Research
Edge-preserving noise for diffusion models
arXiv:2410.01540v4 Announce Type: replace Abstract: Classical diffusion models typically rely on isotropic Gaussian noise, treating all regions uniformly and overlooking structural information importa
arXiv:2410.01540v4 Announce Type: replace Abstract: Classical diffusion models typically rely on isotropic Gaussian noise, treating all regions uniformly and overlooking structural information important for high-quality generation. We introduce an edge-preserving diffusion process that generalizes isotropic models via a hybrid noise scheme with an edge-aware scheduler that smoothly transitions from edge-preserving to isotropic noise. This enables the model to capture fine structural details while generally maintaining global performance. We evaluate the impact of structure-aware noise in both diffusion and flow-matching frameworks, and show that existing isotropic models can be efficiently fine-tuned with edge-preserving noise, making our framework practical for adapting pre-trained systems. Beyond unconditional generation, our method particularly shows improvements in structure-guided tasks such as stroke-to-image synthesis, improving robustness and perceptual quality, as evidenced by consistent improvements across FID, KID, and CLIP-score.
Related
- DMin: Scalable Training Data Influence Estimation for Diffusion Models
- How Noise Benefits AI-generated Image Detection
- Visual Diffusion Models are Geometric Solvers
- Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-Resolution
- RectifiedHR: Enable Efficient High-Resolution Synthesis via Energy Rectification
Source: arXiv cs.CV | 2026-04-17