Soul 2 is an image-generation model described as "hands down the best model right now for generating production ready images." It's positioned as the tool of choice for producing a high-quality Start Frame as the Primary Determinant of AI Video Quality before animating it into video — since a blurry or "plasticky" AI-looking start frame carries that low quality through the whole generated video.
Image generation with Soul 2 pairs with a prompt enhancer toggle: write a simple description (e.g. "a man skating") and turn on the enhancer to automatically expand it into a fuller, more production-detailed prompt, rather than hand-writing every production detail yourself.
See also Brief Image Generation Like a Movie Director, Not a Prompt Engineer for how to brief the image beyond just enabling the enhancer.
Из тем: Prompting Foundations: Briefing the Model Like a Director