DiffGI: Differentiable Geometry Images for High-Fidelity Thin-Shell 3D Generation
Summary
DiffGI introduces a differentiable geometry image representation for high-fidelity thin-shell 3D generation, enabling end-to-end optimization and superior reconstruction quality.
View Cached Full Text
Cached at: 07/21/26, 06:36 AM
Paper page - DiffGI: Differentiable Geometry Images for High-Fidelity Thin-Shell 3D Generation
Source: https://huggingface.co/papers/2607.13365
Abstract
Existing3Dgenerativemodelspredominantlyrelyonimplicitvolumetricrepresentations,whichenforcewatertighttopologyandstruggletorepresentthin-shellandnon-manifoldgeometriessuchasgarments.Geometryimage-basedapproachesofferasurface-centricalternative,butexistingmethodsrelyondiscretebinaryoccupancymapswhoseresolution-dependentboundaryencodingcausesstaircaseartifactsandinformationlossupondownsampling,whilesurfacereconstructionremainsanon-differentiablepost-processingstepdisconnectedfromthelearningpipeline.Toaddressthis,weproposeDifferentiableGeometryImage(DiffGI),anend-to-end3D-to-2Dmappingframeworkthatseamlesslyintegratessurfacerepresentationandgeometricoptimization.DiffGIreplacesbinarymapswithacontinuous2DTruncatedSignedDistanceFunction(TSDF),whichencodesboundarypositionatsubpixelprecisionwithinafixedgridresolution,eliminatingresolution-dependentstaircaseartifactsevenunderaggressivedownsampling.Buildingonthiscontinuousfield,weintroduceadifferentiableMarchingSquaresalgorithmbasedonanalyticallinearinterpolation,allowinggradientsfrom3Dsurfacelossestopropagatebacktothe2Dlatentspace.Leveragingthisdifferentiablepipeline,wetrainaDiffGI-VAEaugmentedwithageometry-awarenormalrenderinglosstocompresscomplex3Dsurfacesintoanultra-compact32X32latentspace,andinstantiateatransformer-basedlatentdiffusionmodelwithaflow-matchingobjectiveontopofthisspaceforconditional3Dgeneration.Extensiveexperimentsongarmentandobjectdatasetsdemonstratethatourmethodachievessuperiorreconstructionfidelityandboundaryprecisioncomparedtopriorgeometry-imageandvoxel-basedapproaches,whilerequiringsignificantlyfewercomputationalresources.
View arXiv pageView PDFProject pageAdd to collection
Get this paper in your agent:
hf papers read 2607\.13365
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2607.13365 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2607.13365 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2607.13365 in a Space README.md to link it from this page.
Collections including this paper1
Similar Articles
CGGS: Consistency-Augmented Geometric Gaussian Splatting for Ego-centric 3D Scene Generation
CGGS is a text-to-3D framework that improves geometric consistency and quality in ego-centric 3D scene generation using a multi-stage approach with consistency-augmented loss, layout decoration, and geometric refinement via entropy-based depth loss.
VGGT-Diff: Visual Geometry Meets Diffusion for Sparse-View Novel View Synthesis
VGGT-Diff is a geometry-routed multi-view diffusion model that integrates visual geometry latents with a pretrained diffusion model to improve sparse-view novel view synthesis, achieving competitive or state-of-the-art performance.
FLUX3D: High-Fidelity 3D Gaussian Generation with Diffusion-Aligned Sparse Representation
FLUX3D introduces a framework for high-fidelity image-to-3D Gaussian Splatting generation by enhancing representation learning and cross-modal alignment with diffusion-aligned structured latents and a sparse-structure-aware diffusion transformer, achieving state-of-the-art results.
T3lescope: Arbitrary-Resolution High-Fidelity Generative Surface Reconstruction from Images
T3lescope提出一种单一固定分辨率生成器的推理时粗到细级联方法,从多视角图像实现跨尺度(从单物体到大型户外场景)的任意分辨率高保真3D场景网格重建,无需逐场景优化,在室内、户外和城市规模场景上超越了前馈与生成基线。
Geometry-Aware Representation Denoising for Robust Multi-view 3D Reconstruction
Introduces GARD, a diffusion-based framework that operates in the feature space of a feed-forward 3D reconstructor to jointly recover scene geometry and high-quality imagery from degraded inputs.