Stable diffusion dreams of "blueberry spaghetti" for one night
Vložit
- čas přidán 15. 08. 2022
- Prompt: "blueberry spaghetti"
Stable diffusion takes a noise vector as input and samples an image. To create this video I simply smoothly interpolate between randomly chosen noise vectors and render frames along the way.
Links
- Stable diffusion: stability.ai/blog
- Code used to make this video: gist.github.com/karpathy/0010...
- My twitter: / karpathy
Some slow restaurant Jazz would fit nicely, but 2013 Dubstep also did the job.
"hey guys welcome to my minecraft survival gameplay"
If blueberry spaghetti ever catches on, we can all say we were here at the start of the revolution
This is a very cool demo, thanks!
Hey Andrej,
you are such an interesting and nice guy. Me and i think many others would like if you could do videos about topics like machine learning theory, philosophy, projects you are working on and your opion on current developments in machine learning for example.
I am originally from the Lac St Jean region where there is a blueberry festival every year and i have never heard of a blueberry Spaghetti intesresting
Blueberry spaghetti? I'd try it once.
Regardless, AI proposals of new flavor combinations seems like a very innovative way to push the boundaries of the modern culinary experience.
I love how Andrej Karpathy is using the default iMovie music from like 2010. 😂😂
All in good fun; you are awesome Andrej keep up the great work.
Does Andrej read his comments or reply hi?
I really liked when the noodles were stained purple with the basil garnish. You could make this with some sort of pancake noodles and it would be pretty good. Once I did a blueberry omelette w/ white cheddar in real life and it was ok, not great.
We miss you Andrej.
Great vid also love the early 2010’s French electro
Waiter, "blueberry spaghetti" please; don't ask why! ;-)
I genuinely don't know what these videos represent or how they were created - I'm assuming via A.I. from a computer? If so, it would be amazing to have a bit more context to your videos so the average person can understand how important / crazy cool they are! I think you would gain many more views and subscribers that way :-)
hmmm ok, may do :)
@@AndrejKarpathy Does this mean we will get a stable diffusion from scratch like video series since that besides chatGPT is what has got everyone hooked and following the ai trends now? ^^ I'm sure everyone is curious about the inner machinery (I know my sister and father are) :)
I'd like to see an artist start with the same noise, and see how close the results come to SD sampling. (Human artists use this technique all the time as a concepting base to get started.)
It's a great demo because you would never see a combo like this in real life.
If you wanted to make this much blueberry spaghetti with human labour, it would take a lifetime (and a massive waste of spaghetti and blueberries... unless?).
Maybe use the unregistered hypercam music on the next one.
Let the music be the noise source?
Wow 😳
This feels like a shitpost and I like because of it
inference time well used?
corn dog based imagery please :-))
@andrej will you be covering diffusion models especially the stable diffusion?
I just bought another share of Tesla
Glad they’re not peas. I hate peas
that's disgusting! :)