Controllability in video generation is difficult when relying on text prompts alone. What if you could just show what you mean? We present MotionShop! With a reference video, seamlessly transfer camera movements or single/multi-object motions. Bring your creative vision to life!

Meet the last GEM Lab pre-print of the year: Explaining in Diffusion! From an all star women-in-STEM team, done by my graduate students Tahira Kazimi and Ritika Allada! We explain the decision-making process of classifiers using diffusion semantics.

FluxSpace is accepted to #CVPR2025! Shoutout to my collaborator Kavana Venkatesh and my advisor Pinar Yanardag for their support. It was a different kind of a ride :D Looking forward to meet with anyone interested in our work at Nashville :)

Super excited to announce that 3 papers from GEMLAB got into CVPR! I am proud of all my PhD students <3 See you in Nashville! #CVPR2025 #CVPR #CVPR2025

Congrats to Yusuf Yusuf Dalva, Tahira Tahira Kazimi, Ritika Ritika, Kavana Kavana Venkatesh and Enis Enis Simsar! If you are looking brilliant summer research interns in GenAI, Tahira Kazimi and Kavana Venkatesh are looking for positions!

👨‍🎨 LoRAShop is now out! We introduce LoRAShop, that enables both generation and editing with multiple personalized concepts (no training), pushing the boundaries of the task of image editing! Kudos to my long time collaborator Hidir Yesiltepe and my advisor Pinar Yanardag !