Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2503.09864 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917109081047040 |
|---|---|
| author | Laria, Héctor Gomez-Villa, Alexandra Qin, Jiang Butt, Muhammad Atif Raducanu, Bogdan Vazquez-Corral, Javier van de Weijer, Joost Wang, Kai |
| author_facet | Laria, Héctor Gomez-Villa, Alexandra Qin, Jiang Butt, Muhammad Atif Raducanu, Bogdan Vazquez-Corral, Javier van de Weijer, Joost Wang, Kai |
| contents | Recent advances in text-to-image (T2I) diffusion models have enabled remarkable control over various attributes, yet precise color specification remains a fundamental challenge. Existing approaches, such as ColorPeel, rely on model personalization, requiring additional optimization and limiting flexibility in specifying arbitrary colors. In this work, we introduce ColorWave, a novel training-free approach that achieves exact RGB-level color control in diffusion models without fine-tuning. By systematically analyzing the cross-attention mechanisms within IP-Adapter, we uncover an implicit binding between textual color descriptors and reference image features. Leveraging this insight, our method rewires these bindings to enforce precise color attribution while preserving the generative capabilities of pretrained models. Our approach maintains generation quality and diversity, outperforming prior methods in accuracy and applicability across diverse object categories. Through extensive evaluations, we demonstrate that ColorWave establishes a new paradigm for structured, color-consistent diffusion-based image synthesis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_09864 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Leveraging Semantic Attribute Binding for Free-Lunch Color Control in Diffusion Models Laria, Héctor Gomez-Villa, Alexandra Qin, Jiang Butt, Muhammad Atif Raducanu, Bogdan Vazquez-Corral, Javier van de Weijer, Joost Wang, Kai Graphics Computer Vision and Pattern Recognition Machine Learning Recent advances in text-to-image (T2I) diffusion models have enabled remarkable control over various attributes, yet precise color specification remains a fundamental challenge. Existing approaches, such as ColorPeel, rely on model personalization, requiring additional optimization and limiting flexibility in specifying arbitrary colors. In this work, we introduce ColorWave, a novel training-free approach that achieves exact RGB-level color control in diffusion models without fine-tuning. By systematically analyzing the cross-attention mechanisms within IP-Adapter, we uncover an implicit binding between textual color descriptors and reference image features. Leveraging this insight, our method rewires these bindings to enforce precise color attribution while preserving the generative capabilities of pretrained models. Our approach maintains generation quality and diversity, outperforming prior methods in accuracy and applicability across diverse object categories. Through extensive evaluations, we demonstrate that ColorWave establishes a new paradigm for structured, color-consistent diffusion-based image synthesis. |
| title | Leveraging Semantic Attribute Binding for Free-Lunch Color Control in Diffusion Models |
| topic | Graphics Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2503.09864 |