Research map / Image generation / Editing and style transfer
PhotoAgent: Exploratory Visual Aesthetic Planning with Large Vision Models
Mingde Yao, Zhiyuan You, King Man Tam, Menglu Wang, Tianfan 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 |
|---|---|---|---|
| add the UGC reward-model evaluator (ablation) | PhotoAgent · UGC-Edit test set | Laion-Reward | Improved |
| run more MCTS simulations than 10 (ablation) | PhotoAgent · UGC-Edit test set | aesthetic metrics | Improved |
| use MCTS search depth greater than 1 instead of greedy selection (ablation) | PhotoAgent · multi-step edits | aesthetic metrics | Improved |
| add MCTS planning to closed-loop ReAct-style editing | PhotoAgent · UGC-Edit test set | BRISQUE: 0.6422 -> 0.6217 | Improved |
| add MCTS planning to closed-loop ReAct-style editing | PhotoAgent · UGC-Edit test set | CLIP Similarity: 0.6027 -> 0.6254 | Improved |
| close the loop with evaluator feedback in ReAct | ReAct · UGC-Edit test set | UGC Score: 3.175 -> 3.258 | Improved |
| use GPT-4o single-step editing instead of PhotoAgent | GPT-4o · UGC-Edit test set | UGC Score: 4.210 vs 4.176 | Mixed |
| select top-K actions and evaluate real outputs instead of the single best simulated action | PhotoAgent executor · UGC-Edit test set | robustness | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- InstructPix2Pix — InstructPix2Pix: Learning To Follow Image Editing Instructions (CVPR 2023) · better on 4 of 4 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
- JarvisArt: Liberating Human Artistic Creativity via an Intelligent Photo Retouching Agent (NeurIPS 2025 · similarity 0.88)
- RetouchIQ: MLLM Agents for Instruction-Based Image Retouching with Generalist Reward (CVPR 2026 · similarity 0.87)
- VisionDirector: Vision-Language Guided Closed-Loop Refinement for Generative Image Synthesis (CVPR 2026 · similarity 0.86)
- UltraEdit: Instruction-based Fine-Grained Image Editing at Scale (NeurIPS 2024 · similarity 0.86)
- Guiding Instruction-based Image Editing via Multimodal Large Language Models (ICLR 2024 · similarity 0.86)
- EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing (ICLR 2026 · 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.
