Research map / Image generation / Autoregressive and latent generation
ViewMask-1-to-3: Multi-View Consistent Image Generation via Multimodal Discrete Diffusion Models
Ruishu Zhu, Zhihao Huang, Jiacheng Sun, Ping Luo, Hongyuan Zhang, Xuelong Li
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 |
|---|---|---|---|
| use cosine mask schedule instead of linear (ablation) | ViewMask-1-to-3 · GSO subset | PSNR: 17.9882 -> 18.0974 | Improved |
| use cosine mask schedule instead of quadratic (ablation) | ViewMask-1-to-3 · GSO subset | LPIPS: 0.1865 -> 0.1834 | Improved |
| use cosine mask schedule instead of linear (ablation) | ViewMask-1-to-3 · GSO subset | LPIPS: 0.1882 -> 0.1834 | Improved |
| scale from 4 to 6 views | ViewMask-1-to-3 · GSO | LPIPS: 0.1497 -> 0.1405 | Improved |
| scale from 4 to 6 views | ViewMask-1-to-3 · GSO | PSNR: 20.6112 -> 20.5956 | No clear effect |
| scale from 4 to 8 views, beyond training token length | ViewMask-1-to-3 · GSO | PSNR: 20.6112 -> 20.5120 | No clear effect |
| use discrete 2D diffusion instead of continuous diffusion | ViewMask-1-to-3 vs Zero-1-to-3 XL · GSO | PSNR: 19.6839 -> 20.6112 | Improved |
| use discrete 2D diffusion instead of continuous diffusion | ViewMask-1-to-3 vs Zero-1-to-3 XL · GSO 3D reconstruction | Chamfer distance: 0.0159 -> 0.0149 | Improved |
| use discrete diffusion model instead of Zero123++ | ViewMask-1-to-3 vs Zero123++ · 3D-FUTURE | PSNR: 23.5001 -> 19.9953 | Hurt |
| use discrete diffusion unified model instead of MMaDA | ViewMask-1-to-3 vs MMaDA · GenEval | overall score: 0.63 -> 0.72 | Improved |
Closest papers by meaning
- Dual Diffusion for Unified Image Generation and Understanding (CVPR 2025 · similarity 0.90)
- Vision as a Dialect: Unifying Visual Understanding and Generation via Text-Aligned Representations (NeurIPS 2025 · similarity 0.90)
- Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokens (CVPR 2025 · similarity 0.90)
- MVDream: Multi-view Diffusion for 3D Generation (ICLR 2024 · similarity 0.90)
- ViewDiff: 3D-Consistent Image Generation with Text-to-Image Models (CVPR 2024 · similarity 0.89)
- Mask^2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation (CVPR 2025 · similarity 0.89)
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.
