Research map / Video / Video generation
V-RGBX: Video Editing with Accurate Controls over Intrinsic Properties
Ye Fang, Tong Wu, Valentin Deschaintre, Duygu Ceylan, Iliyan Georgiev, Chun-Hao Paul Huang, Yiwei Hu, Xuelin Chen, Tuanfeng Yang Wang
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 |
|---|---|---|---|
| add Intrinsic Type Embedding (ITE) (ablation) | V-RGBX forward renderer · X-to-RGB (Evermotion) | FVD: 532.21 -> 405.79 | Improved |
| add Intrinsic Type Embedding (ITE) (ablation) | V-RGBX forward renderer · X-to-RGB (Evermotion) | PSNR: 20.96 -> 21.47 | Improved |
| add reference condition in training on top of ITE (ablation) | V-RGBX forward renderer · X-to-RGB (Evermotion) | PSNR: 21.47 -> 22.42 | Improved |
| add reference condition in training on top of ITE (ablation) | V-RGBX forward renderer · X-to-RGB (Evermotion) | FVD: 405.79 -> 367.89 | Improved |
| add reference condition in training on top of ITE (ablation) | V-RGBX forward renderer · X-to-RGB (Evermotion) | SSIM: 0.7994 -> 0.7952 | Hurt |
| provide keyframe reference at inference (ablation) | V-RGBX · X-to-RGB (Evermotion) | FVD: 401.62 -> 367.89 | Improved |
| provide keyframe reference at inference (ablation) | V-RGBX · X-to-RGB (Evermotion) | Smoothness: 0.9814 -> 0.9805 | No clear effect |
| provide reference when albedo channel is dropped (ablation) | V-RGBX · X-to-RGB, drop albedo | PSNR: 17.18 -> 20.83 | Improved |
| provide reference when irradiance channel is dropped (ablation) | V-RGBX · X-to-RGB, drop irradiance | FVD: 702.16 -> 441.05 | Improved |
| give dropped albedo channel in first frame only (ablation) | V-RGBX · X-to-RGB, albedo | PSNR: 20.83 -> 21.65 | Improved |
| replace per-frame RGBX with V-RGBX for inverse rendering | inverse renderer · RGB-to-X albedo (Evermotion) | LPIPS: 0.2872 -> 0.2406 | Improved |
| replace DiffusionRenderer with V-RGBX for forward rendering | forward renderer · X-to-RGB (Evermotion) | PSNR: 12.66 -> 22.42 | Improved |
Closest papers by meaning
- Video Generation Models Are Inherent Lighting Estimators (ECCV 2026 · similarity 0.88)
- Light-X: Generative 4D Video Rendering with Camera and Illumination Control (ICLR 2026 · similarity 0.88)
- TransPixeler: Advancing Text-to-Video Generation with Transparency (CVPR 2025 · similarity 0.87)
- TC-Light: Temporally Coherent Generative Rendering for Realistic World Transfer (NeurIPS 2025 · similarity 0.86)
- IllumiCraft: Unified Geometry and Illumination Diffusion for Controllable Video Generation (NeurIPS 2025 · similarity 0.86)
- IntrinsiX: High-Quality PBR Generation using Image Priors (NeurIPS 2025 · similarity 0.86)
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.
