Photo AI uses its own generation pipeline which is trained for high photorealism. The main generation model we use now is Flux. We use many other models to perform other functions like post-processing, upscaling, taking videos, adding audio to videos, adding voice to videos etc.

We keep working on improving our pipeline using reinforcement learning from human feedback (RLHF) by users like you which constantly improves our models.

All of this to get you better more photorealistic photos and videos!