What is Training Data
Training Data consists of extra footage of the character whose dialogue you are modifying, which is uploaded as reference material for DeepEditor to study and generate a more accurate Vub.
NOTE: Additional Training Data can only be added to Standard, Professional and Theatrical Vubs.
How Training Data Works
Training data enables DeepEditor to learn how the face of the person you are vubbing moves, so it can produce a better vub.
DeepEditor will go through and add extra detail to the model based on the frames of video in your Training Data.
Training Data Requirements
To get the best results use footage that is really similar to your Source media (e.g. alternate takes) and meets the following criteria:
- The person's face is clear and unobstructed
- The person is speaking (this is required so DeepEditor can learn the way the person's face moves when they speak)
- The person's face position and angle is very similar to the Source media
- Lighting and background is very similar to the Source media
- Each clip features the same actor as the Source media
- The media is compliant with the DeepEditor Supported Media Formats
Uploading Training Data
Drag and drop training data files or click "Add Training Data" in the Training Data section of the Media tab. Currently, you can upload a maximum of 40 video files of up to 1,440 frames each.
Once you have uploaded them, you will be prompted to select which files you want to use before creating your Vub.
Your Training Data uploads can take a bit longer to upload than your Source Media & Driving Data.
NOTE: Vubs with Training Data will take longer to render. Completion can take up to 12 hours, depending on the amount of Training Data you uploaded.
However, the results are much better when you add Training Data.
Here is an example of how adding Training Data can improve your Vub.
Source
Vub without Training Data
Vub with Training Data
In the final image, with Training Data, you can see the teeth and the inside of the mouth are much more clearly defined.