PhotoMaker is an efficient and personalized text-to-image generation method that primarily encodes any number of input ID images into stacked ID embeddings to preserve ID information. This embedding, as a unified ID representation, can not only comprehensively encapsulate the features of the same input ID but also accommodate the features of different IDs for subsequent integration. This paves the way for more interesting and practically valuable applications. Moreover, to drive the training of our PhotoMaker, we propose an ID-oriented data construction pipeline to assemble training data. Nurtured by the dataset constructed via the proposed pipeline, our PhotoMaker demonstrates better ID retention capability compared to methods based on test-time fine-tuning, while simultaneously providing significant speed improvement, high-quality generation results, strong generalization ability, and wide applicability.