Research map / Image generation / Text-to-image diffusion
UniversalBooth: Model-Agnostic Personalized Text-to-Image Generation
Songhua Liu, Ruonan Yu, Xinchao Wang
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 square and shared key-value mappings (vs identity/no KV mappings) (ablation) | UniversalBooth · Small Diffusion (unseen) | DINO-I: .424 -> .482 | Improved |
| add square and shared key-value mappings (vs identity/no KV mappings) (ablation) | UniversalBooth · Seen architecture (SD-v1.4) | DINO-I: .650 -> .667 | Mixed |
| share square KV mappings within a scale (vs unshared) (ablation) | UniversalBooth · Tiny Diffusion (unseen) | DINO-I: .163 -> .479 | Improved |
| share square KV mappings within a scale (vs unshared) (ablation) | UniversalBooth · Seen architecture (SD-v1.4) | DINO-I: .683 -> .667 | Hurt |
| add hierarchical cross-attention (ablation) | UniversalBooth · Small Diffusion (unseen) | DINO-I: .291 -> .482 | Improved |
| add hierarchical cross-attention (ablation) | UniversalBooth · Seen architecture (SD-v1.4) | DINO-I: .629 -> .667 | Improved |
| add optimal transport prior loss (ablation) | UniversalBooth · Tiny Diffusion (unseen) | DINO-I: .291 -> .479 | Improved |
| add optimal transport prior loss (ablation) | UniversalBooth · Seen architecture (SD-v1.4) | DINO-I: .584 -> .667 | Improved |
| add optimal transport prior loss (ablation) | UniversalBooth · Small Diffusion (unseen) | CLIP-I: .644 -> .687 | Improved |
| add hierarchical cross-attention (ablation) | UniversalBooth · Tiny Diffusion (unseen) | CLIP-I: .613 -> .675 | Improved |
| use UniversalBooth instead of ELITE for cross-architecture transfer | SD-v1.4-trained encoder · SD-XL (small-to-large) | DINO-I: .336 -> .580 | Improved |
| use UniversalBooth instead of ELITE for cross-architecture transfer | SD-v1.4-trained encoder · SD-3-Medium (small-to-large) | DINO-I: .446 -> .581 | 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.
- ELITE — ELITE: Encoding Visual Concepts into Textual Embeddings for Customized Text-to-Image Generation (ICCV 2023) · better on 8 of 9 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
- InstantBooth: Personalized Text-to-Image Generation without Test-Time Finetuning (CVPR 2024 · similarity 0.90)
- JeDi: Joint-Image Diffusion Models for Finetuning-Free Personalized Text-to-Image Generation (CVPR 2024 · similarity 0.90)
- MS-Diffusion: Multi-subject Zero-shot Image Personalization with Layout Guidance (ICLR 2025 · similarity 0.89)
- PaRa: Personalizing Text-to-Image Diffusion via Parameter Rank Reduction (ICLR 2025 · similarity 0.89)
- AttnDreamBooth: Towards Text-Aligned Personalized Text-to-Image Generation (NeurIPS 2024 · similarity 0.89)
- R&B: Region and Boundary Aware Zero-shot Grounded Text-to-image Generation (ICLR 2024 · similarity 0.89)
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.
