Face to Model (new)
- Introduces a new endpoint that transforms face images into try-on-ready upper-body avatars.
- Accepts cropped headshots or selfies and generates full upper-body representations.
- Designed for virtual try-on scenarios where full-body photos are not available.
- Preserves facial identity during generation.
Background Change (improved)
- Improved realism of generated backgrounds.
- Enhanced color and lighting integration between the subject and the new background.
- Foreground subjects now adapt more naturally to environmental changes (e.g., moving from an indoor setting to a sunny outdoor scene updates lighting and shadows accordingly).