FaceChain trains a unique digital likeness of a person from just two photos. You can then batch generate portraits in various styles. FaceChain uses the image generation capabilities of a diffusion model, combines LoRA training to merge the person's likeness with styles, and applies post-processing to create portraits that are realistic, visually appealing, and closely resemble the subject.
The following topics describe how to call the FaceChain portrait API on the large model service platform.