Commit Graph

5 Commits

Author SHA1 Message Date
58445340c8 feat: remove gpu support and simplify to cpu-only
All checks were successful
Build and Push Docker Image / build (push) Successful in 1m13s
2026-02-19 12:50:10 +01:00
3812bff509 fix: replace libgl1-mesa-glx with libgl1
libgl1-mesa-glx is a transitional package removed in Debian Bookworm.
libgl1 is the correct replacement for both CPU (Bookworm) and GPU (Ubuntu 22.04).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 13:28:49 +01:00
0d51204862 fix: set DEBIAN_FRONTEND=noninteractive for GPU build
Prevents tzdata interactive prompt when installing deadsnakes PPA.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 13:18:49 +01:00
4856599b57 fix: align dependency versions with FaceFusion 3.5.3 and fix GPU build
- Update onnxruntime to 1.24.1, numpy to 2.2.1, opencv-python to 4.13.0.92,
  scipy to 1.17.0, onnx to 1.20.1 to match facefusion/requirements.txt
- GPU Dockerfile: install Python 3.12 from deadsnakes PPA since CUDA base
  image ships Python 3.10 which is too old for onnxruntime-gpu 1.24.1
- Copy both requirements files in COPY step so -r reference resolves

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 13:16:21 +01:00
800edc08ea Initial commit: FaceFusion REST API
FastAPI wrapper around FaceFusion v3.5.3 submodule with:
- Sync and async (job-based) processing endpoints
- FaceFusion bridge with manual key registration and Lock-serialized processing
- Multi-target Dockerfile (CPU + CUDA GPU)
- Docker Compose configs for dev, prod-cpu, and prod-gpu
- Gitea CI/CD workflow with dual image builds
- All 11 FaceFusion processors supported via options API

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 12:58:33 +01:00