this is about 10 iterations down the line, suggesting developments.
it’s very limited now, you have to persistently trick it to do anything interesting, i wish “photo” requests created the amazing images it manifests using only photographs from it’s training library, but this appears to have been severely restricted (probably for privacy concerns) and seems it only does it by mistake once in a while…
my concern is users thumb up generic results that meet their existing expectations, therefore any new forms and concepts are thumbed down and they’re less likely to do anything mad (unless you push and push).
Thanks! Will hopefully have some fleshed out music to post soonish too!
@cupfungus Thanks - this particular one was made with Zeroscope (an open source text or video to video model) in Comfy UI. Different methods for the other stuff I’ve posted in this thread.
@Naboo for text to video the main online services are Runway Gen-2 and Pika. JING, a competitor to Sora has just been released in China but you need a Chinese phone number to access it.
I prefer to use open source tools on my own computer so have Comfy UI as my Stable Diffusion GUI and do most stuff in there.
There’s a fancy new text/image to video thing out today (expect long wait times while everyone clambers to try it out):
Edit: adding my 1st attempt at using LumaDM. More motion than recent text to video models when trying to get photographic outputs but struggles with anything abstract:
IMO people creating stuff with AI can and will be considered as artists, but they shouldn’t be put together with traditional artists like painters, musicians etc. I think what is developing is a new category of AI artists, where it’s not about the type of art you produce (e.g. pictures, music…) but more of how creative you can be with AI tools to produce something unique that differs from what those tools spit out by a generic prompt. Think of how Holodeck programs are developed in Star Trek
I’m a programmer and I sometimes consider myself an artist, although I’m probably in the minority But it’s not for what is the final product of coding but for a code itself. It’s not that rare that I restructure it for pure cosmetic purposes or just because I don’t like how I programmed a solution, even if it’s working perfectly fine.
I agree that programers are artists. I felt that when I was designing a house and than building it was an artistic expression.
I would not call it art if what comes out of a computer by way of giving it prompts like: create a song in Taylor Swift’s style or paint a picture in Van Gogh’s style. Perhaps that should be called artificial art.
Q for Raskal.X - Should photographers consider themselves artists? They are just (sometimes setting up a scene then) clicking a button.
Q for tha_man - Can photographers be considered artists but not traditional artists like painters or musicians? Or if photography can be considered ‘traditional’ now, could a point in the future exist where AI is considered traditional too?
Not too many years ago Photography was a new evil.
