brucethemoose

joined 2 years ago
[–] brucethemoose@lemmy.world 2 points 42 minutes ago* (last edited 42 minutes ago) (1 children)

I dont think any spitting is necessary, nor is it productive.

[–] brucethemoose@lemmy.world 1 points 55 minutes ago* (last edited 52 minutes ago)

Some models do use open datasets and largely reproducible training, with the Nvidia Nemotron series being the highest profile example. You are right that many are opaque about training data, though, and this is not okay.

Using them is just as unethical as using the models from OpenAI.

And I vehemently disagree with this.

OpenAI is a different order of magnitude of moral depravity. Thats like saying using Lemmy is just as bad as using Facebook or Palantir.

[–] brucethemoose@lemmy.world 2 points 1 hour ago

That is called luma noise, yes. Heh. Its nothing new.

The thing that I wish was better understood is that “slop” = “low effort.” ...that means that not all AI is slop, and thus if you really want to detect AI, you can’t rely on it always being slop.

Yes! I would wear this on my forehead on Lemmy, if I could.

[–] brucethemoose@lemmy.world 3 points 1 hour ago

I mean, the customers are the ones with the purchasing power here, and most of the shame was put on the studio.

[–] brucethemoose@lemmy.world 3 points 2 hours ago* (last edited 1 hour ago) (2 children)

Also... you could just add grain via post processing?

Its actually a reasonable technique, as it softens diffusion blocking artifacts the same way it softens video codec blocking artifacts.


The YouTuber sorta kinda has a point, depending on the context. Spammers tend to use the absolute laziest lowest common denomenator, so its quite possible theyre largely using the same model, with the same artifacts, and not bothering to edit output. So maybe they did have a common noise pattern? Though the way thats presented still sounds fishy to me.

[–] brucethemoose@lemmy.world 2 points 2 hours ago (3 children)

You put this incredibly abrasively, but there is a grain of truth there. Gamers, in aggregate, repeatedly preorder or buy hyped garbage, so perhaps we dont have much of a leg to stand on when it comes to boycotts.

[–] brucethemoose@lemmy.world 4 points 2 hours ago* (last edited 2 hours ago) (2 children)

Well it is actually complicated. It appears the tiny studio ran out of time/budget trying to get usable CGI, so they did some diffusion real-quick as an act of desperation, and got slop as a result.

Not everyone is exactly up to speed on diffusion, either. I bet they Googled and used the first video diffusion service they found, and hence got bad results, but its also fair to criticize them for publishing them. They either didnt bother to proof the output, or they did and knowingly published anyway.

Im not sure anyone is the villain here. But I dont like people leaning into outrage culture, and trying to paint it as black-and-white.

[–] brucethemoose@lemmy.world 4 points 2 hours ago* (last edited 2 hours ago)

Unfortunately it may be politicized beyond recovery, at least in the US with the trajectory of things.

Its going to be associated with the corporations doing awful things with it, and thats going to blot out the sun.

[–] brucethemoose@lemmy.world 4 points 2 hours ago* (last edited 2 hours ago) (4 children)

AI has a certain look to it

I think this is a dangerous misconception.

I could feed locally run Minimax H3 a set of reference art, like the painted stills from this game, and it would exactly match the style. I could use start/end frames, motion masks, pose control, character sheets, voice and music samples, reference videos; I can get it to do whatever I want with extreme precision, and as long as I checked the output it would be indistinguishable from CGI.


The point I'm making is that a lot of people percieve diffusion as mashing a Stable Diffusion keyword prompt into a text box, and getting a sloppy result. Maybe thats still how shit corporate services like ChatGPT work.

But a LOT has changed over two years.

Theres no way people can definitively detect generative output anymore. We can IF its done lazily, but its quite possibly indistinguishable now, too.

[–] brucethemoose@lemmy.world 9 points 2 hours ago* (last edited 2 hours ago) (4 children)

Quality control, for one.

Artifacts like this would be unnaceptable if they were from CGI or whatever, but somehow generative models get a pass? That makes no sense to me, yet it happens all the time; heavily distorted output somehow meets the bar for publishing.

Honestly it has nothing to do with "AI." It would be worth calling out however it was made. But for some reason generative diffusion models always seem to lower the standards of output for user.


If it was indistinguishable from CGI and fit the rest of the hand-painted style, the argument would be very different, IMO.

[–] brucethemoose@lemmy.world 1 points 16 hours ago (1 children)

I guess people would rather eat shit than vegetables.

It's not entirely their choice. The algorithm decides what's presented.

[–] brucethemoose@lemmy.world 5 points 1 day ago* (last edited 1 day ago) (1 children)

It definitely wasn’t an intentional leak to generate buzz. Nope...

view more: next ›