English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Enhancing a low-light noisy RAW image into a well-exposed and clean sRGBimage is a significant challenge in computational photography. Due to thelimitation of large-scale paired data, prior approaches have difficulty inrecovering fine details and true colors in extremely low-light regions.Meanwhile, recent advancements in generative diffusion models have shownpromising generating capabilities, which inspires this work to exploregenerative priors from a diffusion model trained on a large-scale open-domaindataset to benefit the low-light image enhancement (LLIE) task. Based on thisintention, we propose a novel diffusion-model-based LLIE method, dubbedLDM-SID. LDM-SID aims at inserting a set of proposed taming modules into afrozen pre-trained diffusion model to steer its generating process.Specifically, the taming module fed with low-light information serves to outputa pair of affine transformation parameters to modulate the intermediate featurein the diffusion model. Additionally, based on the observation of dedicatedgenerative priors across different portions of the diffusion model, we proposeto apply 2D discrete wavelet transforms on the input RAW image, resulting individing the LLIE task into two essential parts: low-frequency contentgeneration and high-frequency detail maintenance. This enables us to skillfullytame the diffusion model for optimized structural generation and detailenhancement. Extensive experiments demonstrate the proposed method not onlyachieves state-of-the-art performance in quantitative evaluations but alsoshows significant superiority in visual comparisons. These findings highlightthe effectiveness of leveraging a pre-trained diffusion model as a generativeprior to the LLIE task. The project page is available athttps://csqiangwen.github.io/projects/ldm-sid/

Taming Latent Diffusion Models to See in the Dark