Train an avatar LoRA for video | Everypixel Workroom

Avatar LoRA training in Cinema Motion is unavailable. It ran on WAN 2.2, and that engine has been retired from Workroom. We're bringing character training back on newer video models — we're not announcing a date yet. Character LoRAs for images still work: see Train an avatar in Picture Lab.

WAN 2.2 supported LoRA training directly in Cinema Motion — Workroom's AI video generation studio. You uploaded reference photos of a person, a visual style, or a product, and the model learned to apply that look consistently across your video generations, without first creating a reference image in Picture Lab. Here's how the workflow ran.

Cinema Motion with WAN 2.2 selected, showing LoRa Library and Start Frame buttons in the control bar and the Train Avatar toggle at the top-right

Before you begin

You needed 15–30 clear photos of the same person. The model learned from variety — different angles, expressions, and lighting conditions. Sunglasses, hats, or anything obscuring the face had to be avoided.

Select WAN 2.2

You opened Cinema Motion and clicked the model selector in the top-left, then chose WAN 2.2 — it showed a Trainable badge. The control bar updated to show LoRA Library, Start Frame, and the Train Avatar toggle in the top-right.

Enable Train Avatar mode

Clicking the Train Avatar toggle at the top-right switched the interface from the generation view to the avatar creation screen.

Upload your photos

The upload screen showed example reference photos and a drop zone.

You dragged your photos onto the drop zone or clicked Upload to browse. At least 15 photos were required for training to start. The screen showed your upload count as you added files.

Photo requirements:

Face clearly visible, no heavy accessories
Different angles and expressions
Various lighting conditions
Single person per training session

Start training

With 15–30 photos uploaded, you gave the avatar a name and submitted. Training took 20–50 minutes and ran in the background — you could close the tab and come back later. When training completed, you received an email notification and the avatar appeared in your LoRA Library under the Avatar tab.

Use the trained avatar

After training, you opened Cinema Motion with WAN 2.2 selected and clicked LoRA Library in the control bar, then picked your LoRA from the list. The LoRA applied to your next generation — keeping the character, style, or product consistent across all your videos without going through Picture Lab first.

You could also start a generation directly from the LoRA Library: the Generate button on any trained LoRA card opened Cinema Motion with that LoRA already applied.