Litcius/Paper detail

ArtAdapter: Text-to-Image Style Transfer using Multi-Level Style Encoder and Explicit Adaptation

Dar-Yen Chen, Hamish Tennent, Ching-Wen Hsu

202427 citationsDOI

Abstract

This work introduces ArtAdapter, a transformative text-to-image (T2I) style transfer framework that transcends tra-ditional limitations of color, brushstrokes, and object shape, capturing high-level style elements such as composition and distinctive artistic expression. The integration of a multi-level style encoder with our proposed explicit adaptation mechanism enables ArtAdapter to achieve unprecedented fidelity in style transfer, ensuring close alignment with tex-tual descriptions. Additionally, the incorporation of an Aux-iliary Content Adapter (ACA) effectively separates content from style, alleviating the borrowing of content from style references. Moreover, our novel fast finetuning approach could further enhance zero-shot style representation while mitigating the risk of overfitting. Comprehensive evaluations confirm that ArtAdapter surpasses current state-of-the-art methods.

Topics & Concepts

Style (visual arts)Computer scienceAdaptation (eye)EncoderTransfer (computing)Image (mathematics)Artificial intelligenceNatural language processingComputer visionPsychologyArtVisual artsParallel computingOperating systemNeuroscienceGenerative Adversarial Networks and Image SynthesisHandwritten Text Recognition TechniquesMusic and Audio Processing
ArtAdapter: Text-to-Image Style Transfer using Multi-Level Style Encoder and Explicit Adaptation | Litcius