- Sep 19, 2019
- 260
- 300
- Thread starter TwistedScarlett60
- Start date
-
- Tags
- 2d game 2dcg adventure animated big ass big tits cheating combat corruption creampie fantasy futa/trans futa/trans protagonist group sex handjob harem humor internal view lesbian management masturbation milf mobile game monster girl multiple protagonist oral sex pregnancy religion rpg sex toys titfuck transformation trap turn based combat vaginal sex virgin
- Apr 22, 2020
- 488
- 286
I'm sorry, but an uncaught exception occurred.
Uh?
Code:
While running game code:
File "game/6 - Hangouts Emi.rpy", line 1198, in script call
call dareend
File "game/1 - Dungeon.rpy", line 6, in script call
call dungeonsetup from _call_dungeonsetup
File "game/2 - Manor.rpy", line 7, in script call
call morningevent from _call_morningevent
File "game/6 - MorningEvents.rpy", line 3518, in script
if emi6 == 1:
File "game/6 - MorningEvents.rpy", line 3518, in <module>
if emi6 == 1:
NameError: name 'emi6' is not definedYou don't have permission to view the spoiler content.
Log in or register now.
- Apr 18, 2022
- 5
- 4
- Mar 24, 2020
- 5
- 7
- Apr 22, 2020
- 488
- 286
- Nov 23, 2022
- 199
- 376
- Mar 24, 2020
- 5
- 7
I'm 85% sure that the maids and "Hero" were also created by the AI. Of course, the author then finalized them in his own style, but lights, shadows and design is very characteristic of the AI. Also fingers. In the prologue, you can see that sometimes the maids are missing some.
Go compare to MTA or FWB. This has been the dev's art style since long before AI artwork came on the scene, at least in the capacity it has now.
Just rechecked the prologue, all fingers are accounted for, no idea why you think some are missing. Closest is Ophelia who's obviously just got her pinky tucked into her elbow.
- Apr 2, 2022
- 68
- 64
I've seen a lot of talk about the art of this game and AI, and I don't know why I'm bothering to do this but I'm just going to say that I'm pretty much positive the art in this game is AI traced.
If you just look at it with your eyes, the art looks exactly like every other AI image that gets made, especially the ones that get posted here. Plenty of people have noticed it, and the primary defense of this I've seen is that Scarlet has been an illustrator for years and has a massive back catalogue of art, and a lot of it is very good. It's exactly the sort of art that maybe AI models are trained off of, so maybe it's actually the AI copying Scarlet and not the other way around? They linked their Pixiv, but I checked their twitter instead, since everything is stamped with a date. The thing that really surprised me is just how often they upload. Like, new drawing every single day, at the absolute most every other day. This is an insane pace for the level of quality and effort. I get that they're a full-time artist at this point from their Patreon, but this is not a normal pace. But the real thing that struck me is how everything looks completely different from their previous work, Monster Girl 1000. For reference, I attached a picture of the Steam Banner Image of Monster Girl 1000 (Scarlet1)
Looking over Scarlet's work, it's kind of clear that there's two different brands of images that get posted. One of them is images that are traced, and the other one is images that aren't traced (they're usually images from the game), and they look like they're made by two completely different people. The attached Scarlet2 is a drawing that is like...clearly drawn by hand. I think they drew that one themselves, no help. Scarlet3 is a drawing from one month later. I just don't think they improved that much in such a short amount of time. They went from not really understanding how abs work and really basic shading to suddenly having a fully rendered body? I think the image is traced. Now, let's also contextualize this a little bit. Scarlet already openly uses AI art in this game for the backgrounds and certain characters, and they've said as much. But, in looking around a little bit, I also found a link to one of their Patreon posts. I can't link it, but you can find it by Googling "2B Mini-Set." In it, they talk about using StableDiffusion to generate images and then painting over them. In the comments, they talk about how this would be a really good way to generate base images to paint over, but laments that the tech is a little hard to use. That post was made in October 2022, so way before most people had an eye for what AI art looked like and could identify it easily. One of their tweets in August 2022 is about how they've started generating lewd AI images. So, someone who has been in on the ground floor of generating AI and has openly advocated for its use suddenly gets significantly better at drawing at around the point where StableDiffusion becomes readily available and easily usable. They also said that the paintover process for their initial tests took around two hours per image, which makes the upload rate make significantly more sense.
There's a few other reasons why I don't think they completely drew the images themselves. There's a specific genre of artist that have cropped up in the last few years that I don't know how to describe other than "tracing, but then had to draw the head themselves". They draw in an anime style, but use things like cosplay or model photography to draw the body, so it looks really good, but then they can't exactly trace anime eyes or hair, so they have to draw that part themselves, and it gives away the illusion. Scarlet4 is a good example of this - you can just tell the jacket was traced (either from photo reference or AI) but then the head doesn't match that level of detail at all. Inconsistent detail levels is another big thing that sticks out to me, especially when it comes to hands. Which is like, the one thing that the AI is well-known to be bad at generating. The hands look totally fine and make sense in the older images (1 and 2) because it matches the rest of the image. But all of a sudden, when you contrast that to the rest of the painstaking rendering in Scarlet5, it looks comically out of place.
So, big issue with this theory though. All this being considered, if you go back to even before the most baseline of AI tools were developed, they were still making images in a kind of similar style to what they're making now. It's not as good, but that can just be explained as getting better over time, right? Well, there's still the issues I mentioned above which persist even back then - heavily varied styles, weird disconnects between detail levels. Sometimes they know how anatomy and perspective works, sometimes they don't. So, what's the deal? Have they ALWAYS been tracing?
Well, yeah. They've been doing this for years, since at least 2018, according to the sixth image. People have called them out on it before but they basically just did then what they do now - say, "oh, well, sure I used it as an inspiration, but I changed some stuff so it wasn't traced." But like...come on. The general outline is changed a little bit, sure, but the big thing is that the general shading and tonal patterns are exactly the same. Which is, you know, the actual really difficult and time-consuming part of rendering out something like this. The hands are directly traced, again, revealing Scarlet doesn't really know how to draw hands, and we also see the exact same strategy of "tracing, but had to draw the head themselves." So, it totally makes sense now why their art in the games they released is so different and how their style changes so much - it depends what image they're "referencing." I'm sure for their older stuff you'd be able to go back and find the original version of plenty of their works if you wanted to do the detective work, but I'm sure that totally coincidentally around October 2022 you are going to, conveniently for Scarlet, not be able to find any original reference images anymore, which proves its not traced, right? And people publically called them out for tracing in the past, which I'm sure was inconvenient when they're trying to develop a business making 20k a month on Patreon, right?
So yeah, I think everything here points to the art in this game being AI traced.
If you just look at it with your eyes, the art looks exactly like every other AI image that gets made, especially the ones that get posted here. Plenty of people have noticed it, and the primary defense of this I've seen is that Scarlet has been an illustrator for years and has a massive back catalogue of art, and a lot of it is very good. It's exactly the sort of art that maybe AI models are trained off of, so maybe it's actually the AI copying Scarlet and not the other way around? They linked their Pixiv, but I checked their twitter instead, since everything is stamped with a date. The thing that really surprised me is just how often they upload. Like, new drawing every single day, at the absolute most every other day. This is an insane pace for the level of quality and effort. I get that they're a full-time artist at this point from their Patreon, but this is not a normal pace. But the real thing that struck me is how everything looks completely different from their previous work, Monster Girl 1000. For reference, I attached a picture of the Steam Banner Image of Monster Girl 1000 (Scarlet1)
Looking over Scarlet's work, it's kind of clear that there's two different brands of images that get posted. One of them is images that are traced, and the other one is images that aren't traced (they're usually images from the game), and they look like they're made by two completely different people. The attached Scarlet2 is a drawing that is like...clearly drawn by hand. I think they drew that one themselves, no help. Scarlet3 is a drawing from one month later. I just don't think they improved that much in such a short amount of time. They went from not really understanding how abs work and really basic shading to suddenly having a fully rendered body? I think the image is traced. Now, let's also contextualize this a little bit. Scarlet already openly uses AI art in this game for the backgrounds and certain characters, and they've said as much. But, in looking around a little bit, I also found a link to one of their Patreon posts. I can't link it, but you can find it by Googling "2B Mini-Set." In it, they talk about using StableDiffusion to generate images and then painting over them. In the comments, they talk about how this would be a really good way to generate base images to paint over, but laments that the tech is a little hard to use. That post was made in October 2022, so way before most people had an eye for what AI art looked like and could identify it easily. One of their tweets in August 2022 is about how they've started generating lewd AI images. So, someone who has been in on the ground floor of generating AI and has openly advocated for its use suddenly gets significantly better at drawing at around the point where StableDiffusion becomes readily available and easily usable. They also said that the paintover process for their initial tests took around two hours per image, which makes the upload rate make significantly more sense.
There's a few other reasons why I don't think they completely drew the images themselves. There's a specific genre of artist that have cropped up in the last few years that I don't know how to describe other than "tracing, but then had to draw the head themselves". They draw in an anime style, but use things like cosplay or model photography to draw the body, so it looks really good, but then they can't exactly trace anime eyes or hair, so they have to draw that part themselves, and it gives away the illusion. Scarlet4 is a good example of this - you can just tell the jacket was traced (either from photo reference or AI) but then the head doesn't match that level of detail at all. Inconsistent detail levels is another big thing that sticks out to me, especially when it comes to hands. Which is like, the one thing that the AI is well-known to be bad at generating. The hands look totally fine and make sense in the older images (1 and 2) because it matches the rest of the image. But all of a sudden, when you contrast that to the rest of the painstaking rendering in Scarlet5, it looks comically out of place.
So, big issue with this theory though. All this being considered, if you go back to even before the most baseline of AI tools were developed, they were still making images in a kind of similar style to what they're making now. It's not as good, but that can just be explained as getting better over time, right? Well, there's still the issues I mentioned above which persist even back then - heavily varied styles, weird disconnects between detail levels. Sometimes they know how anatomy and perspective works, sometimes they don't. So, what's the deal? Have they ALWAYS been tracing?
Well, yeah. They've been doing this for years, since at least 2018, according to the sixth image. People have called them out on it before but they basically just did then what they do now - say, "oh, well, sure I used it as an inspiration, but I changed some stuff so it wasn't traced." But like...come on. The general outline is changed a little bit, sure, but the big thing is that the general shading and tonal patterns are exactly the same. Which is, you know, the actual really difficult and time-consuming part of rendering out something like this. The hands are directly traced, again, revealing Scarlet doesn't really know how to draw hands, and we also see the exact same strategy of "tracing, but had to draw the head themselves." So, it totally makes sense now why their art in the games they released is so different and how their style changes so much - it depends what image they're "referencing." I'm sure for their older stuff you'd be able to go back and find the original version of plenty of their works if you wanted to do the detective work, but I'm sure that totally coincidentally around October 2022 you are going to, conveniently for Scarlet, not be able to find any original reference images anymore, which proves its not traced, right? And people publically called them out for tracing in the past, which I'm sure was inconvenient when they're trying to develop a business making 20k a month on Patreon, right?
So yeah, I think everything here points to the art in this game being AI traced.
- Apr 22, 2020
- 488
- 286
- Apr 22, 2020
- 488
- 286
I've personally seen artists make the mistake of adding a 6th finger to a characters hand so...
Since the next update is the big story-ending one, I wanted to make sure I fixed a few bugs in the meantime. Every bug mentioned in this thread has been fixed. There were a few suggestions I liked such as a "purchase 5x button" and the combat log automatically scrolling to the bottom, but I ended up running into a lot of difficulties trying to implement them, since I'm only an amateur coder.
OP has been updated to 0.9.0c:
Minor:
The next update won't be the last, but it's pretty close. We'll have a few post-game things to finish up such as Octavia's route, some what-if scenarios for Beatrice and Seraphina and of course, Queenie herself
OP has been updated to 0.9.0c:
Minor:
- Combat log is now draggable in addition to scrollable
- All upgades with 50+ or infinite levels now have a +5x purchase button appear if you can afford it.
- Added an erect variation of Lily's office attire
- Fixed Hazel's final hangout requiring two more Seraphina hangouts than necessary
- Fixed the potion event not showing the first four possible options
- Fixed Demon Encampment offering tons of MP on the MP potion.
- Spells no longer glow if you don't have them, but you can still get reactions from these.
- Entering nothing on the market screen no longer results in a crash
- Various small fixes
The next update won't be the last, but it's pretty close. We'll have a few post-game things to finish up such as Octavia's route, some what-if scenarios for Beatrice and Seraphina and of course, Queenie herself
Oh, and since we are coming to the end of Unveiling the Unknown's development cycle, your feedback is more important than ever! I read all suggestions and implement what I can to try and make the game the best it can be.
A huge thanks to everyone that has left reviews, bug reports and critique so far. This game couldn't have become what it is now without your help. As I said in my Monster Girl 1,000 thread, sometimes I feel like my games are one massive collab with the vast community it builds up ♡
A huge thanks to everyone that has left reviews, bug reports and critique so far. This game couldn't have become what it is now without your help. As I said in my Monster Girl 1,000 thread, sometimes I feel like my games are one massive collab with the vast community it builds up ♡
- May 6, 2017
- 384
- 356
Any plans to build out Lady Balethorns route? would love properly truning her against the Hero.
- Mar 24, 2020
- 5
- 7
The lack of fingers isn't really noticeable, I only pointed that out in tandem with the rest. But there's something about the lighting and the shadows of the characters. Like an uncanny valley effect. Also, if the faces aren't particularly different from the classic author's style, the bodies and costumes are very different. Notice the glare in the generated picture, in the picture from the old artwork, and from this game
Maybe I'm just crazy. But I'm assuming artist generated picture and then did some work on it.
- Jan 13, 2020
- 220
- 233
Same, i really hope Moma Balethorns gets more more scenes she is GYAAT DAMN HOT.