Formatting Images for use in Scenario

Ensuring that images used throughout Scenario meet the minimum size requirements will prevent failed trainings and generations. It may sometimes be useful to simply increase the size of the image's canvas when upscaling multiple assets would be more time consuming.

In any instance where you do not want your generations to fill in the background, please ensure you use a solid color background in your reference image or datasets. The background color should provide as much contrast from the subject/object represented in the image.

Avoid pixelated or low quality images

When using ControlNet Pose mode, avoid using cartoon or pixel art characters. Life like textures or renders work well, and real life images of human poses are ideal.