Guiding Diffusion with Deep Geometric Moments: Balancing Fidelity and Variation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908369318576128 |
|---|---|
| author | Jung, Sangmin Nath, Utkarsh Yang, Yezhou Pedrielli, Giulia Biswas, Joydeep Zhang, Amy Ghasemzadeh, Hassan Turaga, Pavan |
| author_facet | Jung, Sangmin Nath, Utkarsh Yang, Yezhou Pedrielli, Giulia Biswas, Joydeep Zhang, Amy Ghasemzadeh, Hassan Turaga, Pavan |
| contents | Text-to-image generation models have achieved remarkable capabilities in synthesizing images, but often struggle to provide fine-grained control over the output. Existing guidance approaches, such as segmentation maps and depth maps, introduce spatial rigidity that restricts the inherent diversity of diffusion models. In this work, we introduce Deep Geometric Moments (DGM) as a novel form of guidance that encapsulates the subject's visual features and nuances through a learned geometric prior. DGMs focus specifically on the subject itself compared to DINO or CLIP features, which suffer from overemphasis on global image features or semantics. Unlike ResNets, which are sensitive to pixel-wise perturbations, DGMs rely on robust geometric moments. Our experiments demonstrate that DGM effectively balance control and diversity in diffusion-based image generation, allowing a flexible control mechanism for steering the diffusion process. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_12486 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Guiding Diffusion with Deep Geometric Moments: Balancing Fidelity and Variation Jung, Sangmin Nath, Utkarsh Yang, Yezhou Pedrielli, Giulia Biswas, Joydeep Zhang, Amy Ghasemzadeh, Hassan Turaga, Pavan Computer Vision and Pattern Recognition Text-to-image generation models have achieved remarkable capabilities in synthesizing images, but often struggle to provide fine-grained control over the output. Existing guidance approaches, such as segmentation maps and depth maps, introduce spatial rigidity that restricts the inherent diversity of diffusion models. In this work, we introduce Deep Geometric Moments (DGM) as a novel form of guidance that encapsulates the subject's visual features and nuances through a learned geometric prior. DGMs focus specifically on the subject itself compared to DINO or CLIP features, which suffer from overemphasis on global image features or semantics. Unlike ResNets, which are sensitive to pixel-wise perturbations, DGMs rely on robust geometric moments. Our experiments demonstrate that DGM effectively balance control and diversity in diffusion-based image generation, allowing a flexible control mechanism for steering the diffusion process. |
| title | Guiding Diffusion with Deep Geometric Moments: Balancing Fidelity and Variation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.12486 |