Why don't indie devs use AI-generated images as art?

[luj1]

We are not using it because it looks like horseshit

 

[Humanophage]

We are not using it because it looks like horseshit

I think I'm beginning to understand why someone decided it was cool to use 3D heads and drop portraits in NWN2. People have the oddest conceptions.

 

[Narushima]

Automatically generated things are the opposite of creation and originality, which are usually a source of pride for creators and what people look for.

 

[luj1]

NwN 1 shits on NwN 2

 

[Peachcurl]

Well... are they public domain?

In USA, there's nothing limiting them from being released as such. There's been a few landmark cases that have found machine learning to be a transformative work .

You still have to look at the platform. Not necessarily all of them place the generated images in public domain.

And things may become a bit tricky, as in the TOS from artflow:

All content created on the Platform is licensed under the most accommodating Creative Commons license **CC BY** that allows both commercial and non-commercial use, including derivative work, as long as you attribute Artflow. This does not however cover the right to use someone's likeness for other than editorial use (see below).

In a situation where an image resembles a real individual, the person in question retains full publicity rights to their likeness. They, or their representative can submit a request to take down the image and Artflow will abide to those request regardless of the image creators intentions. Even in a situation where such a request was not submitted, you do not hold the right to use someone's likeness for other than editorial purposes.

https://artflow.ai/about#tos

 

[whydoibother]

This is just to save money. You already generate the 3D faces, might as well use them and not pay an artist for 2D portraits. Its being cheap.

 

[The_Sloth_Sleeps]

I've not actually seen anything remotely capable, with non curated results. I've only seen YT videos/academic papers which supposed outputs which are to be treated as lies until you try the program yourself

The whole OP are images generated within a few minutes. What is wrong with this image, for example, that makes it worse than typical indie game art? Why are the two images on the right capable, while the image on the left is not? To my mind, it looks just fine, even very good for an indie game. It's not like they are selected at random - rather, the writer generates appropriate images and picks which ones fit the narrative, the monster, the character, the interface screen, etc. They don't have to be picked at random as some procedural exercise.

Existing art already looks even more like everything is melting, even in more high budget games.

Why not skip the middleman and just generate entire games?

Because it's impossible, while creating decent AI-generated images is perfectly possible. Surely it's more fitting than stuff like that or terrible 3D heads.

For a start, as I said unless *you* have personally generated that image on the left I would not trust it to be real/indicative of typical performance. This sort of stuff is heavily curated and only good looking results shown (unfortunately these kind of selective demos is common place in Academia/Industry these days, e.g. Boston Dynamics).

Second, consistency will always be a problem. Do you really want to deal with inconsistency?

Third, it will be less headache to just get a human to do the art for you.

Having said that I'm sure it would be possible to make some kind of game from the computer generated "art" but I think its a fringe use case.

 

[Humanophage]

Automatically generated things are the opposite of creation and originality, which are usually a source of pride for creators and what people look for.

Not if they assist the creator and allow the individual author to tailor everything to his liking instead of being constrained by budget. Ideally, making a game should become as accessible as writing a book.

It also has the additional bonus of reducing the influence of SJWs on games, since artists tend to be that and they try to sneak in that trash all the time (see the artistic collapse from AD&D through D&D 3E to Pathfinder).

Suppose your party goes into a forest. Does it matter if it is illustrated by images in the upper rows or the images in the bottom rows? Personally, I think it looks dramatically better than most renders and better than most amateur art.

 

[HeroMarine]

So all I have to do to generate infinite images is to join this "Midjourney" site an pay $30 per month?

 

[Ol' Willy]

Arcanum with AI generated portraits

 

[Late Bloomer]

Nevergrind Online touts using AI to generate its character portraits.

Choose from more than 200 unique avatars drawn by artflow.ai, an avatar tool powered by artificial intelligence and machine learning! This allows for an exciting future with unlimited avatar variety!

https://store.steampowered.com/app/853450/Nevergrind_Online/

 

[gurugeorge]

Because it's not art.

I could see it being used very well for variations on human-made art or snapshots of models (e.g. bloodied portrait face, changing mood of portrait, changing background as a result of the ravages of war, etc.), IOW to add variety; but other than that, it doesn't have a use on its own.

 

[Humanophage]

For a start, as I said unless *you* have personally generated that image on the left I would not trust it to be real/indicative of typical performance. This sort of stuff is heavily curated and only good looking results shown (unfortunately these kind of selective demos is common place in Academia/Industry these days, e.g. Boston Dynamics).

Second, consistency will always be a problem. Do you really want to deal with inconsistency?

Third, it will be less headache to just get a human to do the art for you.

Having said that I'm sure it would be possible to make some kind of game from the computer generated "art" but I think its a fringe use case.

Naturally, you would be using the good results rather than any random result. It's a replacement for hiring an artist, not a procedural gimmick. If decent artists were so cheap, everyone could be using them in abundance.

I have personally seen the image on the left being generated on the chat. I used similar prompts and received similar imagery.

Consistency is just fine. For comparison, two "atmospheric dark elf villages in the sunset, in the style of Edmund Dulac" (I like Edmund Dulac) and same prompt, but a goblin village. First the basic images to upscale, then the upscaled images. Seems quite consistent, and also consistent with the Wizard/Goblin in the OP. Could be made a bit more dissimilar, but that's a matter of regenerating it or adding some words to the prompt.

 

[J1M]

Humanophage

The last time I looked into this was the Dream app. I thought it would be easier to work with an AI than to tease out correct results from an Arts grad. I suspect the problem I ran into is the same as others: most of the companies that have built this tech both have not thought of how to license the output in a way that enables commercial use AND are terrified of leaving money on the table, so the terms they are willing to work with until they figure out how to price commercial use are unclear and allow them to arbitrarily revoke the license at any time.

 

[rusty_shackleford]

Because it's not art.

Cope
Nobody cares if it's not made by humans except for art*sts put out of a job.

 

[gurugeorge]

Because it's not art.

Cope
Nobody cares if it's not made by humans except for art*sts put out of a job.

It doesn't matter whether they care, it matters only whether they find it attractive or repulsive, whether they tend to buy it or tend to not buy it so much. At the end of the day, AI stuff is just uncannily off most of the time. I mean you get a few things that pass, but not many, or not for long. It's maybe a bit easier with graphics because graphics themselves (even the human variety) are only just barely art when they're at their best, they're more utilitarian generally speaking. But actual art, poetry and writing, forget about it. It can pass muster for a while, but then there's always something that sticks out as demonstrating the thing doesn't actually understand what it's doing.

 

[rusty_shackleford]

Because it's not art.

Cope
Nobody cares if it's not made by humans except for art*sts put out of a job.

It doesn't matter whether they care, it matters only whether they find it attractive or repulsive, whether they tend to buy it or tend to not buy it so much. At the end of the day, AI stuff is just uncannily off most of the time. I mean you get a few things that pass, but not many, or not for long. It's maybe a bit easier with graphics because graphics themselves (even the human variety) are only just barely art when they're at their best, they're more utilitarian generally speaking. But actual art, poetry and writing, forget about it. It can pass muster for a while, but then there's always something that sticks out as demonstrating the thing doesn't actually understand what it's doing.

stuff in op is more attractive than 99% of """art""" in video games for the past decade.

 

[HumanTotemPole]

Because it's not art.

Cope
Nobody cares if it's not made by humans except for art*sts put out of a job.

It doesn't matter whether they care, it matters only whether they find it attractive or repulsive, whether they tend to buy it or tend to not buy it so much. At the end of the day, AI stuff is just uncannily off most of the time. I mean you get a few things that pass, but not many, or not for long. It's maybe a bit easier with graphics because graphics themselves (even the human variety) are only just barely art when they're at their best, they're more utilitarian generally speaking. But actual art, poetry and writing, forget about it. It can pass muster for a while, but then there's always something that sticks out as demonstrating the thing doesn't actually understand what it's doing.

Judging by the aesthetic illiteracy evident in this and the AI voice acting thread a lot of gamers won't know or care about the difference between human-made and AI generated assests. Discernment has been boilied alive by Anglophone cosumer culture.

Ebert vindicated from beyond the grave:

"But for most gamers, video games represent a loss of those precious hours we have available to make ourselves more cultured, civilized and empathetic"

 

[rusty_shackleford]

Discernment has been boilied alive by Anglophone cosumer culture.

do I sense some anger at anglo superiority?

 

[The_Sloth_Sleeps]

For a start, as I said unless *you* have personally generated that image on the left I would not trust it to be real/indicative of typical performance. This sort of stuff is heavily curated and only good looking results shown (unfortunately these kind of selective demos is common place in Academia/Industry these days, e.g. Boston Dynamics).

Second, consistency will always be a problem. Do you really want to deal with inconsistency?

Third, it will be less headache to just get a human to do the art for you.

Having said that I'm sure it would be possible to make some kind of game from the computer generated "art" but I think its a fringe use case.

Naturally, you would be using the good results rather than any random result. It's a replacement for hiring an artist, not a procedural gimmick. If decent artists were so cheap, everyone could be using them in abundance.

I have personally seen the image on the left being generated on the chat. I used similar prompts and received similar imagery.

Consistency is just fine. For comparison, two "atmospheric dark elf villages in the sunset, in the style of Edmund Dulac" (I like Edmund Dulac) and same prompt, but a goblin village. First the basic images to upscale, then the upscaled images. Seems quite consistent, and also consistent with the Wizard/Goblin in the OP. Could be made a bit more dissimilar, but that's a matter of regenerating it or adding some words to the prompt.

I don't understand your erection for this stuff. Its just doing (essentially) random generation based off existing samples its been fed. Garbage in garbage out.

Its not going to be consistent in that if you "need" it make some change (for example make a continuation in that style across the whole game) that it will make the desired change because it understands absolutely nothing about...anything.

 

[HumanTotemPole]

Discernment has been boilied alive by Anglophone cosumer culture.

do I sense some anger at anglo superiority?

More specifically the pervasive NPC consumer aspect of it Rusty. Makes dullards of us all.

 

[rusty_shackleford]

Discernment has been boilied alive by Anglophone cosumer culture.

do I sense some anger at anglo superiority?

More specifically the pervasive NPC consumer aspect of it Rusty. Makes dullards of us all.

might want to check what kind of products asians love to consoom
apparently you think mobile games & gacha is superior

 

[HumanTotemPole]

Discernment has been boilied alive by Anglophone cosumer culture.

do I sense some anger at anglo superiority?

More specifically the pervasive NPC consumer aspect of it Rusty. Makes dullards of us all.

might want to check what kind of products asians love to consoom
apparently you think mobile games & gacha is superior

Barking up the wrong tree here Rusty, well aware of consumerism's status as Earth's state religion.

 

[rusty_shackleford]

codexers are like

"wow this is so much better than anything computers could make because of the HUMAN factor..."

 

[HeroMarine]

codexers are like

"wow this is so much better than anything computers could make because of the HUMAN factor..."