In this repository, we look at several approaches to creating avatars, changing backgrounds, and deep image processing for more detailed output. It is also worth noting that we use both ready-made solutions, namely the Stability AI SDK (paid solution), and direct interaction with predefined models (free solution).
5 Steps to Create AI Selfies Generator like Lensa: Stable Diffusion In Action
- Text-to-Image
- Image-to-Image
- Replace-Background
- Depth-to-Image
- iOS app - SwiftUI
- Server - Python
Magic Avatars app requires latest Xcode version to run.
Initial state:
Magic Avatar:
Replace Background:
Install all dependencies (/app/requirements.txt) In the app directory run
conda activate {conda env name}
uvicorn api:app --host {your ip-address} --port {port}- port - 8000 or smth else
Processing devices speed from fast to slow:
- cuda - Nvidia GPU video cards
- mps - Other GPU video cards
- cpu - CPU video cards
MIT