iiinova — where your digital self keeps evolving

Acfusion @acfusion.bsky.social · 17/09/2026

Gunbuster's ending always hit me like a truck when I go back to watch it.

010 0

Acfusion @acfusion.bsky.social · 04/08/2026

Damn didn't realize you were a fellow NC resident that's also to be into fighting games.

000 0

Acfusion @acfusion.bsky.social · 21/06/2026

I'm sorry but the fact godot is open find AI assisted contributions is a part of the problemm we shoudn't be excused in anyway. AI assisted assets are just as much as a problem as ones that's generated entirely especially in the context of the current state of Gen AI.

120 0

Acfusion @acfusion.bsky.social · 21/06/2026

The thing is that while Godot isn't outright pro-AI the fact they still allow AI gen code in the first place under any reason is enough of a red flag to me. As soon you start allowing it is going to snowball what ever you like it or not.

150 0

Acfusion @acfusion.bsky.social · 10/06/2026

Not to mention when I mean the windows I mean the ones in the gap between the sign like you see in the model there. You can clearly see windows on top in the model version but it was omitted & have this random patch in the image.

110 0

Acfusion @acfusion.bsky.social · 10/06/2026

A photoshop filter wouldn't do this. it wouldn't added in details that weren't there in the original design suck as the fucked up smearing on the base level building on top of the gap where there should be windows. Same issues show up with the clock tower.

230 0

Acfusion @acfusion.bsky.social · 10/06/2026

My bad in terms of that, but even if we assume that would be the case there's still a lot constant smearing & random details on the that doesn't even exist on the tower if you compare the model & 2D image. I'm sorry I just but I think there are more red flags in a lot of aspects of this box art.

100 0

Acfusion @acfusion.bsky.social · 07/06/2026

By the way the title is clickbait BS here what he actually said & it's far from what the title suggest.

640154 0

Acfusion @acfusion.bsky.social · 26/05/2026

The funny part of all this is there is already a video title & thumbnail that's pretty much this.

9802 0

Acfusion @acfusion.bsky.social · 25/05/2026

@roadietripp.bsky.social go ahead and tell me what your defense for shit like this ? Being autistic & trans doesn't give you the asshole card and it certainly doesn't give her the right to act like she's innocent & meanwhile acting like a snake.

100 0

Acfusion @acfusion.bsky.social · 25/05/2026

No he doesn't and it really highlights you don't care about queer people as much as you think. Just want to sling mud

120 0

Acfusion @acfusion.bsky.social · 08/05/2026

That's nothing to do with references.The problem is that they were doing this along with every other evidence. There's already Image 2 Image software on top of Gen Ai model can & has produced work that is simlar to existing works.

010 0

Acfusion @acfusion.bsky.social · 30/04/2026

No not really. It can has shown to directly copy stuff.

110 0

Acfusion @acfusion.bsky.social · 26/03/2026

Just posting this here to let people know she said this to a black person

000 0

Acfusion @acfusion.bsky.social · 06/03/2026

They literally using comfyUl which is connected to Gen AI.

000 0

Acfusion @acfusion.bsky.social · 10/02/2026

You can't say it's nonsense when that's kinda of the point. People will assume it makes no sense and that it's not generated because the amount of layers shown they had already been caught before on top of that. Scamers like them doing what they can to muddy up the waters. x.com/kariha_moon5...

100 0

Acfusion @acfusion.bsky.social · 05/02/2026

It did about as well as most previous metroid had performed sales wise so I'll never get why people took it as selling over a million a bad sign.

110 0

Acfusion @acfusion.bsky.social · 30/01/2026

First off this is false it's been shown that images in the dataset do become apart of the model and has been shown they can produce images that were part of the dataset. Also it's not copyright maximalist to expect to have some form of consent or permision. spectrum.ieee.org/midjourney-c...

110 0

Acfusion @acfusion.bsky.social · 30/01/2026

That's also false there's no such thing as a model that is competely trained off their own model. Large companies weren't able to do this. Not to mention it's trained under art that was stolen online via pretraining it. futurism.com/artificial-i...

000 0

Acfusion @acfusion.bsky.social · 30/01/2026

Nope not possible futurism.com/artificial-i...

220 0

Acfusion @acfusion.bsky.social · 30/01/2026

No it's obviously not first off becuase of this. 2nd it's not even possible to train exclusively on your own work in the first place. futurism.com/artificial-i...

120 0

Acfusion @acfusion.bsky.social · 10/01/2026

Also giving your data to ChatGPT is stil using other people's work as training data sincet ChatGPT is built on top of data that was scraped online. A functional model will not happen as long you need a so much data for it to work from scratch.

080 0

Acfusion @acfusion.bsky.social · 10/01/2026

Every time some one proclaims to do this is either not be honest about it or doesn't know how training it works.

000 0

Acfusion @acfusion.bsky.social · 10/01/2026

One is that it's not possible for people to have the data to make a model that's just their own data. Anyone who claims to done that are either lying or don't understand what goes into training. So no it's not trained only your own work it's built on top of pretrained work that was stolen.

120 0

Acfusion @acfusion.bsky.social · 09/01/2026

Here's a graph that explains what I mean by this often when people say they are using their own data it's often a smaller group of data finetuned from a base model that's still built on top of stolen data.

000 0

Acfusion @acfusion.bsky.social · 09/01/2026

what they said about using their own data to make assets is outright false you can't make Gen AI assets out from your own data it's just finetuned. Reminder anyone who is saying this they are using their own data is either lying or don't understand how training works.

3685 0

Acfusion @acfusion.bsky.social · 09/01/2026

Also what they said at the end is outright false you can't make Gen AI assets out from your own data it's just finetuned. Reminder anyone who is saying they are using their own data is either lying or don't understand how training works.

161 0

Acfusion @acfusion.bsky.social · 30/12/2025

No the article mentions it's not the only problem and is making it worse. It literally explains how the water is used. Again this you trying hard to discredit it some from trying to assume it was clickbait to persuming everyone in the article is lying.

000 0

Acfusion @acfusion.bsky.social · 26/12/2025

They are using a model built on stolen works that's only fine tuned and can't be seen as okay.

100 0

Acfusion @acfusion.bsky.social · 26/12/2025

Here's a statement from the people behind both Glaze & Nightshade about the recent devlopment.

000 0

Acfusion @acfusion.bsky.social · 22/12/2025

Can you actually read ? They used Gen AI during devlopment. Even assuming if all the placeholder textures are all gone that still marks for disqualification since they still used it.

110 0

Acfusion @acfusion.bsky.social · 22/12/2025

No they didn't they were up front since the start No Gen AI used during devlopment. It's literally apart of the FAQ unless you going to tell me they edit it at last second then give me some proof otherwise you are just digging your heels over nothing.

100 0

Acfusion @acfusion.bsky.social · 18/11/2025

Tried my hand on the trend too. #Frieren

020 0

Acfusion @acfusion.bsky.social · 16/11/2025

Mainly want to mention this because I often seen people say they trained using their own work when questioning the image models they used.

000 0

Acfusion @acfusion.bsky.social · 04/11/2025

Give a source on that & even that were true once again you being this guy and that isn't a good arugement point.

100 0

Acfusion @acfusion.bsky.social · 29/09/2025

Hey I'm an Artist here & code mystics was wrong about generated assets not being used. I already checked both files & the cabnet one is obviously had either AI assets or used an Gen AI tool on top of the draft they posted. 1/X

3251103 0

Acfusion @acfusion.bsky.social · 29/09/2025

Fair warning they achivements do use generated assets I already check the files. They either lying or being lied too.

0122 0

Acfusion @acfusion.bsky.social · 24/08/2025

Sorry to drop in out of nowhere. Keep in mind that they already pointed this out on their website's Q&A. nightshade.cs.uchicago.edu/faq.html

120 0

Acfusion @acfusion.bsky.social · 20/08/2025

And when it comes to the Ad itself the very fact this was it's introduction removes any back pedaling that happens latter in said AI. No matter how you slice it that's inital statement is their foot forward. And beyond again there's plenty of other times AI companies had done this type of Ads.

100 0

Acfusion @acfusion.bsky.social · 14/05/2025

Post a robot not from Starwars or Transformers Well, no one said not from Gundam soooo. How about some Kapool

160 0

Acfusion @acfusion.bsky.social · 04/05/2025

The source comes from the author's guild themselves. Also they already literally back by google & none of them are there for are professional authors. Please actually read the thing.

020 0

Acfusion @acfusion.bsky.social · 21/04/2025

It was a image of people trying to boycott over wanting to have dedicated servers. But over half of them are playing the game anyways.

130 0

Acfusion @acfusion.bsky.social · 28/03/2025

There's no denying that even in the website they are expecting that to happen. I'm just pointing out that img2img transfers aren't the area glaze & nightshade focus on it's at the training stage. nightshade.cs.uchicago.edu/whatis.html

110 0

Acfusion @acfusion.bsky.social · 28/03/2025

I need further context on what did they do. before I make the judgement due to how often I seen people claim to broke glaze. If it's img2img than they didn't broke it.

0130 0