Theme editor

  • RequestStream Instant movie, tv show and anime streaming platform • Trial for 1 week
  • Unfortunately, ATF is shutting down.
    It’s genuinely sad to see it go.

    If you’re joining us from ATF, welcome to LewdCorner. Please read our rules here: LewdCorner Rules

    If you want to talk about ATF and the shutdown, use this thread: Anti-Social Lurker Hideout Spot

AI Art in Games - Discussion Thread

  • Thread starter Thread starter Gertz
  • Start date Start date
  • Replies Replies 153
  • Views Views 9K

Gertz

Registered
Lewd
Joined
Nov 29, 2022
Threads
1
Messages
52
Vouches
0
Likes
199
Activity Coin
267
Donation Coin
0
Platinum Coin
0
1/3
‎3 Years of Service‎
Thread owner
I've seen quite a few games (mostly visual novels) with AI generated art being made and there seems to be more each passing day (see AI Art Games).

What's your opinion on AI art in games at its current state?

I personally feel like most of them struggle with one or many of the following points:
  • Maintaining a Unified Artistic Vision: There's often a noticeable disconnect between the art of the background and the foreground. Characters just feel out of place sometimes.
  • Character Consistency Issues: Characters also tend to morph. Their proportions and the level of detail changes between scenes
  • Lack of Environmental Continuity: Even if the background and foreground (characters, objects, etc) are rendered separately, iterative changes on a given scene/environment (say a new hidden object appears in the background) are non-existent in a lot of AI games.
  • Animated Scenes: Visual novels that use another game engine to render the scenes can also easily generate smooth animated scenes. AI visual novels just can't.
This is often why I avoid AI art like a plague in games, individual images are fine but it breaks my immersion instantly whenever it is used within a game.
 
The crux is,
Most devs won´t put much work in AI art. You have to put much photoshop hours in good AI atm. If you do it. It looks amazing.
So it is like every game. It depends on the work of the dev.
Lazy dev horrible game. Hard worker dev and creative. Good game.
 
I've seen quite a few games (mostly visual novels) with AI generated art being made and there seems to be more each passing day (see AI Art Games).

What's your opinion on AI art in games at its current state?

I personally feel like most of them struggle with one or many of the following points:
  • Maintaining a Unified Artistic Vision: There's often a noticeable disconnect between the art of the background and the foreground. Characters just feel out of place sometimes.
  • Character Consistency Issues: Characters also tend to morph. Their proportions and the level of detail changes between scenes
  • Lack of Environmental Continuity: Even if the background and foreground (characters, objects, etc) are rendered separately, iterative changes on a given scene/environment (say a new hidden object appears in the background) are non-existent in a lot of AI games.
  • Animated Scenes: Visual novels that use another game engine to render the scenes can also easily generate smooth animated scenes. AI visual novels just can't.
This is often why I avoid AI art like a plague in games, individual images are fine but it breaks my immersion instantly whenever it is used within a game.
You listed most of the reasons I avoid games that use AI art with the big one for me being I plain just don't like the look of it
 
Thread owner
The crux is,
Most devs won´t put much work in AI art. You have to put much photoshop hours in good AI atm. If you do it. It looks amazing.
So it is like every game. It depends on the work of the dev.
Lazy dev horrible game. Hard worker dev and creative. Good game.

I agree up to an extent. I think if the dev just can't outright make any art by themselves. No matter how lazy/productive they are, the game would still suffer from the points I mentioned above. I think this is where an experienced artist makes a big difference.
 
I've seen quite a few games (mostly visual novels) with AI generated art being made and there seems to be more each passing day (see AI Art Games).

What's your opinion on AI art in games at its current state?

I personally feel like most of them struggle with one or many of the following points:
  • Maintaining a Unified Artistic Vision: There's often a noticeable disconnect between the art of the background and the foreground. Characters just feel out of place sometimes.
  • Character Consistency Issues: Characters also tend to morph. Their proportions and the level of detail changes between scenes
  • Lack of Environmental Continuity: Even if the background and foreground (characters, objects, etc) are rendered separately, iterative changes on a given scene/environment (say a new hidden object appears in the background) are non-existent in a lot of AI games.
  • Animated Scenes: Visual novels that use another game engine to render the scenes can also easily generate smooth animated scenes. AI visual novels just can't.
This is often why I avoid AI art like a plague in games, individual images are fine but it breaks my immersion instantly whenever it is used within a game.
Even as someone who mucks with AI for fun, I avoid it as much as I can. 99% of it looks bad to me, and I just don't like the work ethic behind it. Like, some devs are coders first and foremost, and may not have the means to do actual art, and I do get that and sympathize, but AI is not the way to go about it. I also have a tendency to prefer 3D over 2D to begin with (except in niche cases where the Devs art tries to stick to the style of a ""real"" series, like early Roguelike:evolution, before Oni started cutting corners, 18Titans, Four Elements trainer, or Something Unlimited). Ultimately, I just don't want to see it, its fine to play with for fun, but AI I don't feel has a place being used "seriously" anywhere.
 
When it comes to AI-Art, the options are endless! I'm all about those super realistic vibes.
But it's like you either get a developer who's a genius AI-Artist or a storytelling wizard.
I've come across games with epic plots but cringy AI-Art, and the other way around too!
Guess it's time to unleash the teamwork magic!
 
I've seen quite a few games (mostly visual novels) with AI generated art being made and there seems to be more each passing day (see AI Art Games).

What's your opinion on AI art in games at its current state?

I personally feel like most of them struggle with one or many of the following points:
  • Maintaining a Unified Artistic Vision: There's often a noticeable disconnect between the art of the background and the foreground. Characters just feel out of place sometimes.
  • Character Consistency Issues: Characters also tend to morph. Their proportions and the level of detail changes between scenes
  • Lack of Environmental Continuity: Even if the background and foreground (characters, objects, etc) are rendered separately, iterative changes on a given scene/environment (say a new hidden object appears in the background) are non-existent in a lot of AI games.
  • Animated Scenes: Visual novels that use another game engine to render the scenes can also easily generate smooth animated scenes. AI visual novels just can't.
This is often why I avoid AI art like a plague in games, individual images are fine but it breaks my immersion instantly whenever it is used within a game.
Very interesting. I'm learning more about AI art reading your post. I think many people myself included had the perception that AI is so advanced that it could write/draw the perfect VN and it's artwork. You make some very good points on how AI struggles. Is that because of the limits of AI or more because of the mistakes of the developer? Also I'd like to try these VNs to better understand what you mean. Would you please suggest a few examples? Thank you
 
Very interesting. I'm learning more about AI art reading your post. I think many people myself included had the perception that AI is so advanced that it could write/draw the perfect VN and it's artwork. You make some very good points on how AI struggles. Is that because of the limits of AI or more because of the mistakes of the developer? Also I'd like to try these VNs to better understand what you mean. Would you please suggest a few examples? Thank you
The post reads almost like it was written by those AI chatbots.
LLMs, bleh
 
As long as it's a text game, then still AI images don't bother me so much. If it's an RPGmaker game with AI art for characters, I wouldn't bother at all.
That said, AI art filling in some niches such as background art wouldn't bother me that much.
 
Even now it's almost impossible to tell if it's AI or not for high quality graphics, the limitation is more the access to the tools. I'd try to make a game if creating AI graphics was less of a drag.
 
Even now it's almost impossible to tell if it's AI or not for high quality graphics, the limitation is more the access to the tools. I'd try to make a game if creating AI graphics was less of a drag.
Is it? I feel that a little bit of exposure to AI art either through games under the AI tag on here or pics from civit.ai will make you be able to instantly recognize the style.
Installing Stable Diffusion on your own PC is quite straightforward nowadays, though generating something decent would take time and patience.
 
Is it? I feel that a little bit of exposure to AI art either through games under the AI tag on here or pics from civit.ai will make you be able to instantly recognize the style.
Installing Stable Diffusion on your own PC is quite straightforward nowadays, though generating something decent would take time and patience.
Not just knowledge (mostly tho) my laptop is also a potato. I could try on my PC but the card is older too.
 
Not just knowledge (mostly tho) my laptop is also a potato. I could try on my PC but the card is older too.
Same here, but you might be able to run it in "lowvram" mode. Could be worth a shot, if you're curious.
 
Same here, but you might be able to run it in "lowvram" mode. Could be worth a shot, if you're curious.
I just read a guide how to go around the NSFW filter in Stable Diffusion and it says 6 VRAM are needed. I'm gonna try to find out if lower RAMs also work.
 
I just read a guide how to go around the NSFW filter in Stable Diffusion and it says 6 VRAM are needed. I'm gonna try to find out if lower RAMs also work.
Never heard that there's a filter and thought that it's based on the model you're using. Here's some instructions about the lowvram command and other optimizations you could try.

In addition to that, you might have to first generate the image, and then run the upscaler separately, so it doesn't crash in the process.
 
Never heard that there's a filter and thought that it's based on the model you're using. Here's some instructions about the lowvram command and other optimizations you could try.

In addition to that, you might have to first generate the image, and then run the upscaler separately, so it doesn't crash in the process.
Thanks, that helps a ton. I think my card has 4 GB. I'll try this soon
 
Back
Top Bottom