I'm glad to see another project tackling the challenge of speeding up image generation. After seeing the bang-for-buck jump from sdxl to sdxl-turbo, I assume there's tons more room for improvement on the speed side of things.
MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
11–20 of 63 posts
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#12This is 30 times faster than SD (on its own, as opposed to LCM methods) right?
Noting the "50" steps they referenced- whereas I've been using LCMs with 3-8 steps for months.
Just trying to understand context of the paper / comparisons etc.
90ms is fast, but like 2-3x speedup (over what I've been seeing) which is still huge, but not 30x
Can someone with more context explain?
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#13Makes me wonder that on any given day someone could find way to improve something, and OpenAI value will evaporate quickly. Seems like these startups with high valuations, are more tenuous even than the startups of years past.
Safe to say that FedEx has had to add another flight or two to carry documents from OpenAI to the USPTO.
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#14Am I missing it or is the source code not available?
I don't think they open sourced it. Thinking back, most AI breakthrough papers I've read don't include source code unfortunately. Researchers want their names out there explaining what they did and that they did it first, but MIT might want to license the implementation IP or Adobe (sounds like their interns discovered this during a summer) lawyers might be hanging onto it for a business edge.
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#15Is the text prompt input the only common thing between SD and MIT models when comparing outputs?
If so I am surprised at how SIMILAR the outputs of both models look in the general layout / framing / composition of the image.
How can for example both fox astronaut images have near identical backdrop of earth and earth alone on the same side of image at same apparent size. Virtually the same shade of deep blue for deep space.
"Lightshow at the Dolomities" output is virtually identical. So similar that they almost look like iPhone versus Samsung Galaxy camera comparison shots of same scene (I am exaggerating of course but almost there).
What results in such a close similarity in outputs? Same training data set?
It is almost like they have the same DNA!
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#16There are so many similar diffusion model distillation techniques these days that it's become hard to tell the difference between them, this is probably the fourth example I've seen that uses adversarial loss to distill the model, the others being UFOgen, some other work by StabilityAI on SDXL-turbo and similar, SDXL-Lightning by ByteDance. I found this blog page that explains some of the differences: https://sander.ai/2024/02/28/paradox.html
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#17Makes me wonder that on any given day someone could find way to improve something, and OpenAI value will evaporate quickly. Seems like these startups with high valuations, are more tenuous even than the startups of years past.
Could be. Could also be that LLM is to OpenAI what information retrieval is to Google. A lot is publicly known in the information retrieval space, but google still dominates.
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#18Is the text prompt input the only common thing between SD and MIT models when comparing outputs? If so I am surprised at how SIMILAR the outputs of both models look in the general layout / framing / composition of the image. How can for example both fox astronaut images have near identical backdrop of earth and earth alone on the same side of image at same apparent size. Virtually the same shade of deep blue for deep…
I believe this is similar to the Latent Consistency Modeling approach, where it’s a replacement for the “diffusion” process, not the underlying weights. Basically, they have a more efficient process for pulling images out of the weights, not necessarily a set of new weights.
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#19Am I missing it or is the source code not available?
I don't think they open sourced it. Thinking back, most AI breakthrough papers I've read don't include source code unfortunately. Researchers want their names out there explaining what they did and that they did it first, but MIT might want to license the implementation IP or Adobe (sounds like their interns discovered this during a summer) lawyers might be hanging onto it for a business edge.
Pretty ironic given this comment at the end of the page
> While image generators like DALL-E and Meta AI's Imagine can produce extremely impressive results, these groups are highly protective of their technology and jealously guard it from curious public eyes. Meanwhile, you can go read about MIT's findings at this link.[0]
Re: MIT Unveils Gen AI Tool That Generates High Res Images 30 Times Faster
#20[dead]