Search papers, labs, and topics across Lattice.
This paper introduces PIPBench, a novel benchmark designed to evaluate personalized image generation by aligning outputs with individual aesthetic preferences derived from historical user data. The authors employ a unique data construction pipeline that incorporates psychological and demographic profiling to enhance both real-user data collection and scalable agent-based data generation. Their evaluation of existing methods reveals significant limitations, highlighting the need for improved approaches in personalized text-to-image synthesis.
Existing text-to-image models struggle to capture individual aesthetic preferences, but PIPBench reveals critical gaps in their performance that could redefine personalized image generation.
Recent text-to-image models such as DALLE-3 excel at following diverse prompts yet remain blind to individual aesthetic preferences. We study personalized image generation, where models must align outputs with a user's implicit visual preferences based on a few historically preferred images and a short prompt. To this end, we introduce PIPBench, the first profile-inclusive benchmark for evaluating personalized image generation. We further propose a novel data construction pipeline that leverages psychological and demographic profiling dimensions for both real-user data collection and scalable agent-based data generation. Using PIPBench, we conduct a thorough evaluation of representative line of methods. Our experiments reveal key limitations in existing methods, suggesting new challenges and opportunities for personalized text-to-image synthesis. Project page: https://wuyuhang05.github.io/PIPBench/
Tsinghua AI