AI Paper Atlas

Research map / 3D vision / Face and shape

RnG: A Unified Transformer for Complete 3D Modeling from Partial Observations

Mochu Xiang, Zhelun Shen, Xuesong Li, Jiahui Ren, Jing Zhang, Chen Zhao, Shanshan Liu, Haocheng Feng, Jingdong Wang, Yuchao Dai

CVPR 2026 · 3D vision › Face and shape · approach: Single-view 3D reconstruction · 2 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
initialize from pretrained VGGT weights (ablation)RnG · GSO ablation (15K)PSNR: 20.784 -> 24.856Improved
initialize from pretrained VGGT weights (ablation)RnG · GSO ablation (15K)camera pose AUC@30: 33.654 -> 82.631Improved
add camera head supervision (ablation)RnG · GSO ablation (15K)PSNR: 24.850 -> 24.856No clear effect
use causal mask instead of full bidirectional attention (ablation)RnG · GSO ablation (15K)PSNR: 24.861 -> 24.856No clear effect
use KV-cache with causal attention (ablation)RnG · single novel view inferencetime: 213ms -> 85msImproved
use KV-cache with causal attention (ablation)RnG · single novel view inferenceFLOPS: 12.26 T -> 2.29 TImproved
scale training data and steps (15K on 44.5K objects to 40K on 113.5K objects) (ablation)RnG · GSOPSNR: 24.856 -> 26.276Improved
train RnG for 40K steps vs LVSM for 40K stepsRnG vs LVSM · GSOPSNR: 24.619 -> 26.276Improved
increase number of input views from 1 to 8RnG · GSOPSNR: 19.452 -> 27.608Improved
increase number of input views from 4 to 2RnG · GSOPSNR: 26.276 -> 23.403Hurt
replace unposed VGGT+LVSM pipeline by RnGRnG · GSOPSNR: 19.394 -> 26.276Improved
use RnG instead of Matrix3D (unposed)RnG · GSOChamfer Distance: 0.0670 -> 0.0067Improved

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