Putting the 'role' back in role-playing games since 2002.
Donate to Codex
Good Old Games
  • Welcome to rpgcodex.net, a site dedicated to discussing computer based role-playing games in a free and open fashion. We're less strict than other forums, but please refer to the rules.

    "This message is awaiting moderator approval": All new users must pass through our moderation queue before they will be able to post normally. Until your account has "passed" your posts will only be visible to yourself (and moderators) until they are approved. Give us a week to get around to approving / deleting / ignoring your mundane opinion on crap before hassling us about it. Once you have passed the moderation period (think of it as a test), you will be able to post normally, just like all the other retards.

Is AI the future of Indie RPGs?

tritosine2k

Erudite
Joined
Dec 29, 2010
Messages
1,490

yeah so it goes, they just pulled some kinda pseudo (cardboard) depth out of the wizard hat
cool_story_bro.png
cool_story_bro.png
cool_story_bro.png
cool_story_bro.png
cool_story_bro.png

cool_story_bro.png
cool_story_bro.png
cool_story_bro.png
cool_story_bro.png
cool_story_bro.png

"infers the depth of an input image", lmao
 

Non-Edgy Gamer

Grand Dragon
Patron
Glory to Ukraine
Joined
Nov 6, 2020
Messages
14,946
Strap Yourselves In
yeah so it goes, they just pulled some kinda pseudo (cardboard) depth out of the wizard hat
This is just a 3d artist's shortcut using depth maps. This has nothing to do with AI. He does it in Photoshop and then puts it all together in a 3d program.
"infers the depth of an input image", lmao
And? It just draws a depth map, dude.

https://medium.com/@bilawal/stable-...on-of-depth-heres-why-it-matters-988c557e74dd
So how do we generate a depth map? Well, Stable Diffusion uses MiDaS for monocular depth estimation in their Depth2Image feature. MiDAS is a state-of-the-art model created by Intel and ETH Zurich researchers that can infer depth using a single 2D photo as an input.
Then it adds that to its denoising calculations in order to keep the basic shape of the image. That's it.

Sorry, I should put this in the strawman format you're expecting, since you'll probably do it anyway in your response:

>WOW STABLE DIFFUSION UNDERSTANDS THE VERY CONCEPT OF DEPTH AND MAKES 3D MODELS AUTOMATICALLY WITH MAGIC
>THIS REQUIRES NO HUMAN INTERACTION BECAUSE IT'S A REAL AI WITH AN IQ OF 10000
>STABLE DIFFUSION CAN SCULPT NEW FACES ONTO MT RUSHMORE IN REAL LIFE
>STABLE DIFFUSION WILL 3D PRINT SPACE SHIPS AND HOOKERS FOR YOU WITH REAL HYPERDRIVES AND BIRTH-CAPABLE VAGINAS IN BOTH

There. Several strawman arguments for you to respond to instead of what I and the press release are actually saying. Enjoy.
 

tritosine2k

Erudite
Joined
Dec 29, 2010
Messages
1,490
Interestingly in smartphone sensors they realize they can't get away without dedicated depth sensor, but leave it to "software engineers" to conjure and redditors to lap it up, with some hamfisted "muh img" " it was chicken first" "argument".
 

Non-Edgy Gamer

Grand Dragon
Patron
Glory to Ukraine
Joined
Nov 6, 2020
Messages
14,946
Strap Yourselves In
Interestingly in smartphone sensors they realize they can't to get away without dedicated depth sensor, but leave it to "software engineers" to conjure and redditors to lap it up, with some hamfisted "muh img" " it was chicken first" "argument".
What are you even saying?

It's Intel that invented the AI depth estimator. All Stability did was figure out how to make the product a factor in denoising calculations. It's a new feature because they presumably trained it into the model.
https://pytorch.org/hub/intelisl_midas_v2/
 

Non-Edgy Gamer

Grand Dragon
Patron
Glory to Ukraine
Joined
Nov 6, 2020
Messages
14,946
Strap Yourselves In
It has the depth of a
cardboard
because
Monocular depth estimation is an underconstrained problem, i.e. geometrically it is impossible to determine the depth of each pixel in the image
Added that last bit 38 minutes later lol.

All it's there for is to keep the relative geometry and estimated depth in the new image. It doesn't need to be perfect. It's just a guideline to make changing the image easier.

Why does this make you so upset?
 

Tweed

Professional Kobold
Patron
Joined
Sep 27, 2018
Messages
2,876
Location
harsh circumstances
Pathfinder: Wrath
I am both afraid and excited for the the dark future of porn that is about to unfold.
lol, the "porn industry" is already flooded with shovelware games, badly animated short movies and hopw roewur ne cartoon. You're just going to see a triple amount of it, and more for even the most niche shit.
Build a better fetish and the world will beat its meat to your door.
 

tritosine2k

Erudite
Joined
Dec 29, 2010
Messages
1,490
Last edited:

RobotSquirrel

Arcane
Developer
Joined
Aug 9, 2020
Messages
1,960
Location
Adelaide
GPT3/ChatGPT is scary. The code it outputs is better than most you'd find in the official documentation for most game engines. It also comments like a champ and then outputs a small step-by-step instruction on what each part of the code does.
That's definitely going to be a massive change in workflow.

The other thing I like is that you can talk to it like its a real person, so if it does get something wrong you can say hey um that part wasn't right you need to correct it here and here. Ai goes back and fixes it.
Then you can run on a tangent, what if I want to take that idea and expand it to then support 3D Meshes? AI is like Oh ok well you just have to do this simple formula to convert it.
Ok can you make it look pretty though?
Sure I wrote this custom shader for you.
Nice.

The challenge is knowing very specifically what you want, how far you can run with it and what to do if things go horribly wrong.
Very cool AI.
The only negative is that it really sucks with politics. It's very biased. It will nearly always side with the Media.
will be nice when it can be run locally.
 

Kamaz

Pahris Entertainment
Developer
Joined
Feb 16, 2004
Messages
1,035
Location
The Glorious Ancient City of Loja
All of the character portraits in Space Wreck are AI-generated.

patrick.jpg


The real reason behind it is rights - real people's faces belong to them and it is complicated, and expensive to get their model signoff, especially for 100+ pictures.
Additionally, we have sexual deviants, pornstars, maniacs, pimps, and billionaires in the game and this may represent the model in a "bad light", so additional problems.

We used a variety of services - Generated Photos, ArtBreeder, custom, and even got a referral as a "use case" - https://generated.photos/solutions.
 

Non-Edgy Gamer

Grand Dragon
Patron
Glory to Ukraine
Joined
Nov 6, 2020
Messages
14,946
Strap Yourselves In
Yeah, I figured you might have, or at least I hoped that you had, given the art style. It'd be a waste to spend the money on a stock model and then distort it this much, or to pay someone to draw the pixel art from scratch.

In the old days, devs would just use people they knew or team members for small budget projects. Can you imagine what getting 100 people you know to pose for pictures and sign off on using them in a game would be like?
 

v1c70r14

Educated
Joined
Feb 8, 2023
Messages
161
Whether or not AI art is the future of indie RPGs I would like to see more developers experiment with it. If I can throw together something decent within a couple of minutes with Microsoft's free service then video game developers should be able to use this to bring half decent art back into video games. This isn't an excuse for me to demo the service and show off what I generated, video games should go back to having art like this and I think the places where people share their AI art is giving it a bad name.

qpC7pax.png

That's mostly an issue with the users and the sort of people at the wheel. Or maybe you think it's all anime girls. I'm into old book covers and nerd stuff like that so I'm going to show off what I've been making for fun testing the model, for free, and within minutes. It does have some issues with faces and hands, but paid models don't seem to suffer as much from that and you could always fix it in post. These are untouched and raw from the model.

oJQUzrz.jpg


r2ggJqU.jpg

BavGath.jpg

EVbjHj6.jpg

FMgkI80.jpg

nD4g2EN.jpg

aeRgiD3.jpg

jTPmu7a.jpg

hHf7svD.jpg

TU8VeVW.jpg

Owm7sC6.jpg

ekCFu4y.jpg

POBod19.jpg

5yrlbMG.jpg

ITLt062.jpg

nSPKvph.jpg

4z76Ok1.jpg

XsJyxx8.jpg

SIfw5mz.jpg

iECQjii.jpg

JggBV2g.jpg

NEPqGvO.jpg

FcRcUXg.jpg

3kBL5kA.jpg

Zi7RoSE.jpg

RtZMDS7.jpg

K2qZq7d.jpg

pjZgqD2.jpg

bSwEvnr.jpg

So if you need concept art or are using static images in your game, why haven't you outsourced your art process away from a greedy "artist" that is stealing everything from references anyway to a much cheaper and better computer model? Why doesn't your game have the cover art of a 90's softcover fantasy novel?
 

KeighnMcDeath

RPG Codex Boomer
Joined
Nov 23, 2016
Messages
13,058
"WHERE ARE HER NIPPLES?"

I'd love to visit some of those places. VR games gonna get crazy.
 

anvi

Prophet
Village Idiot
Joined
Oct 12, 2016
Messages
7,551
Location
Kelethin
I was trying to do the same recently but as video. My dream would be to record a video of me playing EverQuest and have the AI convert it into as photorealistic as it could. (So people would watch :P)
 

Krice

Arcane
Developer
Joined
May 29, 2010
Messages
1,330
GPT3/ChatGPT is scary. The code it outputs is better than most you'd find in the official documentation for most game engines.
It doesn't take much to write better code than in 99% of games. I don't know why people think GPT is "scary". If it can write some code it's just a positive thing, right? It gives more time for creativity and game design when you don't have to think about coding. But we still have to see how useful it is in practice and if we even want to always use it. Imagine if some large open source projects were cleaned up with AI... but I highly doubt it will happen, because open source developers are so stuck up how they want to write code. Not to mention if GPT even knows what good programming is, because it's "learning" from those sucky open source styles itself I guess.
 

v1c70r14

Educated
Joined
Feb 8, 2023
Messages
161
My dream would be to record a video of me playing EverQuest and have the AI convert it into as photorealistic as it could. (So people would watch :P)
szq6Rw3.jpg
 

As an Amazon Associate, rpgcodex.net earns from qualifying purchases.
Back
Top Bottom