I'm interesting in playing around with a project that involves text-to-video. Basically ~20 second video clips based on a text summary.

I've looked at e.g. gemini which seems to refuse to generate even the most mundane things without explanation, some chatgpt options, and runwayml, the last of which seemed to be the best option but appears to be ridiculously expensive. I think runwayml also is limited to like ~10 second videos maximum, I think? Which is not enough.

Anyone got any ideas?