This is something I really would love to play around with but it looks like at this time you need a graphics card with some pretty big horsepower. I would expect that to change and it to work with more cards in the future. Can't wait!
This is something I really would love to play around with but it looks like at this time you need a graphics card with some pretty big horsepower. I would expect that to change and it to work with more cards in the future. Can't wait!
I would expect to just run it mostly in the cloud. I'm not sure what type of cards are being used by Midjourney etc, but I know when was I was previously using Disco Diffusion I would sometimes be running AI in the cloud on very high-power Tesla V100 GPUs, which are something like $3,000+. It just seems like the obvious way to go in centralizing it unless you have very niche needs.
__________________
"If stupidity got us into this mess, then why can't it get us out?"
Hey sorry had to cut those out, they were each 2+MB PNG files (40 of them), waaay too much for a page (the tag to hide just hides it, it doesn't prevent the page from loading them all).
Could either switch them to JPGs to be smaller or just have links. Thanks.
__________________ Uncertainty is an uncomfortable position.
But certainty is an absurd one.
I can understand using it as a bit of a tool, but his discussion of it being used to create stories and music just sounds unappealing. Computer generated content is mostly bad. It's got a sameness to it, and lets be honest, AI doesn't have much creativity, not compelling anyway. Have you read AI generated stories? If all studios lean to heavily on these tools it'll kill the industry. And the temptation will be there, given the massive cost savings.
Those Stranger Things pics, those faces look way better than I thought AI could do. Crazy.
Fascinating part to me is the more descriptive you are the better the result. I'm talking ultra detailed, don't know the exact prompts I used but they generally contained at least 20 different parameters. Even things like F-stop and focal length.
Plus it's very entertaining, you have little to no idea what the image will look like.
Quote:
Originally Posted by Fuzz
Nice cheap dunks, but really, see if you can even make it all the way through this:
Fascinating part to me is the more descriptive you are the better the result. I'm talking ultra detailed, don't know the exact prompts I used but they generally contained at least 20 different parameters. Even things like F-stop and focal length.
Plus it's very entertaining, you have little to no idea what the image will look like.
So? This is just the beginning.
Sure, but it's got a long way to go to match human skill and creativity in story telling. If this is the kind of nonsense our games are going to be built on, that's not going to make for very good games. But studios are going to be very tempted to use it anyway, given the cost savings.
Sure, but it's got a long way to go to match human skill and creativity in story telling. If this is the kind of nonsense our games are going to be built on, that's not going to make for very good games. But studios are going to be very tempted to use it anyway, given the cost savings.
I would think they are messing with a hybrid human/AI type of thing
Sure, but it's got a long way to go to match human skill and creativity in story telling. If this is the kind of nonsense our games are going to be built on, that's not going to make for very good games. But studios are going to be very tempted to use it anyway, given the cost savings.
Did you read the thread? He says in it that artists aren't going to just be replaced but that they will make use of these tools. This is already happening.
__________________
"If stupidity got us into this mess, then why can't it get us out?"
Did you read the thread? He says in it that artists aren't going to just be replaced but that they will make use of these tools. This is already happening.
Ya, until management asks "The computer made this all on its own? Well what do we need you for?"
Ya, until management asks "The computer made this all on its own? Well what do we need you for?"
When computers started generating sounds which were turned into music people said we won't need artists anymore. 50 years ago Pink Floyd was accused of letting the electronics do all the work. Same thing happened when CGI first hit, we won't need people anymore. Puppeteers thought they would be out of a job, turns out they are needed to create realistic motion only thing that changed is the input method.
There may be a time when A.I. can do absolutely everything from beginning to end and do it better, not sure how I feel about that. In the mean time semi-intelligent A.I. is simply another tool among many. For me personally it opens up an entirely new way to create, I don't have the mechanical skills in my hands to create art but now I can create at least on some level.
What about people with physical limitations, now they can express themselves artistically. I think you should try tools like Stable Diffusion before passing judgement. I strongly disagree that A.I. generates images that all look the same, you can get incredible variety it's mostly up to how well you can describe a scene, both technically and artistically.
That is my favourite, it has the likeness but also a vibe of its own. The hair is amazingly detailed. There is usually chaos in the renders which can be incredible or downright disturbing. I kept using the tag amulet which the A.I. doesn't know how to piece together properly most of the time. But sometimes it looks great, or it turns amulet into a skull. The permutations are basically endless talk about unleashing imagination.
I'd love to see what people with legit artistic talent can do with tools like SD.