Research map / Image generation / Text-to-image diffusion
Draw Your Mind: Personalized Generation via Condition-Level Modeling in Text-to-Image Diffusion Models
Hyungjin Kim, Seokho Ahn, Young-Duk Seo
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 guidance mechanism G (ablation) | DrUM · PIP, Stable Diffusion V1 | CLIP score (target): 19.04 -> 25.13 | Improved |
| add the guidance mechanism G (ablation) | DrUM · ML, Stable Diffusion V1 | CLIP score (target): 22.74 -> 29.49 | Improved |
| add coreset sampling S (ablation) | DrUM · PIP, Stable Diffusion V1 | CLIP score (target): 25.49 -> 25.13 | No clear effect |
| add coreset sampling S (ablation) | DrUM · ML, Stable Diffusion V1 | CLIP score (target): 29.56 -> 29.49 | No clear effect |
| add sampling and guidance together (ablation) | DrUM · PIP, Stable Diffusion V1 | CLIP score (target): 18.13 -> 25.13 | Improved |
| use DrUM instead of TV | personalized generation · PIP, Stable Diffusion V1 | CLIP score improvement (%): +22.40% -> +34.83% | Improved |
| use DrUM instead of TV | personalized generation · ML, Stable Diffusion V3 | CLIP score improvement (%): DrUM -5.60% vs TV -4.47% | Hurt |
| use DrUM instead of TV and PMG | personalized generation · ML, OpenCLIP ViT-L | Text align improvement (%): -12.42% (TV) -> +3.13% | Improved |
| use coreset sampling instead of random or uniform sampling | user profiling · PIP/ML, OpenCLIP ViT-L | Text align | Improved |
| use DrUM on ML dataset with complex product information | DrUM · ML (Stable Diffusion V3) | CLIP score: +4.26% (SDXL) vs -5.60% (SD V3) | Hurt |
Closest papers by meaning
- JeDi: Joint-Image Diffusion Models for Finetuning-Free Personalized Text-to-Image Generation (CVPR 2024 · similarity 0.89)
- PaRa: Personalizing Text-to-Image Diffusion via Parameter Rank Reduction (ICLR 2025 · similarity 0.89)
- Hollowed Net for On-Device Personalization of Text-to-Image Diffusion Models (NeurIPS 2024 · similarity 0.88)
- Minority-Focused Text-to-Image Generation via Prompt Optimization (CVPR 2025 · similarity 0.88)
- T2I-Adapter: Learning Adapters to Dig Out More Controllable Ability for Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.87)
- Mining your own secrets: Diffusion Classifier Scores for Continual Personalization of Text-to-Image Diffusion Models (ICLR 2025 · similarity 0.87)
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.
