Research map / Image generation / Text-to-image diffusion
Multivariate Diffusion Transformer with Decoupled Attention for High-Fidelity Mask-Text Collaborative Facial Generation
Yushe Cao, Dianxi Shi, Xing Fu, Xuechao Zou, Haikuo Peng, Xueqi Li, Chun Yu, Junliang Xing
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 unified tokenization of mask instead of channel concatenation (ablation) | MDiTFace · MM-CelebA test | Mask accuracy (%): 88.73 -> 94.64 | Improved |
| use unified tokenization instead of channel concatenation (ablation) | MDiTFace · MM-CelebA test | IRS: 0.3699 -> 0.6993 | Improved |
| use unified tokenization instead of channel concatenation (ablation) | MDiTFace · MM-CelebA test | CMMD: 0.476 -> 0.482 | No clear effect |
| replace holistic tri-stream attention with improved decoupled attention (ablation) | MDiTFace · MM-CelebA test | Mask accuracy (%): 94.72 -> 94.64 | No clear effect |
| use hard decoupled attention instead of improved decoupled attention (ablation) | MDiTFace · MM-CelebA test | Mask accuracy (%): 94.64 -> 91.22 | Hurt |
| increase LoRA rank beyond 8 (ablation) | MDiTFace · MM-CelebA test | Mask accuracy (%): 94.64 -> 94.73 (r=32) | No clear effect |
| raise condition dropout probability p from 0.1 to 0.9 (ablation) | MDiTFace · MM-CelebA test | Mask accuracy (%): 94.64 -> 84.00 | Hurt |
| raise condition dropout probability p from 0.1 to 0.9 (ablation) | MDiTFace · MM-CelebA test | CLIP.T (%): 26.75 -> 27.02 | Mixed |
Closest papers by meaning
- Rare Text Semantics Were Always There in Your Diffusion Transformer (NeurIPS 2025 · similarity 0.87)
- Omni-Diffusion: Unified Multimodal Understanding and Generation with Masked Discrete Diffusion (ICML 2026 · similarity 0.87)
- When StyleGAN Meets Stable Diffusion: a W+ Adapter for Personalized Image Generation (CVPR 2024 · similarity 0.87)
- Dual Diffusion for Unified Image Generation and Understanding (CVPR 2025 · similarity 0.87)
- Mask^2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation (CVPR 2025 · similarity 0.87)
- Exploring the Deep Fusion of Large Language Models and Diffusion Transformers for Text-to-Image Synthesis (CVPR 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.
