 69] . These approaches generalize well to unseen scenes by leveraging strong geometric priors learned from large-scale data [49, 66, 34, 30, 19, 64, 12, 50, 2, 55, 54, 21, 73, 6, 44, 13] . Yet, handling dynamic humans and achieving photorealistic rendering quality at the same time remains a significant challenge [23, 10] .  \nAs summarized in Table 1, existing feed-forward methods either do not explicitly model dynamic humans or do not provide photorealistic renderable representations. In this work, we take a first step toward feed-forward photorealistic, renderable dynamic human-scene reconstruction from monocular videos. To this end, we leverage geometric and human priors [10] and lift them into a unified 3DGS representation. Our representation consists of dynamic human Gaussians and static scene Gaussians, whose appearance is predicted by respective decoders for humans and scenes. Our method enables feed-forward reconstruction of dynamic human-scen","cbCaiiYXVOwp30lH","https://ap.wps.com/l/cbCaiiYXVOwp30lH","pdf",6608180,1,15,"English","en",105,"# Introduction\n## Problem setting and motivation\n## Related work\n### 3D reconstruction\n# Conceptual overview\n## Method comparison table","[{\"question\":\"What problem does GUSH3R address?\",\"answer\":\"GUSH3R targets dynamic human-scene reconstruction from a monocular video, requiring recovery of scene geometry, camera motion, and non-rigid human dynamics with photorealistic rendering support.\"},{\"question\":\"How does GUSH3R differ from optimization-based methods?\",\"answer\":\"Optimization-based approaches iteratively optimize 3D/4D representations and are computationally costly. GUSH3R is feed-forward, performing reconstruction in a single forward pass for faster inference.\"},{\"question\":\"What outputs does GUSH3R produce and how are they used?\",\"answer\":\"Given a monocular human-scene video, GUSH3R reconstructs dynamic humans and static scenes as 3D Gaussian Splatting primitives, allowing geometrically consistent novel view synthesis.\"}]",1784191493,38,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"gush3r-everyone-everywhere-all-at-once-as-gaussians","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/gush3r-everyone-everywhere-all-at-once-as-gaussians/83927/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does GUSH3R address?","Question",{"text":74,"@type":75},"GUSH3R targets dynamic human-scene reconstruction from a monocular video, requiring recovery of scene geometry, camera motion, and non-rigid human dynamics with photorealistic rendering support.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does GUSH3R differ from optimization-based methods?",{"text":79,"@type":75},"Optimization-based approaches iteratively optimize 3D/4D representations and are computationally costly. GUSH3R is feed-forward, performing reconstruction in a single forward pass for faster inference.",{"name":81,"@type":72,"acceptedAnswer":82},"What outputs does GUSH3R produce and how are they used?",{"text":83,"@type":75},"Given a monocular human-scene video, GUSH3R reconstructs dynamic humans and static scenes as 3D Gaussian Splatting primitives, allowing geometrically consistent novel view synthesis.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]