Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Stable Diffusion 2.0

stability.ai

Post reply on HN
Print view View original

341–350 of 519 posts

Re: Stable Diffusion 2.0

#341

Post by smeagull »

TaylorAlexander wrote: post #154
kmeisthax wrote: post #58

Is there a good explanation of how to train this from scratch with a custom dataset[0]? I've been looking around the documentation on Huggingface, but all I could find was either how to train unconditional U-Nets[1], or how to use the pretrained Stable Diffusion model to process image prompts (which I already know how to do). Writing a training loop for CLIP manually wound up with me banging against all sorts of stra…

Ah I am glad to see someone else talking about using public domain images! Honestly it baffles me that in all this discussion, I rarely see people discussing how to do this with appropriately licensed images. There are some pretty large datasets out there of public images, and doing so might even help encourage more people to contribute to open datasets. Also if the big ML companies HAD to use open images, they would…

Yup, it really is a good thing they haven't been forced to use open images.

Re: Stable Diffusion 2.0

#342

Post by smeagull »

LunaSea wrote: post #320
bowsamic wrote: post #228

Earlier quoted context omitted.

No one is ever going to stop using all the available images until there is a law against it. Why would they?

There are already laws against it but enforcement is laking, as always.

Can't wait for law enforcement to start arresting people for looking at images.

Re: Stable Diffusion 2.0

#343

Post by smeagull »

liuliu wrote: post #41

Seems the structure of UNet hasn't changed other than the text encoder input (768 to 1024). The biggest change is on the text encoder, switched from ViT-L14 to ViT-H14 and fine-tuned based on https://arxiv.org/pdf/2109.01903.pdf . Seems the 768-v model, if used properly, can substantially speed-up the generation, but not exactly sure yet. Seems straightforward to switch to 512-base model for my app next week.

Oh shit, I think that means it's CPU only for me now.

Re: Stable Diffusion 2.0

#344

Post by snowram »

jeroenhd wrote: post #266
NKosmatos wrote: post #199

Wow, just wow! Newbie question, why can’t someone just take a pre-trained model/network with all the settings/weights/whatever and run it on a different configuration (at a heavily reduced speed)? Isn’t it like a Blender/3D studio/Autocad file, where you can take the original 3D model and then render it using your own hardware? With my single GOU it will take days to raytrace a big scene, whereas someone with multipl…

The main limitation for running these AIs is that you need tons of VRAM available for your GPU to get any good performance out of them. I don't have a video card with 12GiB of VRAM and I don't know anyone who does. If you're willing to wait more (30 seconds per image, assuming limited image sizes) there are repositories that will run the model on the CPU instead, leveraging your much cheaper RAM. In theory you could…

12GiB VRAM cards are common places nowadays. A RTX3060 is around ~450$ and available to everyone.

Re: Stable Diffusion 2.0

#345

Post by postexitus »

LASR wrote: post #128
hackernewds wrote: post #53

Earlier quoted context omitted.

Strongly believe selection bias among the folks you're getting this impression from. The avg user is not our circle.

This is exactly it. Honestly I was quite surprised at how regular people are impressed by this tech. I was also surprised by how little regular people are aware of this tech even existing. We, on hackernews, on a thread about Stable Diffusion, are of course not too unimpressed. But that’s not the vast majority of people.

Your insistence on not understanding the point made makes me think you, my friend, are also an AI.

Re: Stable Diffusion 2.0

#346

Post by phtrivier »

ajuc wrote: post #325
acadapter wrote: post #235

Earlier quoted context omitted.

Human artists derive their inspiration and styles from a large set of copyrighted works, but they are free to produce new art despite of that. Art would have developed much slower and be much poorer if, for example, Impressionism or Cubism had been entangled in long ownership confrontations in courts. Then there's the fact that humanity has been able to develop and share art and literary works for thousands of years…

> It would be interesting to see if this technology can erode the copyright concept a bit Copyright law (especially in US) only ever changes in the direction that suits corporations. So - no. What I expect instead is artists being sued by a big tech company for copyright violations because that big tech company used the artist Public Domain image for training their copyrighted AI and as a result it created a copyrigh…

In this situation, it would seem like the suit would end up at "comparing the timestamp at which the public domain and copyrighted versions were published", wouldn't it ?

There is nothing that the generative AI can do in this process that's legally different from copy pasting the image, editing it a bit by hand, and somehow claiming intellectual property of the _initial_ image, no ?

Re: Stable Diffusion 2.0

#347

Post by fragmede »

foobazgt wrote: post #220
fragmede wrote: post #190

Earlier quoted context omitted.

I agree. Cars break down and crash, they'll never replace horses.

I think this analogy doesn't hold water - horses aren't exactly a beacon of reliability (having owned one). I've already seen tools that support workflows where you compose art by iteratively generating a piece of it, performing some correction, and repeating. So, I think there's room in the art world for less than perfectly generated art. That said, let's not kid ourselves that the typical failure modality of ML tod…

It's only an analogy, but it serves to underscore the last point you make. Initial versions of the technology can make some genuine horrors but you're blinding yourself to progress if you can't see the potential in it.

Re: Stable Diffusion 2.0

#348

Post by klqwart »

thot_experiment wrote: post #285
TaylorAlexander wrote: post #154

Earlier quoted context omitted.

Ah I am glad to see someone else talking about using public domain images! Honestly it baffles me that in all this discussion, I rarely see people discussing how to do this with appropriately licensed images. There are some pretty large datasets out there of public images, and doing so might even help encourage more people to contribute to open datasets. Also if the big ML companies HAD to use open images, they would…

The problem is not that people aren't owning ideas hard enough, ideas shouldn't be ownable in this way, the problem is that we've created a system that's obsessed with scarcity and collecting rents. Being able to own and trade ideas a la copyright/patents helps people who can buy copyrights and patents stifle creativity more than it helps artists gather reward for their creation (though it does both). Human endeavor…

Perhaps no one wants "your art"? 99% of artists who produce something worthwhile very much care about money/copyright.

The there still is the question of attribution, which 100% of real artists care about.

Re: Stable Diffusion 2.0

#349

Post by fragmede »

klqwart wrote: post #348
thot_experiment wrote: post #285

Earlier quoted context omitted.

The problem is not that people aren't owning ideas hard enough, ideas shouldn't be ownable in this way, the problem is that we've created a system that's obsessed with scarcity and collecting rents. Being able to own and trade ideas a la copyright/patents helps people who can buy copyrights and patents stifle creativity more than it helps artists gather reward for their creation (though it does both). Human endeavor…

Perhaps no one wants "your art"? 99% of artists who produce something worthwhile very much care about money/copyright. The there still is the question of attribution, which 100% of real artists care about.

So anybody who just wanted a thing to exist, and don't care who gets the credit, aren't "real artists"? You must not work on any large art projects that involve other people.

Re: Stable Diffusion 2.0

#350

Post by otabdeveloper4 »

FeepingCreature wrote: post #251
otabdeveloper4 wrote: post #236

Earlier quoted context omitted.

> when you consider that the model is a couple gigabytes itself, it can't memorize 240TB of data, so it "learned". This is just lossy compression with a large and well-tuned (to the expected problem domain) dictionary. Video compression codecs can achieve a 500x compression ratio, and they are general-purpose.

The dataset, LAION-5B, is 240TB of already compressed data. (5 billion pairs of text to 512x512 image.) Uncompressed, LAION-5B would be 4PB, for a compression ratio into SD of ~780kx, or one byte per picture.

The point is that there's is no practical limit on compression. You don't need "AI" or anything besides very basic statistics to get astronomical compression ratios. (See: "zip bomb".)

The only practical limit is the amount of information entropy in the source material, and if you're going to claim that internet pictures are particularly information-dense I'd need some evidence, because I don't believe you.

Post reply on HN