Models and settings

Pick the models

1. Open the editor

   Press Edit on the design’s card.

2. Text

   On the Text tab, choose a Text Model. It writes descriptions and suggests storyboard panels. While the design is active, the AI Assistant uses it too. Account default means the design doesn’t set one. Anything in Text System Prompt goes to that model first, each time it writes a description or suggests panels. The AI Assistant does not get it.

3. Image

   On the Image tab, click the Image Model box and type to search. Leave it empty to use the image model from your account settings.

4. Video

   On the Video tab, pick a Prompt Template and a Video Model. The Timeline uses them when it makes video clips.

5. Save

   Press Save.

The tags in the model list tell you what a model can do. Ref means it takes reference pictures, Neg a negative prompt, and Seed a seed, so a result can be repeated. $ to $$$ is a rough price per picture.

Every model’s price and features are in the model catalog.

Without a plan that includes video, the Video tab shows a Pro tag and an upgrade prompt.

Tune a model

1. Open Advanced Settings

   Press Advanced Settings on the Text, Image or Video tab. It opens for the model picked on that tab. You can pick a different model in the window’s Model box.

2. Change what you need

   The options depend on the model. Most image models have Seed, which you leave empty for a new result each time. Many also have Negative Prompt, Aspect Ratio, Guidance and more. Each option has a note that explains it.

3. Save

   Press Save. Reset to Defaults puts the model’s own values back.

Settings are saved for each model in each design. Switch a design to another model and back, and your settings for the first model are still there.

LoRAs

If the image model can take a LoRA, a LoRA Integrations button appears next to Advanced Settings. The LoRA guides cover it.