Research map / Image generation / Text-to-image diffusion
Modeling Artistic Workflows for Image Generation and Editing
Hung-Yu Tseng, Matthew Fisher, Jingwan Lu, Yijun Li, Vladimir Kim, Ming-Hsuan Yang
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 AdaIN optimization (ablation) | artwork generation module · Face drawing reconstruction | l1 reconstruction error: 0.094 -> 0.040 | Improved |
| add AdaIN optimization (ablation) | artwork generation module · Chair design reconstruction | FID: 129.2 -> 46.48 | Improved |
| optimize latent z instead of AdaIN parameters (ablation) | artwork generation module · Face drawing reconstruction | l1 reconstruction error: 0.094 -> 0.104 | Hurt |
| add AdaIN optimization (no regularization) (ablation) | artwork generation module · Face drawing editing | FID: 38.68 -> 44.28 | Hurt |
| add AdaIN optimization (no regularization) (ablation) | artwork generation module · Chair design editing | FID: 128.4 -> 97.90 | Improved |
| add strong weight decay (w=1e-2) to AdaIN optimization (ablation) | AdaIN optimization · Anime drawing reconstruction | l1 reconstruction error: 0.042 -> 0.097 | Hurt |
| add strong weight decay (w=1e-2) to AdaIN optimization (ablation) | AdaIN optimization · Face drawing editing | FID: 44.28 -> 38.57 | Improved |
| add learned regularization to AdaIN optimization (ablation) | AdaIN optimization · Face drawing editing | FID: 44.28 -> 39.40 | Improved |
| add learned regularization to AdaIN optimization (ablation) | AdaIN optimization · Face drawing reconstruction | l1 reconstruction error: 0.040 -> 0.045 | No clear effect |
| add learned regularization to AdaIN optimization (ablation) | AdaIN optimization · Anime drawing editing | FID: 37.40 -> 35.73 | Improved |
| add learned regularization to AdaIN optimization (ablation) | AdaIN optimization · Anime drawing reconstruction | l1 reconstruction error: 0.042 -> 0.070 | Hurt |
| add learned regularization instead of fixed weight decay 1e-3 (ablation) | AdaIN optimization · Chair design editing | FID: 91.68 -> 95.25 | Hurt |
Closest papers by meaning
- End-to-End Visual Editing with a Generatively Pre-trained Artist (ECCV 2022 · similarity 0.87)
- EditAR: Unified Conditional Generation with Autoregressive Models (CVPR 2025 · similarity 0.85)
- Pathways on the Image Manifold: Image Editing via Video Generation (CVPR 2025 · similarity 0.85)
- FluxSpace: Disentangled Semantic Editing in Rectified Flow Models (CVPR 2025 · similarity 0.85)
- CREA: A Collaborative Multi-Agent Framework for Creative Image Editing and Generation (NeurIPS 2025 · similarity 0.85)
- Learning to Manipulate Artistic Images (AAAI 2024 · similarity 0.84)
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.
