Research map / Image generation / Text-to-image diffusion
Image Generation from Contextually-Contradictory Prompts
Saar Huberman, Or Patashnik, Omer Dahary, Ron Mokady, Daniel Cohen-Or
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 stage-aware proxy prompts (SAP) | FLUX.1-dev · Whoops-Hard | GPT-4o alignment: 44.3 -> 62.13 | Improved |
| add stage-aware proxy prompts (SAP) | FLUX.1-dev · ContraBench | GPT-4o alignment: 57.16 -> 66.16 | Improved |
| add stage-aware proxy prompts (SAP) | SD3.0 · Whoops-Hard | GPT-4o alignment: 55.73 -> 64.40 | Improved |
| add stage-aware proxy prompts (SAP) | FLUX.1-dev · Whoops | GPT-4o alignment: 78.85 -> 85.10 | Improved |
| add in-context examples to LLM decomposition (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 48.0 -> 62.13 | Improved |
| add explanation (reasoning) requirement for LLM (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 46.46 -> 62.13 | Improved |
| allow three proxies instead of two (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 60.26 -> 62.13 | Improved |
| use proxy prompts vs static prompt (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 44.3 -> 62.13 | Improved |
| shift LLM-predicted interval boundaries by up to 1 step (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 62.13 -> 62.06 | No clear effect |
| shift interval boundaries by up to 2 steps (ablation) | SAP · Whoops-Hard | GPT-4o alignment: 62.13 -> 58.46 | Hurt |
| use R2F alternating prompts on FLUX (reimplementation) | FLUX.1-dev · Whoops | GPT-4o alignment: 78.85 -> 48.68 | Hurt |
| use Ella fine-tuned SD1.5 | FLUX.1-dev · Whoops | GPT-4o alignment: 78.85 -> 69.09 | Hurt |
Closest papers by meaning
- Prompting Hard or Hardly Prompting: Prompt Inversion for Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.89)
- PromptCoT: Align Prompt Distribution via Adapted Chain-of-Thought (CVPR 2024 · similarity 0.88)
- Devil is in the Detail: Towards Injecting Fine Details of Image Prompt in Image Generation via Conflict-free Guidance and Stratified Attention (CVPR 2025 · similarity 0.88)
- Diff-Prompt: Diffusion-Driven Prompt Generator with Mask Supervision (ICLR 2025 · similarity 0.88)
- Towards Understanding and Quantifying Uncertainty for Text-to-Image Generation (CVPR 2025 · similarity 0.88)
- From Denoising to Refining: A Corrective Framework for Vision-Language Diffusion Model (NeurIPS 2026 · similarity 0.88)
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.
