Research map / Image generation / Text-to-image diffusion
A-STAR: Test-time Attention Segregation and Retention for Text-to-image Synthesis
Aishwarya Agarwal, Srikrishna Karanam, K J Joseph, Apoorv Saxena, Koustava Goswami, Balaji Vasan Srinivasan
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 attention retention loss (ablation) | Stable Diffusion · Animal-Object prompts | text-text similarity: 0.78 -> 0.83 | Improved |
| add attention segregation loss (ablation) | Stable Diffusion · Animal-Animal prompts | text-text similarity: 0.76 -> 0.79 | Improved |
| add both segregation and retention losses (ablation) | Stable Diffusion · Animal-Animal prompts | text-text similarity: 0.76 -> 0.82 | Improved |
| add both losses (ablation) | Stable Diffusion · Object-Object prompts | text-text similarity: 0.77 -> 0.82 | Improved |
| A-STAR instead of Attend-and-Excite on top of Stable Diffusion | Stable Diffusion · Animal-Animal prompts | text-text similarity: 0.80 -> 0.82 | Improved |
| A-STAR vs Stable Diffusion on CLIP similarity | Stable Diffusion · all three prompt categories | CLIP full prompt / min object similarity: +7.1% and +10.8% | Improved |
| Composable Diffusion instead of plain Stable Diffusion | Stable Diffusion · Animal-Animal prompts | text-text similarity: 0.76 -> 0.69 | Hurt |
| use A-STAR vs baselines in a user study | Stable Diffusion · User study, animal-animal | preference rate: 2.2% -> 94.8% | Improved |
| apply losses to prompts where baseline already succeeds | Stable Diffusion · qualitative, easy prompts | semantic faithfulness | No clear effect |
Later papers that compared with this method
1 later accepted paper reports A-STAR as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization (CVPR 2024 · better on 1 of 1 shared result)
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
- CONFORM: Contrast is All You Need for High-Fidelity Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.91)
- R&B: Region and Boundary Aware Zero-shot Grounded Text-to-image Generation (ICLR 2024 · similarity 0.90)
- The Hidden Language of Diffusion Models (ICLR 2024 · similarity 0.90)
- Compositional Text-to-Image Synthesis with Attention Map Control of Diffusion Models (AAAI 2024 · similarity 0.90)
- Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.90)
- Towards Understanding Cross and Self-Attention in Stable Diffusion for Text-Guided Image Editing (CVPR 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.
