Research map / Image generation / Text-to-image diffusion
Training-free Diffusion Model Adaptation for Variable-Sized Text-to-Image Synthesis
Zhiyu Jin, Xuli Shen, Bin Li, Xiangyang Xue
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 proposed attention scaling factor instead of 1/sqrt(d) | Stable Diffusion · 224*224 LAION-5B | FID: 74.5742 -> 41.8925 | Improved |
| use proposed attention scaling factor | Stable Diffusion · 448*448 | FID: 19.9039 -> 19.4923 | Improved |
| use proposed attention scaling factor | Stable Diffusion · 768*768 | FID: 29.5974 -> 28.1372 | Improved |
| use proposed attention scaling factor | Latent Diffusion · 128*128 LAION-400M | FID: 65.2988 -> 57.5542 | Improved |
| use proposed attention scaling factor | Latent Diffusion · 384*384 | FID: 20.5466 -> 20.3842 | Improved |
| use proposed attention scaling factor | Stable Diffusion · 224*224 | CLIP score: 0.2553 -> 0.2764 | Improved |
| use proposed attention scaling factor | Stable Diffusion · 448*448 | CLIP score: 0.3176 -> 0.3180 | No clear effect |
| use proposed attention scaling factor | Latent Diffusion · 224*224 | CLIP score: 0.3094 -> 0.3096 | No clear effect |
| use proposed attention scaling factor | Stable Diffusion · 224*224 human eval | consistency score: 5.13 -> 6.77 | Improved |
| use proposed attention scaling factor | Stable Diffusion · 768*768 human eval | consistency score: 6.74 -> 7.56 | Improved |
| use proposed scaling factor instead of MultiDiffusion | Stable Diffusion · 224*224 FID-4K | FID: 154.6925 -> 50.3515 | Improved |
| use proposed scaling factor instead of MultiDiffusion | Stable Diffusion · 768*768 FID-4K | FID: 40.9270 -> 28.1372 | Improved |
Closest papers by meaning
- Adapting Diffusion Models for Improved Prompt Compliance and Controllable Image Synthesis (NeurIPS 2024 · similarity 0.90)
- On the Scalability of Diffusion-based Text-to-Image Generation (CVPR 2024 · similarity 0.90)
- Any-Size-Diffusion: Toward Efficient Text-Driven Synthesis for Any-Size HD Images (AAAI 2024 · similarity 0.90)
- SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis (ICLR 2024 · similarity 0.90)
- ScaleDiff: Higher-Resolution Image Synthesis via Efficient and Model-Agnostic Diffusion (NeurIPS 2025 · similarity 0.90)
- ScaleCrafter: Tuning-free Higher-Resolution Visual Generation with Diffusion Models (ICLR 2024 · similarity 0.90)
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.
