Research map / 3D vision / 3D generation
Vox-E: Text-Guided Voxel Editing of 3D Objects
Etai Sella, Gal Fiebelman, Peter Hedman, Hadar Averbuch-Elor
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace image-space regularization with volumetric density-correlation regularization (L_reg3D) (ablation) | Vox-E · Synthetic objects, local+global edits | CLIP_Sim / FID_Rec: CLIP_Sim 0.29 -> 0.37; FID_Rec 367.53 -> 240.37 | Improved |
| add 3D cross-attention spatial refinement (SR) (ablation) | Vox-E · Synthetic objects, local edits | FID_Rec: 240.37 -> 119.44 | Improved |
| add 3D cross-attention spatial refinement (SR) (ablation) | Vox-E · Synthetic objects, local edits | CLIP_Sim: 0.37 -> 0.36 | Hurt |
| add spatial refinement (ablation) | Vox-E · Synthetic objects, local edits | CLIP_Dir: 0.08 -> 0.06 | Hurt |
| use volumetric correlation instead of volumetric L2 on density features (ablation) | Vox-E · Synthetic objects | CLIP_Dir: 0.05 -> 0.06 | Improved |
| also regularize color features (L_reg3D++) (ablation) | Vox-E · Synthetic objects | CLIP_Dir: 0.06 -> 0.02 | Hurt |
| Vox-E vs Text2Mesh on CLIP direction similarity | Vox-E · Synthetic objects, local edits | CLIP_Dir: 0.08 -> 0.07 | Hurt |
| use view-dependent color via spherical harmonics (ablation) | Vox-E · Qualitative | edit quality | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Text2Mesh — Text2Mesh: Text-Driven Neural Stylization for Meshes (CVPR 2022) · better on 0 of 4 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
- Let 2D Diffusion Model Know 3D-Consistency for Robust Text-to-3D Generation (ICLR 2024 · similarity 0.89)
- Diffusion Handles Enabling 3D Edits for Diffusion Models by Lifting Activations to 3D (CVPR 2024 · similarity 0.89)
- ED-NeRF: Efficient Text-Guided Editing of 3D Scene With Latent Space NeRF (ICLR 2024 · similarity 0.89)
- A3D: Does Diffusion Dream about 3D Alignment? (ICLR 2025 · similarity 0.88)
- SceneFactor: Factored Latent 3D Diffusion for Controllable 3D Scene Generation (CVPR 2025 · similarity 0.88)
- LT3SD: Latent Trees for 3D Scene Diffusion (CVPR 2025 · similarity 0.88)
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.
