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

Recent advances in text-to-image diffusion models have enabled thephotorealistic generation of images from text prompts. Despite the greatprogress, existing models still struggle to generate compositionalmulti-concept images naturally, limiting their ability to visualize humanimagination. While several recent works have attempted to address this issue,they either introduce additional training or adopt guidance at inference time.In this work, we consider a more ambitious goal: natural multi-conceptgeneration using a pre-trained diffusion model, and with almost no extra cost.To achieve this goal, we identify the limitations in the text embeddings usedfor the pre-trained text-to-image diffusion models. Specifically, we observeconcept dominance and non-localized contribution that severely degrademulti-concept generation performance. We further design a minimal low-costsolution that overcomes the above issues by tweaking (not re-training) the textembeddings for more realistic multi-concept text-to-image generation. OurCorrection by Similarities method tweaks the embedding of concepts bycollecting semantic features from most similar tokens to localize thecontribution. To avoid mixing features of concepts, we also apply Cross-TokenNon-Maximum Suppression, which excludes the overlap of contributions fromdifferent concepts. Experiments show that our approach outperforms previousmethods in text-to-image, image manipulation, and personalization tasks,despite not introducing additional training or inference costs to the diffusionsteps.

Multi-Concept T2I-Zero: Tweaking Only The Text Embeddings and Nothing Else