That’s some weird bokeh. What lens is this?

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

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.

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.

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

Whoa nellie.

This was many years ago, when I didnt know much about linux, and yes many lessons were learned. Though I distinctly remember instances where I couldnt update Manjaro packages because of expired certs.

To clarify some points:

• On packages/AUR, I was desperate, because upstream Arch had some critical fixes that werent shipped to Manjaro yet. Clearly this didnt work out well.

• AUR specifically breaks with Manjaro because AUR targets upstream arch as a base. When a dependant Arch package updates, the AUR package immediately updates to that new dependecy if its being maintained. But since Manjaro delays the upstream updates, this AUR package will break if its updated.

• Yes, I know AUR is filled with malware now.

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.

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.

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.

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.

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.

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.

I guess people would rather eat shit than vegetables.

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

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

Other species, too: https://en.wikipedia.org/wiki/Phragmosis

Its sleazy software, and IMO its defaults provide a really terrible 1st time experience. It convinces 1st time users that locally run LLMs are terrible, and goes out of its way to be incompatible with other stuff.

I think LM Studio would be a much better “1st time” recommendation. Its not perfect either, but its lightyears better, not an active con like ollama, and its smoother to “graduate” from it to croco.cpp or llama.cpp or something.

https://sleepingrobots.com/dreams/stop-using-ollama/

And thats just the tip of the iceberg.

Also, while Minstral is fine, it is pretty much obsolete…

You shouldn’t let that discourage you. VLLM is really finicky, but there are easier backends to use (though which one depends on your hardware).

Manjaro was more unstable than Arch ever was, for me.

My experience was:

• They would mess with Arch packages and muck things up.

• It relied on AUR some, which also messed things up.

• 3rd party repos or AUR didn’t work, as its not in sync with upstream arch.

• …Holding back packages didn’t fix anything anyway. It just delayed fixes. I found myself trying to install upstream Arch packages.

It turns out Arch Linux is more tested and “held back” than people give it credit for. There’s little to gain from forking like that, instead of just contributing to upstream arch testing.

Not to mention the cert thing. Manjaro really did let certs expire many times. When I was new to linux, that took me forever to figure out.