this post was submitted on 16 Sep 2024
61 points (96.9% liked)

Asklemmy

43811 readers
853 users here now

A loosely moderated place to ask open-ended questions

Search asklemmy ๐Ÿ”

If your post meets the following criteria, it's welcome here!

  1. Open-ended question
  2. Not offensive: at this point, we do not have the bandwidth to moderate overtly political discussions. Assume best intent and be excellent to each other.
  3. Not regarding using or support for Lemmy: context, see the list of support communities and tools for finding communities below
  4. Not ad nauseam inducing: please make sure it is a question that would be new to most members
  5. An actual topic of discussion

Looking for support?

Looking for a community?

~Icon~ ~by~ ~@Double_A@discuss.tchncs.de~

founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[โ€“] ReCursing@lemmings.world 1 points 1 month ago (1 children)

No it doesn't. It's not theft by any reasonable definition of the word. No images are stored, no artwork is used directly to create other artwork. It';s just not, that's not how latent diffusion works. That's one of most commonly repeated pieces of bullshit which has been refuted so often you would have thought it'd have got through a few of your thick skulls by now.

[โ€“] tetris11@lemmy.ml 1 points 1 month ago (1 children)

(thanks for the insult, stay classy) so the network training stage was pulled out of thin air then? Huh, I didn't know these models could self-bootstrap themselves out of nothing.
I guess inverting models to do a tracing attack is impossible. Huh.

[โ€“] ReCursing@lemmings.world 1 points 1 month ago (1 children)

The insult is justified because you are spouting bollocks. Again. You CANNOT pull any of the training images out of a latent diffusion model, it is simply impossible because they are NOT THERE and if someone says they did they are either lying or spent a fuck of a lot of time and energy on making it look like they did. Either way they are trying to con you. Also the training thing - it's no different to art inspiring human artists except the neural network in the computer is a lot simpler. It's a new medium being used by humans, by artists, to create art. That's all it is.

I don't have the time or energy to explain any more of this to you. Again. Learn how something works before you comment again. Or just shut the fuck up for good. That works too.

[โ€“] tetris11@lemmy.ml 0 points 1 month ago* (last edited 1 month ago) (1 children)

(nice ad hominem) Christ. When you reduce a high dimensional object into an embedded space, yes you keep only the first N features, but those N features are the most variable, and the loadings they contain can be used to map back to (a very good) approximation of the source images. It's akin to reverse engineering a very lossy compression to something that (very strongly) resembles the source image (otherwise feature extraction wouldn't be useful), and it's entirely doable.

[โ€“] ReCursing@lemmings.world 1 points 1 month ago (1 children)

So you can't pull an image out as it went in? Because it's not stored there? Yeah that's what I FUCKING SAID! Stop spreading bullshit. Just stop it.

[โ€“] tetris11@lemmy.ml 0 points 1 month ago* (last edited 1 month ago)

(Ah, the joyful tantrum). Educate yourself on how a simple JPEG works and exactly how little features are needed to produce an image that is almost indistinguishable from the source.