Skip to main content Scroll Top
Copy Face
New

Copy Face

49,99 $

Copy Face replaces a source face (called “source face” or “target face”) with a new face in a temporaly stable manner.

Running fully locally, no server processing.
Powered by blace.ai

Copy Face for After Effects

Fully hardware accelerated. From Full HD 8-bit to 4K 32-bit, rendering speed remains constant.

Rendering happens locally. Your data never leaves your machine.

Completely C++ based for maximum stability, performance and security. No Python environments installed.


Features

New! Instant Model.

The new instant model creates realistic face swaps and renders in under 300ms per frame, even on 32bit 4k footage.

New! Change face with text input.

Simply describe the new face with text input and have it transfered onto the target face.

High-quality Face Swapping

The locally running AI replaces faces in your video with new ones!

Works on real faces (Use Instant Model)

Great for actor replacement and content personalization!

Works on stylized faces (Use Style Model)

Transfer anime stylizations, super heroes and other stylized faces.

Easy to setup

  1. Important: In order to safeguard this technology, we introduce a face-verification system.
  2. Apply the effect to your footage. Make sure only one face is present in the clip, as Copy Face currently only supports single face processing (working on an extension already).
  3. Choose another layer which contains the new face or use a text prompt to describe the new face. The new layer can either be an image or a .cpf file, in case you send it for verification. Faces created with text prompts don't need verification.
  4. Select "New Face" to verify that the new face is aligned properly, so it covers the whole frame. If the face is not aligned automatically, make sure the face is already centered in the selected layer.
  5. We deploy two models: Instant and Style. Instant renders realistic faces quickly and Style excels at swapping of stylized characters. Instant is the default.
  6. The Style model has two modes: In "One Frame Preview", the source face is replaced with the new face frame-by-frame. This allow for quick preview and tweaking of settings, but is not temporaly stable. Choose "Full Sequence" to process multiple frames together. The plugin works on ~50 frames at once, so for the first frame and then every ~50 frames the computation is longer (up to 1 or 2 minutes). All frames in between are rendered very quick. Note: The visual output of "One Frame Preview" and "Full Sequence" will be slightly different. If you want to cancel "Full Sequence" processing, click "Stop Processing".


System Requirements

Windows with NVIDIA GPU >= 12GB Other Windows Apple Silicon with MacOS Sequoia or newer and >= 32GB RAM Other MacOS
✅ very slow ✅ ❌

Documentation


Follow us on socials

Item added to cart View Cart Checkout