Originally Posted by tomash999

I don't know about long form video generation, but you can use vid2vid or img2img, and maybe also generate short gif like videos.

If you're looking for a platform with API, then you'll always have limitation about the content, and no competitive advantage. Instead, try to install locally or use cloud services. Usually the cloud services have tutorials and demos of what you want.

The LLaVA can do better if you change the prompt:
"Describe the model in the image using the following format:
A {age} {hair color} {clothing} {action} {pose} {face expression} {location} {background}"
Result:
"A young woman with blonde hair, wearing a colorful, flowing garment, is standing in a relaxed pose with her hands gently holding the fabric of her dress. She has a soft, neutral expression on her face. She is located in an outdoor setting, possibly a patio or garden, with a rustic stone wall and a wooden barrel planter in the background. The sky is visible, suggesting it might be a sunny day."