AI Paper Atlas

Research map / 3D vision / Face and shape

PersonaCraft: Personalized and Controllable Full-Body Multi-Human Scene Generation Using Occlusion-Aware 3D-Conditioned Diffusion

Gwanghyun Kim, Suh Yoon Jeon, Seunggyu Lee, Se Young Chun

ICCV 2025 · 3D vision › Face and shape · approach: Full-body avatar reconstruction · 3 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
use depth SMPLx conditioning instead of RGB (ablation)SCNet · Pose ablationMPJPE (3D): 82.52 -> 71.20Improved
use depth SMPLx conditioning instead of normal (ablation)SCNet · Pose ablationMPJPE (3D): 79.30 -> 71.20Improved
combine depth and normal SCNet residuals (ablation)SCNet-D · Pose ablationMPJPE (3D): 71.20 -> 63.16Improved
combine depth and normal SCNet residuals (ablation)SCNet-D · Pose ablationAP-0.5 (2D): 0.423 -> 0.494Improved
add OccNet (ablation)SCNet-D + SCNet-N · Pose ablationMPJPE (3D): 63.16 -> 62.92Improved
add OccCFG (ablation)SCNet-D + SCNet-N + OccNet · Pose ablationMPJPE (3D): 62.92 -> 62.64Improved
add OccCFG (ablation)SCNet-D + SCNet-N + OccNet · Pose ablationAP-0.5 (2D): 0.499 -> 0.495Hurt
use SMPLx body shape parameters (ablation)PersonaCraft · Body shape preservation (total)cosine similarity: 0.539 -> 0.615Improved
replace OMG-based InstantID with PersonaCraftInstantID + OMG · Multi-human personalizationFace ID preservation (total): 0.227 -> 0.317Improved
replace 2D pose ControlNet-SDXL with 3D SMPLx conditioningControlNet-SDXL · Pose-controlled generationMPJPE (3D): 114.93 -> 62.647Improved
fine-tune ControlNet-SDXL baseline on MPII data (ablation)ControlNet-SDXL · Pose-controlled generationMPJPE (3D): 114.93 -> 100.817Improved
replace ControlNet-SDXL with PersonaCraft on text alignmentControlNet-SDXL · Pose-controlled generationCLIP sim: 0.285 -> 0.274Hurt

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