Thank you, you see , this is how healthy conversation goes. I assumed he doesn’t back you up since i never see him in your thread, you explained the situation. So again, thank you.
- Thread starter CircleGames
- Start date
Another problem is that in the months between updates a million clones of this game have popped up to the point of oversaturation. So even while it was the original that every AI dev is ripping off, if we are only getting a chapter every 4 to 5 months, by the time we get Chapter 6, there will be 200 games that are really similar, and hype will start to die down. While most of the copycats are slop, as AI gets better and easier to use, a few creators will inevitably make something pretty good, and start getting more attention. It places more pressure on each release to be a home run.
I’m not criticizing the developer here, but I would like to share my opinion. Games that typically take around six months to develop are usually choice-based, offering multiple paths and a complex storyline. In this case, however, the story is linear.
Additionally, the artwork is AI-generated(
While I’m not implying that it’s easier but it is more efficient compared to traditional artistic drawing & coloring or graphics rendering in some cases)
The developer mentioned upgrading their PC with better hardware for faster image generation, and at the same time, AI-generated art has significantly improved.
Also, not sure if anyone noticed it but looking at some of the art from the newer version, there is a slight decline in quality and inconsistency in the faces compared to the older versions.
Now, looking at the data below, you can see the issue. Let's see what v0.5 brings to the table and hope it would shut my mouth.
Last edited:
A main difficulty I see with AI games is that if the script gets more complicated, then the AI demands past a simple 1 or 2 character image become difficult: so we see many AI games abandoned ( awaiting tech catch up ) or finished early with a simple route. Those of you who have tried AI scene creation will know that simple images are improving all the time, but more complex scenes involve many failures and re-generation.
Plus yes, there are new similar AI games weekly now, as you would expect from a self learning tool, the iterative process accelerates with volume demand.
Very true. Especially in the NTS genre, it'll eventually involve multiple character scenes. Making every individual character feel alive and part of the scene is a difficult feat. Circle having an art background really does help and the work he's done in Chapter 5 is very impressive so far. There's a lot of things Pony or SD can do, but there's even more that it can't do, which is why a lot of the games coming out... feel the same? They have the same poses, same facial expressions, ect...
Then you have the games like Moonripple, Mila and MNGF who take things a step further with a lot of custom poses/expressions. I mean this rule counts for every game, anyone can render a scene in Daz and anyone can render an image through AI, but there's a steep learning curve and those who spend the time to really get into it, are able to produce some amazing results.
Even something like this, which might seem pretty straight forward... is incredible hard to generate through regular means. As Pony will always try to make it into a kiss or french kiss. Licking someones cheek? It doesn't know that, so you have to force it in through other means. (Obviously this picture has issues that need to be fixed, but it was more as an example).
- Sep 13, 2020
- 618
- 572
This pressure's a bad thing?
You started me thinginck. Sounds like you're describing the prototypical growth of sophistication of any field. A breakthrough inspires more ventures which begin experimenting with the new potentials. As to what you're implicitly lamenting, formulated in capitalist terms, is that the most viable in the end might well not be the original venture? I can't remember but there's a very prominent old example - was it ketchup? and maybe modern stuff like google, windows and facebook might count- but Galileo's fame's a non0business parallel. IDK what i'm talking about... you've tickled interesting stuff, but I'm uncomfortable with something in the whole dynamic and I don't like that business success is involved in the value of objective achievement...
I am wondering why they're all NTS based too?
- Apr 27, 2024
- 70
- 353
this is exactly why tracking development time is so important for these games. because these problems are inherent to AI development, and there is no obvious solution. as supporters, i think it's important to know "uh oh, this game is starting to hit the limitations of the dev/AI engine," because as we have seen many, many, many times now, this is usually the first step toward a slowdown and abandonment.
of course, the dev (or his many defenders in this thread) will come in and say, "what? no! i'm not going to abandon this!" but nobody admits to abandoning their games beforehand. they always just slow development to a crawl while continuing to take patreon money.
this game has a few more flashing warning signs: it seems like circlegames is franchising out his art style/AI generation techniques, which means there are at least 2 other games with very, very similar art styles to this game, all in the early stages of development, all of which will eventually hit the same "multi-person sex scene AI ceiling" and slowdown around the same time. the authors of these games are very quick to create patreons and pump out AI demos with paper-doll art and bog-standard poses they got from civitai, but the proof is in the payoff: will ANY of these games get to the part of the game where the NTS is being fulfilled, or will they all milk the teasing stage, because it's the only part the AI can handle?
While some people might not agree, but learning to be your own artist and making your own art, not only can you do more but there are less restrictions. It might take a bit longer, but with how energy inefficient a render is with AI. It might be worth just getting into an self-art school or drawing over the art yourself and fixing it instead of praying an AI rendering will get it right with a shotgun method.
Learning something like photoshop or GIMP and just putting poses together of different characters while also learning how to do put two subjects together. Then overtime you can add more and more characters or multiple subjects for a single scene. Instead of rendering dozens of people for an AI the AI is instead focusing on one or two subjects at a time, then you put them together in photoshop with a rendered scene.
Typically what we did was we had two renders going on, one person did one character and basically sequentially (seperately done) added in different pieces of characters and they added lighting at the very end. Especially with rendering which is half the process, it is easier to add in those characters like that than trying to do all of them at once. Cause lighting will basically tie the whole piece together. While also retaining the same type of art style and also basically creating a bones and model for it to use. (3d modelling does this very easy) so instead of rendering everything at once we rendered each individually then combined them in post.
I would suggest for most people unaware of the process.
1. Create Assets
2. Pose Assets
3. Create background
4. Place Lighting overtop assets
5. Render Lighting / Shadows (easily done by hand)
There is solutions but it just means actually learning the same process artists do to make them or game designers. You'll notice with a lot of older RPG maker games they had an image of someones face have it be blank, They would then switch around the eyes or add blushing. Or add details. This can easily be done with LIVE2d or other tools where you just render one character once, and change the face around to suit the subject. This would cut down on development time, and then you can do the big RENDERs of CGs. While also letting you focus on the big moments of renders instead of rendering each scene individually.
It would just mean learning these development processes that have been standard for the last twenty years to twenty five years. I mean no insult in that regard if you have not learned game development in college or studied it you will know about breakpoints and choke points when it comes to development processes, and learning which parts are the most time consuming to develop.
I know a famous project group on here for Hreinn games the reason they had so much trouble with their current project was burning the bridge with their artist and getting TOO many art stuff before actually finishing the game. Last Soverign I know has largely avoided this by a large array of gameplay. Or many of the other complicated projects on here. They have very solid base gameplay that lets people actually play the game while also having very relatively fewer CGs. Essentially a quality over quantity approach.
CG games might be stuck on the multiple scenes with various characters, which most artists IRL don't do because of how complicated they are to do.
I am not taking shots at anyone, not to CG or other devs. But as a consumer and sometimes potential buyer i don't think devs have any right to complain about AI's limitations (while it is true), since i don't see them crying for using a software which technically steals real artists' works.
I have defended devs when some criticized them for using an AI. But i will revile any dev who uses limitations of AI as an excuse. In the end, i don't care if AI is stealing someone else's work or its capacity is limited, i just want frequent updates.
I have defended devs when some criticized them for using an AI. But i will revile any dev who uses limitations of AI as an excuse. In the end, i don't care if AI is stealing someone else's work or its capacity is limited, i just want frequent updates.
Last edited:
- Jul 3, 2021
- 2,039
- 12,663
ControlNet gets rid of most of the issues people are discussing here as it allows to generate and manipulate images to a degree simple prompts couldn't achieve. The real limitation is hardware imo, but with the new 50 series that will be less of a problem. AI is constantly improving aswell. If you compare images from 2 years ago with what's available today, there are worlds between. Video generation is making massive leaps aswell.
TLDR; I don't think we have to be concerned about the game being abandoned over AI limitations.
Also, from what I've been told, CG is doing lots of it by hand. He's not just typing prompts and crossing his fingers.
TLDR; I don't think we have to be concerned about the game being abandoned over AI limitations.
Also, from what I've been told, CG is doing lots of it by hand. He's not just typing prompts and crossing his fingers.
Eh I think just simply photoshoping things together is much easier, if you want to improve the pipeline, along with just shifting how much images you have to generate.
AI generation is not efficient by its very nature. Unless you have a dozen render computers ready to go. The best solution is to use those resources sparringly and use them for the big moments and render a single image and various poses and character movements instead of generating a new image for every motion that is possible.
So that way you can hit the moments you want to hit with limiting the toolkit and not over using a generation thing. When we made our game project we auto-generated hundreds of items and buildings with a toolkit that let us customzie how many windows, and floors for a building.
All these were to be used in the background and didn't have physics attached to them. So they had low priority, a dev of ours literally just made the tool for us, so the level designer could populate the city, while the important objects were handmade as those are the major things a player is more concerned about. A player does not care if a lightpost is slightly sinking into the ground, or that the bottom texture that is invisible to a player glitches out.
As indie devs you gotta be smart with your resources. If a process takes too long its time to investigate other methods of quickening the process.
The sounds of the messages on the mobile in the game, are they the sounds of the application in reality? Because I find it a lot in other games
U have driven our girls in nervous wracks… they need time now to recover… respect this pls!
- Sep 13, 2020
- 618
- 572
My last post said, let's declare a truce until 1st Feb, then we can all judge from actual game released,
I guess no truce is wanted ,,
I guess no truce is wanted ,,
- Jun 1, 2019
- 479
- 2,945
VIP Booth
A little more privacy is always nice!! I will keep you guys posted about the progress on the update. I can't confirm the release date yet, everyday until friday is going to be crucial but either way we are getting pretty close to the end no matter what.
You don't have permission to view the spoiler content.
Log in or register now.
- Jul 3, 2021
- 2,039
- 12,663
What is Vi up to I wonder?! Haven't seen her in those club previews I believe? Surely she's up to no good either way..
- May 20, 2017
- 839
- 1,319
I reas booth and immediately though of those gloryhole cabins. Wouldn't mind if she was passing some time in there. HahahaWhat is Vi up to I wonder?! Haven't seen her in those club previews I believe? Surely she's up to no good either way..