REVIEW 6 cited by
Improving the Stability and Efficiency of Diffusion Models for Content Consistent Super-Resolution
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
The generative priors of pre-trained latent diffusion models (DMs) have demonstrated great potential to enhance the visual quality of image super-resolution (SR) results. However, the noise sampling process in DMs introduces randomness in the SR outputs, and the generated contents can differ a lot with different noise samples. The multi-step diffusion process can be accelerated by distilling methods, but the generative capacity is difficult to control. To address these issues, we analyze the respective advantages of DMs and generative adversarial networks (GANs) and propose to partition the generative SR process into two stages, where the DM is employed for reconstructing image structures and the GAN is employed for improving fine-grained details. Specifically, we propose a non-uniform timestep sampling strategy in the first stage. A single timestep sampling is first applied to extract the coarse information from the input image, then a few reverse steps are used to reconstruct the main structures. In the second stage, we finetune the decoder of the pre-trained variational auto-encoder by adversarial GAN training for deterministic detail enhancement. Once trained, our proposed method, namely content consistent super-resolution (CCSR),allows flexible use of different diffusion steps in the inference stage without re-training. Extensive experiments show that with 2 or even 1 diffusion step, CCSR can significantly improve the content consistency of SR outputs while keeping high perceptual quality. Codes and models can be found at \href{https://github.com/csslc/CCSR}{https://github.com/csslc/CCSR}.
Forward citations
Cited by 6 Pith papers
-
ScaleResfusion: Residual Rectified Flow based on Residual Vector Field
ScaleResfusion modifies rectified flow to start from a noisy low-quality image and learn only a residual velocity field, enabling 4-step image restoration with LoRA fine-tuning of billion-scale text-to-image models.
-
How far have we gone in Generative Image Restoration? A study on its capability, limitations and evaluation practices
Modern generative image restoration has shifted from under-generating details to over-generating them with semantic errors, revealed by multi-dimensional evaluation of 20 models across curated scenes and degradations.
-
Fine-structure Preserved Real-world Image Super-resolution via Transfer VAE Training
A transfer training scheme converts Stable Diffusion's 8x VAE into a 4x VAE that stays compatible with the pretrained UNet, improving fine-structure preservation in real-world super-resolution at lower FLOPs.
-
Optimization-Free Universal Watermark Forgery with Regenerative Diffusion Models
Extracting a watermark latent from a target image and injecting it into a pre-trained regeneration model forges watermarks onto arbitrary cover images without optimization, but only reliably for Gaussian Shading on UN...
-
Efficient Difficulty-Aware Dynamic Routing for Diffusion-Based Real-World Image Super-Resolution
DDR-SR routes each real-world low-resolution image to one of two diffusion experts based on a high-frequency-loss difficulty score, using a low-compression VAE for hard images and a high-compression VAE for easy image...
-
FiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
A one-step diffusion super-resolution model combining detail-aware loss weighting, latent residual refinement, and tunable low/high-frequency injection reports the best fidelity-detail balance among compared diffusion...
Discussion (0). Sign in to comment.