Synthetic Factory / Products / MVE
● Maintenance Mode
● Next Version in Active Development

Machine Video Editor

// Dataset Editor

A professional dataset editor built around the DeepFaceLab pipeline. Inspect and edit masks and landmarks, analyze pose coverage, sort and group faces by similarity, all from a single interface.

MVE is stable and fully functional. Active development has moved to the next generation, no further feature updates are planned for this release.
Download — Windows Download — Linux
MVE main screenshot
Machine Video Editor — Main Interface Live screenshot
// Features
⬡

Pose Heatmap

Visualize pitch and yaw distribution across your entire dataset as a 2D heatmap. Filter the face viewer by angle range to find and fix coverage gaps before training.

⬡

Face Grouping

Automatically cluster faces by visual similarity using VGG embedding vectors. Identify duplicates, outliers, and distinct expression groups at a glance.

⬡

Mask Editor

Inspect and edit segmentation masks per-face. Visualize mask coverage directly on the extracted face grid and correct problem frames before they affect training.

⬡

Landmark Editor

Review and adjust facial landmark placements across your dataset. Catch misaligned extractions that would otherwise degrade model quality.

⬡

Face Set Comparison

Side-by-side comparison screen for source and destination datasets. Compare angle distributions and dataset characteristics to identify mismatches early.

⬡

DFL Script Launcher

GUI bindings for DeepFaceLab's core functions: extract, sort, train, merge. Run the full DFL pipeline without touching a terminal.

The Next Generation is Coming

MVE's successor is in active development and being released in focused, specialized modules as they're completed. The current release remains stable and fully functional in the meantime.

In Development
Follow Updates

Free. No License Required.

Includes a self-contained Python installer. GPU recommended for full functionality.

$0 / forever
Windows Linux GPU Recommended
Download — Windows Download — Linux
Also try MVE Trainer →