Research map / Image generation / Text-to-image diffusion
VisualPrompter: Semantic-Aware Prompt Optimization with Visual Feedback for Text-to-Image Synthesis
Shiyu Wu, Mingzhen Sun, Weining Wang, Yequan Wang, Jing Liu
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 visual-feedback self-reflection and DSG-based concept expansion | VisualPrompter · DSG-1k and TIFA, average over 4 generators | semantic accuracy: 78.4 -> 83.0 | Improved |
| replace DSG-based concept expansion with direct LLM (Qwen) rewriting (ablation) | VisualPrompter · DSG benchmark | semantic accuracy / aesthetic: 77.0 / 5.73 -> 73.0 / 5.45 | Hurt |
| direct Qwen prompt optimization without visual feedback (ablation) | original prompts · DSG benchmark | semantic accuracy: 72.1 -> 67.8 | Hurt |
| add visual-feedback self-reflection to DSG prompt modification (ablation) | VisualPrompter · DSG benchmark | semantic accuracy: 71.9 -> 77.0 | Improved |
| regenerate defective images without modifying prompts (ablation) | original prompts · DSG benchmark | semantic accuracy: 72.1 -> 73.8 | Improved |
| replace the LLM decorator with NeuroPrompts (ablation) | VisualPrompter · DSG benchmark | Aesthetic Score: 5.81 -> 6.24 | Improved |
| replace the LLM decorator with NeuroPrompts (ablation) | VisualPrompter · DSG benchmark | semantic consistency | Hurt |
| apply prior prompt engineers (Promptist, BeautifulPrompt) to generators other than SD1.5 | Promptist, BeautifulPrompt · DSG benchmark | CLIP Score: Promptist Flux 31.58 -> 31.49 | Hurt |
| use VisualPrompter instead of original prompts on Flux-dev | Flux-dev · DSG benchmark | CLIP Score: 31.58 -> 33.81 | Improved |
| add VisualPrompter on top of generator | VisualPrompter · inference time, SD v1.5 | seconds: 7.0s (Promptist) vs 10.7s | Hurt |
Later papers that compared with this method
1 later accepted paper reports VisualPrompter as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- From Structural Feedback to Prompt Policies: Learning Faithful Text-to-Image Prompt Editors (NeurIPS 2026 · better on 2 of 3 shared results)
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
- Imagine Before You Draw: Visual Prompt Engineering for Image Generation (NeurIPS 2026 · similarity 0.92)
- A User-Friendly Framework for Generating Model-Preferred Prompts in Text-to-Image Synthesis (AAAI 2024 · similarity 0.92)
- Tailored Visions: Enhancing Text-to-Image Generation with Personalized Prompt Rewriting (CVPR 2024 · similarity 0.91)
- Visually Guided Decoding: Gradient-Free Hard Prompt Inversion with Language Models (ICLR 2025 · similarity 0.90)
- PromptCoT: Align Prompt Distribution via Adapted Chain-of-Thought (CVPR 2024 · similarity 0.90)
- Textualize Visual Prompt for Image Editing via Diffusion Bridge (AAAI 2025 · 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.
