I'm primarily using the script use case with a story arc. I've done four videos so far.

I like everything about the workflow and U.I., but the scene images I'm getting back are very hit-and-miss. They are people with skewed or backward limbs, mangled hands, and/or the wrong age or gender, so to get a relevant image, I need to write a better prompt and sometimes regenerate more than once.

Unfortunately, it becomes a grind, albeit much less so than working in a traditional video editor.

I bought Tier 2, and I will try to hang in there and see if I can squeeze the economy-of-scale benefits I want. Hopefully, the images will improve organically along with all A.I. in the months ahead.

It would be great to have a Stability API key for images and, beyond branding, be able to prompt an image avatar to generate character-to-video script continuity (e.g., a fit and muscular 55-year-old gray-haired man shot on Kodak Portra 400 film, realism, hyper-photo realistic).