AI Paper Atlas

Research map / 3D vision / Face and shape

Material Magic Wand: Material-Aware Grouping of 3D Parts in Untextured Meshes

Umangi Jain, Vladimir Kim, Matheus Gadelha, Igor Gilitschenski, Zhiqin Chen

CVPR 2026 · 3D vision › Face and shape · approach: Optimization-based shape matching · 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
add isolated-part view (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 86.90 -> 89.74Improved
add part-with-context view (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 87.30 -> 89.74Improved
add full-object view (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 88.89 -> 89.74Improved
add context and full views to isolated-part-only input (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 86.18 -> 89.74Improved
initialize from DINO-v3 small instead of DINO-v2 large (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 86.51 -> 89.74Improved
use pretrained initialization instead of random (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 78.54 -> 89.74Improved
finetune last 5 blocks instead of 3 (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 89.74 -> 89.52No clear effect
retrieve with pre-projection embedding x instead of z (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 87.45 -> 89.74Improved
add data rebalancing (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 78.41 -> 89.74Improved
add part deduplication at inference (ablation)Material Magic Wand · Material-aware grouping benchmark (100 meshes, 241 queries)AUC: 88.13 -> 89.74Improved
scale DINO-v3 backbone from small to large (ablation)DINO-v3 baseline · Material-aware grouping benchmark (100 meshes, 241 queries)AUC PR: 81.14 -> 80.86No clear effect
train material-aware contrastive encoder vs off-the-shelf DINO-v3 smallDINO-v3 small · Material-aware grouping benchmark (100 meshes, 241 queries)F1: 59.36 -> 75.94Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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