#The sheer power I have with this ai video generator
1 messages · Page 1 of 1 (latest)
New golem feature?
/showcat?
unfortunately I don't think the server has those resources
well it does
but
it doesn't have a gpu
so
I can't be bothered setting it up honestly lol
You can try with ollama and see how it performs
But it's going to be very slow without a GPU 💀
how well would it perform with rtx 4060?
yes
that's what I thought 
jay do you know a good voice cloner repo?
I need to clone my voice
🥺
no
what should I generate with my ai
bastion from overwatch
as in a youtube shitty documentary video thing
1968 documentation video thing
whar
like it's a war project documentation
I'm not sure if the ai will like that
yeah no it doesn't like that
but it did generate a script apparently
so we will be getting... something holy fucking shit
I didn't know showcase was that active
it actually didn't crash
despite the insane amount of errors
my code held together
Can it generate bastion schematics?
It can generate videos like the one above
I mean can you make it generate bastion schematics
Just give it an image prompt, not moving thing and let's see what happens
Bastion schematic
On blueprint
a what
💯 Auto 💯 correct 💯
I haven't coded it to allow me to use an image as a prompt
I need a text promp
t
Depends on your ram, more precisely if all of your ram (or even better, vram) can hold all of the AI model
But if it can, then it's going to be fast as fuck
I know like I bought this PC specially for ai stuff; 32gb ram, rtx 4060 8gb
VS2: the documentary
8 GBs of VRAM is kind of ass
for ml
What ai are you talking about 💀
GPT 4
Even 2-4 is okay for most of the stable diffusion models
Nah it's fine I'm using it
you can run stable diffusion xl on a raspberry pi
Yeah
back when I was local hosting llama and GPT-J
mf ate 16 GBs of Vram
I remember generating an image with stable diffusion when that stuff was new and it took like a minute
now I can do it in under a second
on my 2070
wait I have 8gb of vram?
whar
I had no idea
Hmmmm
in anycase, couldn’t you have payed for google cloud for ai shit
which would have been cheaper, while working much faster
I mean yeah
Yeah 8gb vram is not that great (there's consumer level gpus with 20-24gb vram)
But 32 ram is enough for most AIs
But it makes my job easier than before, I sell ai generated baby-child-adult books with images and all 
(I had GTX 1650ti with 4gb VRAM)
