Skip to content

Mimic 1.0

Choose a tag to compare

View all tags
@CursedPrograms CursedPrograms released this 11 Nov 22:51
· 5 commits to main since this release
v1.0.0
b79b05f

Mimic 1.0

FaceFusion's face processors in C++: put one person's face onto the faces in a picture or a video, restore the detail, and see what it did, on your own PC, with no Python. It runs on any DirectX 12 GPU through DirectML (NVIDIA, AMD or Intel), or on the CPU. Part of Atelier, and a LayerTack plugin (Filter > Mimic, LayerTack 1.2 and up).

  • Face swapper: 13 models (inswapper, HyperSwap 1a/1b/1c, SimSwap 256/512, Ghost 1/2/3, HiFiFace, BlendSwap, UniFace), with pixel boost and weight.
  • Face enhancer: GFPGAN 1.2/1.3/1.4, CodeFormer, GPEN 256 to 2048 and RestoreFormer++.
  • Face selector: every face, the first, or faces like a reference you click; 8 orders; age, gender and race filters.
  • Face masker: box, occlusion (hands, hair and glasses stay in front), area and region masks.
  • Detector and landmarker: RetinaFace, SCRFD, YOLO-Face, YuNet; 2DFAN4 and PeppaWutz; angles for sideways faces.
  • Face debugger: boxes, face points and the mask outline.
  • Video: through ffmpeg with the sound kept; x264, x265, VP9, NVENC, AMF or QSV; quality, fps and trim.
  • mimic-cli takes FaceFusion's own option names.

Install

Unzip Mimic-win64.zip and run install.bat inside it (Start menu and Desktop shortcuts; LayerTack finds it there), or just run Mimic.exe. Windows 10/11, 64-bit. You can also install it with the Atelier launcher.

The models download once, the first time each is used, from FaceFusion's model releases. Some are non-commercial only; see NOTICE. Videos need ffmpeg on the PATH.

Only use faces you have the right to use, and say when a picture or video has been changed.

Assets 3
Loading