First Project
Create your first face replacement project and learn the workspace.
This guide walks you through creating your first project in Recaster, from the initial consent dialog through loading media and understanding the workspace layout. By the end, you will have a project ready for face extraction, training, or Quick Recast processing.
BIPA Consent Dialog
When you launch Recaster for the very first time, a consent dialog will appear before any other interface loads. This dialog fulfills the requirements of the Illinois Biometric Information Privacy Act (BIPA), which mandates explicit consent before collecting or processing biometric data such as facial geometry.
Required Consent
The consent dialog explains:
- What biometric data is collected (facial geometry and landmarks)
- The specific purpose of the data collection (face replacement processing)
- The retention period and destruction policy
- Your rights under BIPA
After accepting, your consent is recorded with a timestamp and version number. You will not see this dialog again unless the consent terms are updated in a future release, in which case you will be asked to re-consent.
Consent is Stored Locally
Creating a Project
After accepting the consent dialog, the main Recaster window loads. To create a new project:
Open the New Project dialog
Go to File → New Project in the menu bar, or use the keyboard shortcut Cmd+N (macOS) / Ctrl+N (Windows/Linux).
Choose a project folder
A file browser will open. Navigate to the location where you want to store your project files and either select an existing empty folder or create a new one. Each project needs its own dedicated folder.
Name your project
Enter a descriptive name for your project. This name is displayed in the project panel and title bar. You can rename it later from the project settings.
Project is ready
Recaster creates the project structure inside your chosen folder and opens it in the workspace. The project panel on the left shows your project directory tree.
Opening Existing Projects
Cmd+O / Ctrl+O) and select the project folder. Recaster will detect the project structure automatically.Project Structure
Every Recaster project has a standardized folder structure based on DeepFaceLab conventions. When you create a new project, these directories are created automatically:
my-project/
data_src/ # Source face images (the face you want to use)
aligned/ # Extracted and aligned source faces
data_dst/ # Destination images/video (the face you want to replace)
aligned/ # Extracted and aligned destination faces
model/ # Trained model files (.dat, .npy)
merged/ # Final output after merging
project.json # Project metadata and settingsDirectory Purposes
| Directory | Purpose | Contents |
|---|---|---|
data_src/ | Source faces | Images or video of the person whose face you want to use as the replacement |
data_dst/ | Destination faces | Images or video containing the original face to be replaced |
model/ | Trained models | Model weights and configuration files generated during training |
merged/ | Output | Final merged frames and reassembled video after face replacement |
Loading Media
With your project created, the next step is to load your source and destination media. There are several ways to do this:
Drag and Drop
The fastest way to load media is drag and drop. Drag image or video files from your file manager directly into the Recaster window. Depending on which area you drop the files, they will be placed in the appropriate project directory.
Supported Formats
Recaster supports common image and video formats:
Image Formats
- JPEG (.jpg, .jpeg)
- PNG (.png)
- BMP (.bmp)
- TIFF (.tiff, .tif)
- WebP (.webp)
Video Formats
- MP4 (.mp4)
- AVI (.avi)
- MOV (.mov)
- MKV (.mkv)
- WMV (.wmv)
- FLV (.flv)
Workspace Layout
The Recaster workspace is organized into several key areas:
Sidebar (Left)
The left sidebar contains navigation buttons for the main application views. Click an icon to switch between different tools:
- Face Browser — Browse extracted face thumbnails with overlay options
- Face Editor — Edit face masks and segmentation
- Training — Configure and run model training
- Quick Recast — Fast face replacement with drag-and-drop
- Upscale — AI video upscaling
Project Panel
The project panel shows your project file tree with three tabs:
- Local — Your local project directory tree (always visible)
- Remote — Remote instance file tree (Studio tier, visible when connected)
- Sync — Upload/download file synchronization (Studio tier)
Central Area
The central area displays the active tool. This changes based on which sidebar item you have selected. For example, selecting "Quick Recast" shows the drag-and-drop face replacement interface, while "Training" shows the model training configuration and preview.
Status Bar (Bottom)
The bottom status bar shows:
- Current project name and path
- Active processing status and progress
- License tier indicator (Free or Studio)
- GPU status and memory usage
Quick Start: Quick Recast
If you want to try face replacement right away without training a custom model, Quick Recast is the fastest path. It uses pre-trained models to swap faces instantly:
Open Quick Recast
Click the Quick Recast icon in the left sidebar (lightning bolt icon).
Add a source face
Drag an image of the face you want to use into the Source drop zone on the left side of the interface.
Add a target
Drag an image or video containing the face you want to replace into the Target drop zone on the right.
Process
Click the Process button. Recaster will swap the source face onto the target and show a before/after preview when complete.
For a complete guide to Quick Recast features, including video processing, multi-identity mode, and enhancement options, see the Quick Recast documentation.
Next Steps
Now that you have a project set up, explore these areas:
- Quick Recast Basics — Detailed walkthrough of the face replacement interface
- Local Training — Train custom models for higher quality results
- Mask Editing — Refine face masks for precise blending
Was this page helpful?