Quick Recast

Quick Recast is Recaster's streamlined face replacement tool. Unlike the traditional DeepFaceLab workflow that requires face extraction, training, and merging, Quick Recast uses pre-trained models to swap faces in a single step. Simply drag and drop your source face and target media, select a model, and click process.

How It Works

Quick Recast uses a processing pipeline that orchestrates face detection, swapping, and optional enhancement in a single pass:

The pipeline automatically handles face detection in the target, embeds the source face identity, blends the result into the target frame, and optionally enhances the face quality. For video inputs, the pipeline processes every frame and preserves the original audio.

Key Features

• Drag-and-drop interface — No complex configuration. Drop your source and target to get started.
• Multiple swap models — Choose from InSwapper (Free), Ghost, or BlendSwap (Studio) depending on your quality requirements.
• AI face enhancement — Apply GFPGAN, CodeFormer, GPEN, or RestoreFormer++ to improve face quality after swapping.
• Video processing — Frame-by-frame processing with real-time preview and progress tracking.
• Multi-identity mode — Assign different source faces to different people in the same scene.
• Video trimming — Select a range of frames to process with a dual-handle range slider.
• Audio preservation — Original audio is automatically preserved when processing videos.
• Before/after comparison — Interactive slider to compare original and processed results side by side.
• Remote processing — Process on cloud GPUs with the Studio tier for faster results on large videos.

Available Models

Quick Recast supports several face swap models with different quality and speed characteristics. Models are automatically downloaded from Hugging Face on first use and cached locally for future runs.

For detailed information about each model, swap quality, and enhancement options, see the Basics and Enhancement Options pages.