Unity Abandoned Locus Tenebrarum [2022_12_13] [Rue]

[LocusTenebrarum_Rue]

Would be probably easier/faster to just run filters and touch-ups on real world pictures to produce similar "panting" effect, but with much better control over poses and appearances.

Some kind of style transfer GAN could probably do a decent job at that. That said, I enjoy the creative process of making art this way. I have some traditional art skills, and this still allows me to use a lot of them.

I don't doubt that someone will do this with style transfer at some point. I wish them luck, but I'm enjoying my current art process too much to switch.

 

[phuocdang98]

great art <3

 

[cooperdk]

It takes a lot of time and work. I spend a lot of time in Photoshop editing things, running them back through the AI, doing paint-overs, inpainting the face, hands, clothes, etc.

Things like Dreambooth can help, but it can be tricky. I have a backlog of stuff I need to do right now, but I'm planning to spend more time with Dreambooth once I've got time.

Ahh, so you do img2img and not txt2img - I suspect you feed it with images of your favourite porn stars or smth?

 

[LocusTenebrarum_Rue]

Ahh, so you do img2img and not txt2img - I suspect you feed it with images of your favourite porn stars or smth?

I do txt2img then edit with Photoshop, then feed that through img2img. The workflow differs depending on what I’m trying to accomplish.

Sometimes I’ll bring elements from multiple txt2img generations into Photoshop, blend them together, then run parts of that through img2img. Sometimes I’ll run the whole thing through an upscaler, take the result into Photoshop, then grab chunks of the image and run those through img2img. That can really help with coherence on hands and faces.

Sometimes I’ll even render a hand in Daz Studio, then photoshop that into an image in the right spot and run that through img2img. I don’t use images from the Internet because there are unresolved legal questions around that which I’d rather not have hanging over my game.

And of course there’s a lot of the liquify brush being used in Photoshop to move things around. Sometimes it’s also easier to just paint in limbs and things like that by hand.

If anyone wants to know my secret, it’s that you need to just keep reworking the images. Keep making little improvements to various parts. Know your way around Photoshop. Think outside the box and combine the tools in unexpected ways. Don’t be afraid to spend a lot of time on an image. I know everyone thinks AI art is as simple as finding a good prompt, but results like mine are the result of combining AI with traditional art skills and taking the time to get the image right.

 

[Ackerlight]

Great start to the game so far. Really impressed that you are able to use an AI to generate the 2D art. I have done AI for a while with different video projects (long time supporter of GRisk and Topaz tester since the first alpha), but have not tried Stable Diffusion yet. I will have to look into that one. I know GRisk has been doing a gui but have not played with it.

 

[BoyD.Joy]

Dang that was quick for someone to make a AI generated H-Game. Looks pretty interesting though, gonna check it out.

 

[lutra_lovegood]

Yeah, that’s a pretty good summary. It really does struggle with multiple characters. Any time I need an image with multiple characters, I’m probably going to have to spend a lot of time in photoshop re-working the image, doing paint-over, and running parts of the image back through the AI.

Have you tried generating two different images, cutting and pasting one onto the other and either feeding that into the AI again? I haven't tried that kind of thing myself but kitbashing can get you great results with AIs apparently.
E: Nevermind, saw the later comment.

As for content can we get some bsdm and corruption? And I think this is missing the exhib and voyeurism tags?

 

[Ptah]

Get ready for an increase of these kind of games.....

I'm even thinking of creating a furry, ntr, harem, BBC, game with this kind of software

Nah is not going to happen that much unless the software improves A LOT. I'm using this programs and is not as easy as people thinks It is... It still gives a lot of problems to maintain the same model for multiple creations.

 

[Ptah]

No specific hate for this project but I think it's a great showcase for why AI art will never replace real artists. You just can't get an AI to do the exact details you want, it's just about throwing enough *AHEM* at the wall until you find something that is close enough. I mean that penis looks completely deformed. And the fact that the different girls really look almost nothing alike between pictures, they even change eye color.

Again, this is just a point of view, not directed specifically against this project. And this is still better than the games that just photoshop images of real pornstars, ugh.

The thing is, you're WRONG, AI art doesn't replace real artists NOW, but I wouldn't be as preposterous as saying "will never"...

 

[PornSlut32]

I can get not wanting to hire artists in order to save money and time but have you considered hiring one to at least edit the paintings themselves so they can fix some of the tiny stuff and make the characters look more consistatn through out. I'm just seeing that the one black haired girl tends to look like a completely different person and while hiring someone to edit would cost you money I think it might be a good idea in the long run tho. Keep up with the decent work. But if you are ever looking for an artist to help you out with that I have a portfollio I'm more then willing to show off it it would help.

 

[retardedusername]

There’s a dream fragment with an extended scene with her that you find in one of the 3D environments. Be sure to find it and go to the Dream Emporium to talk to Talulah.

You get a tiny glimpse of that scene the first time you go to the Dream Emporium, but the fragment version has about a dozen images (including the version of the Patreon header image that I’m not allowed to post on Patreon).

A very good extra. The art is a little inconsistent at times but it's still pretty good. I kinda wish there were some options for the game like volume options would be nice and I hate when text comes out slowly... I prefer instant text because I'm a fast reader. Not really sure why there is an option to touch people that I just met especially if I can't touch them anyway... maybe it will be useful later down the line but I doubt it. I'm guessing this is your first game? not bad for a beginner... good luck!

 

[PornSlut32]

It takes a lot of time and work. I spend a lot of time in Photoshop editing things, running them back through the AI, doing paint-overs, inpainting the face, hands, clothes, etc.

Things like Dreambooth can help, but it can be tricky. I have a backlog of stuff I need to do right now, but I'm planning to spend more time with Dreambooth once I've got time.

Are these based off your own paintings?

 

[PornSlut32]

I do txt2img then edit with Photoshop, then feed that through img2img. The workflow differs depending on what I’m trying to accomplish.

Sometimes I’ll bring elements from multiple txt2img generations into Photoshop, blend them together, then run parts of that through img2img. Sometimes I’ll run the whole thing through an upscaler, take the result into Photoshop, then grab chunks of the image and run those through img2img. That can really help with coherence on hands and faces.

Sometimes I’ll even render a hand in Daz Studio, then photoshop that into an image in the right spot and run that through img2img. I don’t use images from the Internet because there are unresolved legal questions around that which I’d rather not have hanging over my game.

And of course there’s a lot of the liquify brush being used in Photoshop to move things around. Sometimes it’s also easier to just paint in limbs and things like that by hand.

If anyone wants to know my secret, it’s that you need to just keep reworking the images. Keep making little improvements to various parts. Know your way around Photoshop. Think outside the box and combine the tools in unexpected ways. Don’t be afraid to spend a lot of time on an image. I know everyone thinks AI art is as simple as finding a good prompt, but results like mine are the result of combining AI with traditional art skills and taking the time to get the image right.

NGL I completely respect that. It's how people should use AI art. Honestly and actually putting effor and work into it is a good thing. AI art can definitely be used to make some cool shit but it's cool to see someone once in a while that also put their own hard work into it. Kinda based.

 

[LocusTenebrarum_Rue]

A very good extra. The art is a little inconsistent at times but it's still pretty good. I kinda wish there were some options for the game like volume options would be nice and I hate when text comes out slowly... I prefer instant text because I'm a fast reader. Not really sure why there is an option to touch people that I just met especially if I can't touch them anyway... maybe it will be useful later down the line but I doubt it. I'm guessing this is your first game? not bad for a beginner... good luck!

I’m working on adding more options. I was going to add volume soon, but text speed is also a great idea. I’ll add that to my list.

 

[LocusTenebrarum_Rue]

NGL I completely respect that. It's how people should use AI art. I find that one of the major problems around AI art communities is people stealing other people art and then putting it out there as if it's just done, but honestly and actually putting effor and work into it is a good thing. AI art can definitely be used to make some cool shit but it's cool to see someone once in a while that also put their own hard work into it. Kinda based.

Yeah, I think of AI as being like power tools. I’m sure carpenters were concerned when power drills and electric saws were invented, but it didn’t kill carpentry. Instead the tools enabled skilled carpenters to make better things faster, but you still need the skill and experience to use them correctly.

This is a big change, and that’s really scary for a lot of artists. Some of my friends have been having an anxiety attack ever since DALL-E 2 was announced. So I want to be kind to these artists who are scared, but I think they’ll realize that these really are advanced tools that require an artist to fully utilize them. If anything, I’m optimistic that AI could bring about a golden age for artists.

 

[kewl smut]

Drawings are crazy clear. Kudos to the creator.

 

[Deniz31]

You must be registered to see the links

You must be registered to see the links

After nearly a month, I have a big release to announce! Here are the highlights:
* New save/load system so you can resume from your last session. This took a ton of work because I'm building out all of the core systems of the game myself, and I would expect some bugs to crop up for a little while. We'll get them ironed out over time.
* New introduction to the game. As I mentioned before, the previous versions didn't start at the beginning of the story. That's been resolved, and there's now a sequence at the beginning that explains the backstory.
* All of the narration text has been re-written. This was very time consuming, but worth it. Instead of using second person present tense (which is common in choose your own adventure novels and interactive fiction), I've switched to a first person past tense perspective, which is more common in novels. This allowed me to add a lot more detail to scenes, especially the lewd scenes. I think you're really going to like this change.
* Dream fragment #2 has been expanded and split up into smaller pieces to make it easier to get to parts of it. This also gives you more things to search for in the world. There are 9 completely new images.
* Part 1 of fragment #2 is now available to free players
* Many existing images have been reworked and improved.
* There's an intro screen when you start the game. It's a small thing, but it feels a lot more polished than launching straight into the game.
* Numerous bugs have been resolved involving interacting with things while various dialogs/menus are being shown.
* More settings! Now you can set the volume of music and sounds, along with the speed of the text. These settings are included in your save file and should be restored when you load from a save.
* Improvements to the debug console.
* You can hit escape to dismiss the settings menu (a small thing, but not doing it really irritated me)
* Fix bug with gallery where you couldn't view the same image twice in a row.

 

[gobman]

Now begins the age of machine.

 

[Kracken36]

Interesting idea as a project but it just seems to be testing AI.

 

[alexglig]

need tag "neuronet"