6 papers
Learn Once, Edit Anywhere: Visual Direction Transfer for Diffusion Models
Yusuf Dalva, Hidir Yesiltepe, Pinar Yanardag
The rapid advancement of diffusion models has enabled the generation of high-fidelity images from textual prompts, yet achieving precise, disentangled control over specific attribu…
AdaState: Self-Evolving Anchors for Streaming Video Generation
Yusuf Dalva, Pinar Yanardag
Autoregressive video diffusion models generate streaming video by producing frames sequentially, conditioning each chunk on previously generated content. These models are structura…
RAVEL: Rare Concept Generation and Editing via Graph-driven Relational Guidance
Kavana Venkatesh, Yusuf Dalva, Ismini Lourentzou +1
Despite impressive visual fidelity, current text-to-image (T2I) diffusion models struggle to depict rare, complex, or culturally nuanced concepts due to training data limitations.…
LoRAShop: Training-Free Multi-Concept Image Generation and Editing with Rectified Flow Transformers
Yusuf Dalva, Hidir Yesiltepe, Pinar Yanardag
We introduce LoRAShop, the first framework for multi-concept image editing with LoRA models. LoRAShop builds on a key observation about the feature interaction patterns inside Flux…
FluxSpace: Disentangled Semantic Editing in Rectified Flow Transformers
Yusuf Dalva, Kavana Venkatesh, Pinar Yanardag
Rectified flow models have emerged as a dominant approach in image generation, showcasing impressive capabilities in high-quality image synthesis. However, despite their effectiven…
LayerFusion: Harmonized Multi-Layer Text-to-Image Generation with Generative Priors
Yusuf Dalva, Yijun Li, Qing Liu +4
Large-scale diffusion models have achieved remarkable success in generating high-quality images from textual descriptions, gaining popularity across various applications. However,…