Been having some success in experimenting with Krea. It has an amazing feature where you can upload a B&W logo or pattern and it guides the rendering for the prompt. This is what I got using my own logo on a prompt for a snowy Nordic birch forest -
I think it would’ve looked better if I’d uploaded an image of Pinhead first.
Pika, which is text to video, opened up its alpha today. It’s just ok, looks like it’s mostly for the odd animated still at this point. There’s a load of shit clips floating about on their Discord though. Like MJ, it’s all down to the prompt and what you base it off.
Speaking to designer and cgi based friends, Ai is a long way off because it’s not creative, it just copies everything, so for the time being will be used for cutting time in programming/coding etc .
But it’s early days and will catch up soon and studios are businesses and saving money usually means cutting crew .