AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Diffusion-Generated Pseudo-Observations for High-Quality Sparse-View Reconstruction

Xinhang Liu, Jiaben Chen, Shiu-Hong Kao, Yu-Wing Tai, Chi-Keung Tang

ECCV 2024 · 3D vision › Gaussian splatting · approach: Sparse-view Gaussian splatting for novel views

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 depth conditioning to the diffusion model (ablation)Deceptive-NeRF · Hypersim 20-viewPSNR: 18.79 -> 22.41Improved
add progressive densification (ablation)Deceptive-NeRF · Hypersim 20-viewPSNR: 19.90 -> 22.41Improved
add the two-stage noise-augmentation training of the diffusion model (ablation)Deceptive-NeRF · Hypersim 20-viewPSNR: 20.49 -> 22.41Improved
add textual inversion embedding (ablation)Deceptive-NeRF · Hypersim 20-viewPSNR: 20.58 -> 22.41Improved
add image captioning text embedding (ablation)Deceptive-NeRF · Hypersim 20-viewPSNR: 21.59 -> 22.41Improved
densify observations fivefold vs tenfold (and none) (ablation)Deceptive-NeRF/3DGS · Hypersim 10-viewvisual qualityImproved
diffusion pseudo-observations instead of image restoration models for coarse renders (ablation)Deceptive-NeRF · qualitativevisual qualityImproved
diffusion pseudo-observations as extra training views instead of diffusion-as-regularizerfew-view NeRF · training speedtraining speed: nearly ten times fasterImproved
Deceptive-NeRF vs FreeNeRF on LLFF 9-viewDeceptive-NeRF · LLFF 9-viewPSNR: 25.13 -> 25.34Improved

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