Is AI the future of Indie RPGs?

[The_Sloth_Sleeps]

...

Welll, I meant it's functions in principle as a tool, like pen is a tool - not that it functions exactly as a pen. The whole point is to cherrypick. That's why you run scripts that canvas over large grids of parameters, pick up the best result and then continue on from there. I see nothing in your propositions that would contradict what I said.

The whole point is to cherrypick.

that's giving it "feedback" because it's not intelligent on its own so you just took I out of AI and all that's left is artifical.

That's semantics, for what it is worth, I don't mind calling it artificial stupidity or whatever, the point is, it's an useful tool.

Could you give a demonstration of how this is a useful tool in the context of game asset generation?

Something plausible and generally useful and not an edge use case would be even better.

 

[Kev Inkline]

I guess this would be a fair and balanced take:

 

[tritosine2k]

Here's what they do to get variations of a specific output:



also tried it, doesn't work.

Similarly you can't even get the SAME sword that's rusty, glowing, sticks out of something etc. etc. because the thing holds no memory. Same with "backdrops" .

 

[Kev Inkline]

What do you mean, what doesn't work?

 

[tritosine2k]

What do you mean, what doesn't work?

it's not the same subject anymore. That's why stable diffusion videos are all messed up because theres no coherence. And in order to get coherence it will need some surface descriptor (3D) computational requirements will be thru the roof, that won't be the same "colab for free" affair either.

 

[Kev Inkline]

This is how I run the rough first grids - the same seed, varying the number of steps and a few different samplers. Now if it produces something interesting, I pick the given instance, and work on from there. Also might a body from one sample and the head from another, collate them in photoshop and start working with img2img.

 

[The_Sloth_Sleeps]

I guess this would be a fair and balanced take:

Interesting I guess that he found a way to use it. Yet also utterly ridiculous. You cannot use this in reasonably repeatable consistent workflow. Its like he is working against the image generation to get a result, it would be so much simpler to not use it. I just cant take this stuff seriously.

 

[Gromlintroid]

AI is lol for the next decade at least

 

[The_Sloth_Sleeps]

AI is lol for the next decade at least

"AI" is a defunct tech thats experienced a bit of a comeback due to higher computing power and larger data sets than what we had in the 80s. I fully expect it to fizzle out as quickly as it came back in. It might be that actually there is zero use for any of this rubbish - yet thanks to sales clowns this nonsense somehow finds its way into our cars and washing machines.

 

[Non-Edgy Gamer]

Nice try. I don't care how I achieve a result.

Which is why you aren't a "real artist" any more than a prompter is.

I did paint the scene - I just use texture brushes and stamps rather than pixel by pixel. I don't think it matters at all!

Cope.

My point isn't that I care if you use "AI", my point is it's that it's pretty shit at getting it to do anything consistently useful.

For tilesets. Yet. Which is what you've been whining about in every page of every thread about AI.

Meanwhile, generated in NovelAI as a joke and then rigged into a full vtuber model by some internet random:



The vtuber generated an entire agency's worth of models btw:

Spoiler

A non-artist being able to work through hundreds of usable concepts at zero cost in a single session.

You really need to be trying VERY hard not to see how this could be useful.

 

[Non-Edgy Gamer]

I guess this would be a fair and balanced take:

He's not even using the best workflow, which is in Krita. No back and forth, which he's obviously used to. It automatically masks generated images, so he could just select and regenerate parts of the object. It generates with multiple batch copies as well.

So, cut the first 18 minutes of the video down to 5 or 10 max.

 

[rusty_shackleford]

AI is lol for the next decade at least

"AI" is a defunct tech thats experienced a bit of a comeback due to higher computing power and larger data sets than what we had in the 80s. I fully expect it to fizzle out as quickly as it came back in. It might be that actually there is zero use for any of this rubbish - yet thanks to sales clowns this nonsense somehow finds its way into our cars and washing machines.

your posts give off the same kind of desperation as a 39 year old single mother on a dating app

 

[Non-Edgy Gamer]

your posts give off the same kind of desperation as a 39 year old single mother on a dating app

"You just can't handle a REAL woman!"

*cats meowing in the background*

 

[Kev Inkline]

I guess this would be a fair and balanced take:

He's not even using the best workflow, which is in Krita. No back and forth, which he's obviously used to. It automatically masks generated images, so he could just select and regenerate parts of the object. It generates with multiple batch copies as well.

So, cut the first 18 minutes of the video down to 5 or 10 max.

Yeah, a Krita plugin is pretty nice feat, now especiallly that they've got one working with AUTOMATIC1111 distro.

 

[The_Sloth_Sleeps]

Proomters seething. This stuff is ridiculous, you have just one video showing a workflow and it's just silly.

Proompter could at least dig up some more workflow videos if you are trying to convince people. So far as I can see, it's a joke.

 

[Non-Edgy Gamer]

Proompter could at least dig up some more workflow videos if you are trying to convince people.

I don't need to convince anyone. The results speak for themselves for the people who actually use them.

And you're never going to be convinced until your particular niche use case is solved for you. Even then, I doubt you'd do a 180 on AI publicly. You'd probably just use it quietly and drop out of all these threads.

 

[The_Sloth_Sleeps]

Proompter could at least dig up some more workflow videos if you are trying to convince people.

The results speak for themselves for the people who actually use them.

Indeed they do.

I have yet to find any convincing use case. Sure I do see a lot of lolli and furry portraits etc and occasionally some artist goes "this could be useful but Im not sure why" but thats about it. Theres no convincing evidence. No body of work done, no concrete pipeline for how to actually achieve an outcome.

It would be far more useful to have a tool that could accomplish the basics, over a patchy command line process that attempts grandiose artwork and utterly fails.

 

[Non-Edgy Gamer]

I have yet to find any convincing use case.

Plenty have been posted itt.

What you mean to say is that your personal use case of creating tilesets hasn't been met.

Theres no convincing evidence. No body of work done, no concrete pipeline for how to actually achieve an outcome.

Convincing evidence and body of work? What, do you want a peer-reviewed study?

"Oh no, no one has only a few people have made a game with it in the 3 months that it's been out. Gosh, that must mean it's useless!"

Numerous pipelines have been posted, including videos. Do you think people can't scroll up or look at other pages?

It would be far more useful to have a tool that could accomplish the basics, over a patchy command line process that attempts grandiose artwork and utterly fails.

It would be far more useful for you to work on your tilesets than spam every AI thread with your vapid doomer takes.

 

[The_Sloth_Sleeps]

I have yet to find any convincing use case.

Plenty have been posted itt.

What you mean to say is that your personal use case of creating tilesets hasn't been met.

Theres no convincing evidence. No body of work done, no concrete pipeline for how to actually achieve an outcome.

Convincing evidence and body of work? What, do you want a peer-reviewed study?

"Oh no, no one has only a few people have made a game with it in the 3 months that it's been out. Gosh, that must mean it's useless!"

Numerous pipelines have been posted, including videos. Do you think people can't scroll up or look at other pages?

It would be far more useful to have a tool that could accomplish the basics, over a patchy command line process that attempts grandiose artwork and utterly fails.

It would be far more useful for you to work on your tilesets than spam every AI thread with your vapid doomer takes.

No you just spam videos that have no connection to the topic.

I suspect they have been programmed heavily to specialize in face generation (which unsurprisingly is what it does best) and perhaps few other rudimentary scene principles like scene object classification. I dont believe it will ever advance beyond what it has been specifically programmed for. We would need endless programmed modules.

Perhaps the most glaring issue is that the proc generators have no idea of what form is. They are also incapable of applying knowledge across domains, ie no common sense.

Its difficult for me to see any promise for a system that lacks the fundamental requirements of being able to create useful art.

 

[Sarathiour]

So, cut the first 18 minutes of the video down to 5 or 10 max.

Are you that daft ?

It's an edited tutorial video, meaning he spend a bit more time explaining a few things, but he also speed up when nothing noteworthy is being shown. The 1 hour is by no mean supposed to be the real time it take to do such thing, "akshhually with the correct workflow you can gain 10 minutes" is a massive cope and missing the point.

 

[tritosine2k]

Well if a good thing comes out of this might be some investment proompter high on artistry starts seeing perspective in gamedev instead of waiting for overengineered AI or regurgitated stuff from embracer/ MS et al. Yet not only the connection between prompts and games is not clear, the whole prompting might be simply a red herring in grander scheme, but hey it's a start I'll give it that and it kinda has traction too.

 

[Non-Edgy Gamer]

No you just spam videos that have no connection to the topic.

I don't. They do. And they're a lot better than your posts which are the same thing every time:

"WAAAAAAAA! It doesn't do tilesets! It's not real art! I don't like it! WAAAAAAAAAAAAA!"

I suspect they have been programmed heavily to specialize in face generation

It depends on the model. People have trained their own. You know this. Item generation model results have been posted.

Its difficult for me to see any promise for a system that lacks the fundamental requirements of being able to create useful art.

Wow. Great opinion. Good job posting that again for the 100th time.

 

[Non-Edgy Gamer]

Are you that daft ?

Asks the other doom spammer with an AI-shaped stick up his ass.

It's an edited tutorial video, meaning he spend a bit more time explaining a few things

3 minutes. I actually watched the video.

but he also speed up when nothing noteworthy is being shown.

Which would make it take more time not less.

I'm talking about the actual time it would take using the Krita workflow. Since, I know. Because I actually use it and don't just hang out in these threads and doom post for whatever minor ego boost it might give you.

The 1 hour is by no mean supposed to be the real time it take to do such thing, "akshhually with the correct workflow you can gain 10 minutes" is a massive cope and missing the point.

Speaking of cope. Next time at least look at the time codes. It's all there: the parts he's using SD and the parts he's using Blender.

In reality, it takes 5 to 10 minutes to do exactly what he did to generate the image, if you use the better workflow in Krita. That's what I was talking about. That's why I said 18 minutes. Because I looked at the time codes for when he's working on generating the image.

But that doesn't matter to you, since all you see is another post about AI and you freak out. Calm down, dude.

 

[Non-Edgy Gamer]

Well if a good thing comes out of this might be some investment proompter high on artistry starts seeing perspective in gamedev instead of waiting for overengineered AI or regurgitated stuff from embracer/ MS et al. Yet not only the connection between prompts and games is not clear, the whole prompting might be simply a red herring in grander scheme, but hey it's a start I'll give it that and it kinda has traction too.

At present, it's a tool to cut down on time for artists, or cut down on the need for a skilled artist when a less-skilled one will do.

This can translate into hundreds or thousands of dollars in savings for a small project like a vtuber model or a visual novel.

It is a disruptive technology. It will affect the art industry. But it will not completely eliminate artists or anything. At least, not yet. Words can only communicate so much so precisely.

 

[The_Sloth_Sleeps]

Well if a good thing comes out of this might be some investment proompter high on artistry starts seeing perspective in gamedev instead of waiting for overengineered AI or regurgitated stuff from embracer/ MS et al. Yet not only the connection between prompts and games is not clear, the whole prompting might be simply a red herring in grander scheme, but hey it's a start I'll give it that and it kinda has traction too.

At present, it's a tool to cut down on time for artists, or cut down on the need for a skilled artist when a less-skilled one will do.

This can translate into hundreds or thousands of dollars in savings for a small project like a vtuber model or a visual novel.

It is a disruptive technology. It will affect the art industry. But it will not completely eliminate artists or anything. At least, not yet. Words can only communicate so much so precisely.

Dubious claims. I think it's fair to say I am a "less skilled" artist yet I can't make use of it. Nevertheless I don't even believe that's the niche here. I don't think anyone semi serious with even lower end of skill would use this.

I actually think the niche is shovelware. People who have zero ability to draw. Low quality games, comics, low effort stuff where turnover and churn are important. There's definitely a market there.