Introduction to Vub Creation

• What You Need to Create a Vub 

• Shots and Vubs

• The Vub Creation Process

What You Need to Create a Vub

To create a vub in DeepEditor, you need two pieces of media:

• Source media - the original shot containing the actor's performance you want to modify. This should be a single, continuous clip of no more than 1440 frames (= 60 seconds @ 24fps).
• Driving Data - the new dialogue or performance you want to apply to the Source media. This can be new dialogue audio or a shot with the new visual performance.

Optionally, you can also upload Training Data - additional footage of the same actor - to help DeepEditor build a more robust model and improve the image quality of the result.

Shots and Vubs

In DeepEditor, a shot is the container that pairs your Source media with your Driving Data. Think of it as framing the question: here is the original performance, and here is what you want to change it to.

A vub is the output: the visual dub produced once DeepEditor has processed the shot. It is a video file containing the actor's facial expressions and mouth movements adjusted to match the new dialogue or performance.

One shot can produce multiple vubs - for example, testing different Driving Data options, or generating the same shot at different quality levels for different stages of the workflow.

The Vub Creation Process

Once your Source media and Driving Data are in a shot, DeepEditor will:

• Analyse the media and detect every character in the shot
• Let you review, name and organise those characters
• Let you create the vub once you select the character and chosen quality level

The initial vub generation typically takes approximately 1 hour for an average duration shot (~5 seconds) but may vary depending on the duration and complexity of the shot.

Read next: Source media