Research map / Image generation / Text-to-image diffusion
DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Fine-tune the diffusion model to bind a unique identifier to the subject, using class-specific prompts and prior preservation | Stable Diffusion · Subject-driven generation dataset | DINO: 0.569 -> 0.668 | Helped |
| Fine-tune the diffusion model to bind a unique identifier to the subject, using class-specific prompts and prior preservation | Stable Diffusion · Subject-driven generation dataset | CLIP-T: 0.255 -> 0.305 | Helped |
| Add prior preservation loss (ablation) | DreamBooth (Imagen) · Subject-driven generation dataset (15 subjects) | PRES: 0.664 -> 0.493 | Helped |
| Add prior preservation loss (ablation) | DreamBooth (Imagen) · Subject-driven generation dataset (15 subjects) | DIV: 0.371 -> 0.391 | Helped |
| Add prior preservation loss (ablation) | DreamBooth (Imagen) · Subject-driven generation dataset (15 subjects) | DINO: 0.712 -> 0.684 | Hurt |
| Add prior preservation loss (ablation) | DreamBooth (Imagen) · Subject-driven generation dataset (15 subjects) | CLIP-T: 0.306 -> 0.308 | Helped |
| Use the correct class noun in the training prompt (ablation) | DreamBooth (Imagen) · Subject fidelity class-name ablation | DINO: 0.303 -> 0.744 | Helped |
| Use the correct class noun instead of an incorrect one (ablation) | DreamBooth (Imagen) · Subject fidelity class-name ablation | DINO: 0.454 -> 0.744 | Helped |
| Increase the number of input images from 1 to 4 (ablation) | DreamBooth (Imagen), backpack · Backpack personalization | DINO: 0.494 -> 0.604 | Helped |
| Increase the number of input images from 1 to 4 (ablation) | DreamBooth (Imagen), dog · Dog personalization | DINO: 0.798 -> 0.876 | Helped |
| Reduce super-resolution fine-tuning noise augmentation from 10^{-3} to 10^{-5} (ablation) | Imagen 256×256 super-resolution model · Personalized subject generation | subject detail fidelity: 10^{-3} -> 10^{-5} | Helped |
Closest papers by meaning
- DreamDistribution: Learning Prompt Distribution for Diverse In-distribution Generation (ICLR 2025 · similarity 0.90)
- Subject-driven Text-to-Image Generation via Preference-based Reinforcement Learning (NeurIPS 2024 · similarity 0.89)
- AttnDreamBooth: Towards Text-Aligned Personalized Text-to-Image Generation (NeurIPS 2024 · similarity 0.89)
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.88)
- Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.88)
- OneActor: Consistent Subject Generation via Cluster-Conditioned Guidance (NeurIPS 2024 · 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.