DreamMatcher: Appearance Matching Self-Attention for Semantically-Consistent Text-to-Image Personalization
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913325728661504 |
|---|---|
| author | Nam, Jisu Kim, Heesu Lee, DongJae Jin, Siyoon Kim, Seungryong Chang, Seunggyu |
| author_facet | Nam, Jisu Kim, Heesu Lee, DongJae Jin, Siyoon Kim, Seungryong Chang, Seunggyu |
| contents | The objective of text-to-image (T2I) personalization is to customize a diffusion model to a user-provided reference concept, generating diverse images of the concept aligned with the target prompts. Conventional methods representing the reference concepts using unique text embeddings often fail to accurately mimic the appearance of the reference. To address this, one solution may be explicitly conditioning the reference images into the target denoising process, known as key-value replacement. However, prior works are constrained to local editing since they disrupt the structure path of the pre-trained T2I model. To overcome this, we propose a novel plug-in method, called DreamMatcher, which reformulates T2I personalization as semantic matching. Specifically, DreamMatcher replaces the target values with reference values aligned by semantic matching, while leaving the structure path unchanged to preserve the versatile capability of pre-trained T2I models for generating diverse structures. We also introduce a semantic-consistent masking strategy to isolate the personalized concept from irrelevant regions introduced by the target prompts. Compatible with existing T2I models, DreamMatcher shows significant improvements in complex scenarios. Intensive analyses demonstrate the effectiveness of our approach. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_09812 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | DreamMatcher: Appearance Matching Self-Attention for Semantically-Consistent Text-to-Image Personalization Nam, Jisu Kim, Heesu Lee, DongJae Jin, Siyoon Kim, Seungryong Chang, Seunggyu Computer Vision and Pattern Recognition The objective of text-to-image (T2I) personalization is to customize a diffusion model to a user-provided reference concept, generating diverse images of the concept aligned with the target prompts. Conventional methods representing the reference concepts using unique text embeddings often fail to accurately mimic the appearance of the reference. To address this, one solution may be explicitly conditioning the reference images into the target denoising process, known as key-value replacement. However, prior works are constrained to local editing since they disrupt the structure path of the pre-trained T2I model. To overcome this, we propose a novel plug-in method, called DreamMatcher, which reformulates T2I personalization as semantic matching. Specifically, DreamMatcher replaces the target values with reference values aligned by semantic matching, while leaving the structure path unchanged to preserve the versatile capability of pre-trained T2I models for generating diverse structures. We also introduce a semantic-consistent masking strategy to isolate the personalized concept from irrelevant regions introduced by the target prompts. Compatible with existing T2I models, DreamMatcher shows significant improvements in complex scenarios. Intensive analyses demonstrate the effectiveness of our approach. |
| title | DreamMatcher: Appearance Matching Self-Attention for Semantically-Consistent Text-to-Image Personalization |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2402.09812 |