ECCV 2026</p>\n","updatedAt":"2026-06-24T15:11:28.124Z","author":{"_id":"63720d0dcc36c72c0109a8be","avatarUrl":"/avatars/3189c216dcaa6488fc7f2c81ae48b356.svg","fullname":"Ray","name":"BatofGo","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"de","probability":0.3562392294406891},"editors":["BatofGo"],"editorAvatarUrls":["/avatars/3189c216dcaa6488fc7f2c81ae48b356.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.22197","authors":[{"_id":"6a3bf35500eadb5867edffc2","name":"Rui Wang","hidden":false},{"_id":"6a3bf35500eadb5867edffc3","name":"Quentin Lohmeyer","hidden":false},{"_id":"6a3bf35500eadb5867edffc4","name":"Siyu Tang","hidden":false},{"_id":"6a3bf35500eadb5867edffc5","name":"Mirko Meboldt","hidden":false}],"publishedAt":"2026-06-20T00:00:00.000Z","submittedOnDailyAt":"2026-06-24T00:00:00.000Z","title":"Multi4D: High-Fidelity Dynamic Gaussian Splatting via Multi-Level Competitive Allocation","submittedOnDailyBy":{"_id":"63720d0dcc36c72c0109a8be","avatarUrl":"/avatars/3189c216dcaa6488fc7f2c81ae48b356.svg","isPro":false,"fullname":"Ray","user":"BatofGo","type":"user","name":"BatofGo"},"summary":"Dynamic 3D Gaussian splatting faces a fundamental tension between motion consistency and visual fidelity. Deformation-based approaches preserve temporal correspondence but suffer from motion over-factorization, oversmoothing high-frequency dynamics. In contrast, 4D-primitive methods capture fine visual details yet incur temporal overparameterization, breaking object identity and leading to severe storage overhead. To resolve this, we introduce Multi4D, a framework for high-fidelity dynamic Gaussian Splatting based on multi-level competitive allocation. Instead of a monolithic representation, we distribute modeling capacity across three structured levels: static structure, persistent dynamic geometry, and transient appearance primitives. Through shared rasterization and residual-driven optimization, these levels dynamically compete to explain photometric error, enabling adaptive specialization without pre-assigned decomposition. This allocation preserves long-term motion consistency while capturing fine dynamic detail, achieving state-of-the-art rendering quality and real-time performance with significantly fewer dynamic primitives. Furthermore, because our representation explicitly tracks compact persistent Gaussians over time, semantic features can be embedded afterward, enabling Multi4D to achieve state-of-the-art 4D segmentation accuracy with an order-of-magnitude speedup. Project page: https://batfacewayne.github.io/Multi4D.io/","upvotes":0,"discussionId":"6a3bf35500eadb5867edffc6","projectPage":"https://batfacewayne.github.io/Multi4D.io/","githubRepo":"https://github.com/BatFaceWayne/Multi4D","githubRepoAddedBy":"user","ai_summary":"Multi4D addresses the trade-off between motion consistency and visual fidelity in dynamic 3D Gaussian splatting through a multi-level competitive allocation framework that enables adaptive specialization and efficient representation.","ai_keywords":["dynamic 3D Gaussian splatting","motion consistency","visual fidelity","deformation-based approaches","4D-primitive methods","multi-level competitive allocation","static structure","persistent dynamic geometry","transient appearance primitives","rasterization","residual-driven optimization","photometric error","adaptive specialization","temporal correspondence","motion over-factorization","oversmoothing","temporal overparameterization","object identity","storage overhead","state-of-the-art rendering quality","real-time performance","4D segmentation accuracy"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":0},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.22197.md","query":{}}">
Multi4D: High-Fidelity Dynamic Gaussian Splatting via Multi-Level Competitive Allocation
Published on Jun 20
· Submitted by Ray on Jun 24 Abstract
Multi4D addresses the trade-off between motion consistency and visual fidelity in dynamic 3D Gaussian splatting through a multi-level competitive allocation framework that enables adaptive specialization and efficient representation.
Dynamic 3D Gaussian splatting faces a fundamental tension between motion consistency and visual fidelity. Deformation-based approaches preserve temporal correspondence but suffer from motion over-factorization, oversmoothing high-frequency dynamics. In contrast, 4D-primitive methods capture fine visual details yet incur temporal overparameterization, breaking object identity and leading to severe storage overhead. To resolve this, we introduce Multi4D, a framework for high-fidelity dynamic Gaussian Splatting based on multi-level competitive allocation. Instead of a monolithic representation, we distribute modeling capacity across three structured levels: static structure, persistent dynamic geometry, and transient appearance primitives. Through shared rasterization and residual-driven optimization, these levels dynamically compete to explain photometric error, enabling adaptive specialization without pre-assigned decomposition. This allocation preserves long-term motion consistency while capturing fine dynamic detail, achieving state-of-the-art rendering quality and real-time performance with significantly fewer dynamic primitives. Furthermore, because our representation explicitly tracks compact persistent Gaussians over time, semantic features can be embedded afterward, enabling Multi4D to achieve state-of-the-art 4D segmentation accuracy with an order-of-magnitude speedup. Project page: https://batfacewayne.github.io/Multi4D.io/
Community
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.22197 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.22197 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.22197 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.