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/

Zendy Plus

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

Zendy Tools

Zendy Open

In-domain generation aims to perform a variety of tasks within a specificdomain, such as unconditional generation, text-to-image, image editing, 3Dgeneration, and more. Early research typically required training specializedgenerators for each unique task and domain, often relying on fully-labeleddata. Motivated by the powerful generative capabilities and broad applicationsof diffusion models, we are driven to explore leveraging label-free data toempower these models for in-domain generation. Fine-tuning a pre-trainedgenerative model on domain data is an intuitive but challenging way and oftenrequires complex manual hyper-parameter adjustments since the limited diversityof the training data can easily disrupt the model's original generativecapabilities. To address this challenge, we propose a guidance-decoupled priorpreservation mechanism to achieve high generative quality and controllabilityby image-only data, inspired by preserving the pre-trained model from adenoising guidance perspective. We decouple domain-related guidance from theconditional guidance used in classifier-free guidance mechanisms to preserveopen-world control guidance and unconditional guidance from the pre-trainedmodel. We further propose an efficient domain knowledge learning technique totrain an additional text-free UNet copy to predict domain guidance. Besides, wetheoretically illustrate a multi-guidance in-domain generation pipeline for avariety of generative tasks, leveraging multiple guidances from distinctdiffusion models and conditions. Extensive experiments demonstrate thesuperiority of our method in domain-specific synthesis and its compatibilitywith various diffusion-based control methods and applications.

Image is All You Need to Empower Large-scale Diffusion Models for In-Domain Generation