AI Paper Atlas

Research map / 3D vision / Gaussian splatting

GAP: Gaussianize Any Point Clouds with Text Guidance

Weiqi Zhang, Junsheng Zhou, Haotian Geng, Wenyuan Zhang, Yu-Shen Liu

ICCV 2025 · 3D vision › Gaussian splatting · approach: Regularization and primitive design for 3DGS · 4 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
add scale loss (ablation)GAP · Objaverse ablationFID: 214.63 -> 40.39Improved
add scale loss (ablation)GAP · Objaverse ablationCLIP score: 26.25 -> 27.26Improved
add distance loss (surface anchoring) (ablation)GAP · Objaverse ablationFID: 161.04 -> 40.39Improved
add distance loss (surface anchoring) (ablation)GAP · Objaverse ablationCLIP score: 24.30 -> 27.26Improved
add diffuse-based Gaussian inpainting (ablation)GAP · Objaverse ablationFID: 46.37 -> 40.39Improved
add diffuse-based Gaussian inpainting (ablation)GAP · Objaverse ablationCLIP score: 27.21 -> 27.26No clear effect
add diffuse-based Gaussian inpainting (ablation)GAP · Objaverse ablationKID (x10^-3): 8.77 -> 5.28Improved
add scale loss (ablation)GAP · Objaverse ablationKID (x10^-3): 79.04 -> 5.28Improved
replace mesh reconstruction (BPA) + UV pipeline with direct Gaussian optimization vs SyncMVDGAP · Objaverse text-driven appearance generation (410 objects)FID: 60.29 -> 40.39Improved
replace CAP-UDF reconstructed mesh + Text2Tex with GAPGAP · Objaverse text-driven appearance generation (410 objects)FID: 52.78 -> 40.39Improved
replace SyncMVD (artist UV meshes) with GAPGAP · Objaverse text-driven appearance generation (410 objects)FID: 40.85 -> 40.39Improved
use reconstructed meshes (BPA) instead of artist UV meshes for SyncMVDGAP · Objaverse text-driven appearance generation (410 objects)FID: 40.85 -> 60.29Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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