A: We're using open source models for image generation from StableDiffusion, but we're using different techniques to change the default model behaviour.
You can see the prompt used to generate image in the edit mode of a single frame.