AI Paper Atlas

Research map / 3D vision / Face and shape

DiffGI: Differentiable Geometry Images for High-Fidelity Thin-Shell 3D Generation

Eungjune Shim, Hansol Lee, Eunjung Ju

ECCV 2026 · 3D vision › Face and shape · approach: Textured mesh generation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace binary occupancy with continuous TSDF (no normal loss) (ablation)DiffGI-VAE · GarmageSetCD (x1e-3): 1.503 -> 0.595Improved
replace binary occupancy with continuous TSDF (no normal loss) (ablation)DiffGI-VAE · GarmageSetJSD (x1e-3): 4.539 -> 2.169Improved
add geometry-aware normal rendering loss (ablation)DiffGI-VAE (TSDF) · GarmageSetNC: 0.921 -> 0.961Improved
add geometry-aware normal rendering loss (ablation)DiffGI-VAE (TSDF) · GarmageSetCD (x1e-3): 0.595 -> 0.461Improved
add normal rendering loss to occupancy representation (ablation)DiffGI-VAE (Occ) · GarmageSetNC: 0.906 -> 0.947Improved
initialize VAE from SD1.5 instead of random (ablation)DiffGI-VAE · GarmageSetCD (x1e-3): 0.47 -> 0.46No clear effect
compress 256x256 geometry image into 32x32x4 latent vs uncompressed 64x64x4 OmagesDiffGI-VAE · ABONC: 0.89 -> 0.83Hurt
compress into 32x32x4 latent vs OmagesDiffGI-VAE · ABOCD (x1e-3): 0.89 -> 0.83Improved
use DiffGI diffusion vs Omages for label-conditioned generationDiffGI · ABOmean P-FID: 31.14 -> 19.91Improved
use DiffGI diffusion vs Omages for Table categoryDiffGI · ABO Table categoryP-KID: 0.10 -> 0.11No clear effect
geometric UV augmentation (3,800 -> 500K samples) for larger DiT (ablation)DiT-Base · ABOtraining stabilityImproved
TSDF vs occupancy at resolution 64/128 without VAE (ablation)geometry image · ABO and GarmageSetHausdorff distanceImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy