California Strip Poker
CSP AI-generated players?
I have been playing with AI-generated images for a few weeks now. I was curious to see how easy it might be to create assets for CSP that way intead of using DAZ Renders. DAZ Renders can take an hour to set up a shot, and then another hour to actually render it. Not to mention all those assets cost a lot. Turns out it certainly isn't as easy as they make it look, but gen-AI can do some pretty fancy things. I've been using Fooocus, which is free and easy to set up, and pretty quick if you have a decent graphics card.
The requirements for making a CSP opponent is that we need a bunch of pictures, in various states of undress, of the SAME PERSON, with the SAME FACE, and the SAME BODY, and the SAME CLOTHES, in the SAME PLACE. These are all things AI generation doesn't do really well. Getting one gorgeous shot of a lady on a beach is easy. Getting 50 more of the same person on the same beach with the same clothes, not so much.
FaceSwap and Inpainting goes a long way though. Start with a picture of a naked lady, then use faceswap to get the same lady again in the next shot. Use inpainting on the areas where the clothes should be with extremely detailed descriptions of the clothing, generate 50 pictures, and keep the 3 that actually work out. It almost works. Here are a few "Vivian V3s" that worked out well, with outfits and poses matching her original beach set:
Are the shirts and the bow exactly the same? no, but close enough. The location is somewhat different too, but whatever, beach! (also I didn't try very hard) Is it exactly the same woman? No, but only if you look real close. Also I had to in-paint the breasts like 50 times to get nipples that looked similar (but it was worth it). Using PyraCanny with an original Vivian shot kept the body consistent. I turned her into a downloadable CSP opponent for your entertainment. Enjoy the uncanny valley.
Getting more shots of that same girl proved tricky though. I tried a bunch but eventually gave up. They all looked like sisters, but not the same person.
So here we are:
SAME BODY: solved(ish), but only if you have 3d rendered shots to start from, and a lot of patience with nipples. And no tattoos.
SAME CLOTHES: solved(ish)
SAME LOCATION: solved(ish), at least for outdoor locations
SAME FACE: not really.
COMPOSITE SHOTS: forget about it.
Pro tips:
- start with the naked shot and add clothes, or your bodies will be all over the place
- Do all the cleanups you need on the first shot before adding any clothes, because other things change too and make it impossible to re-apply the fix to the other shots in photoshop.
- Hands suck. Be ready to redo them over and over. Or hide them.
- Don't even think of holding poker cards. Hands are hard.
- Composite shots? Yeah, no idea how to ever do those.
- AI gets glasses right, but the eyes behind them, almost never. Sorry Denise.
FaceSwap from CSP images did also get me a few extremely striking photoreal portraits of some pretty recognizable ladies. The catch here is I couldn't get very many, and couldn't FaceSwap them in consistently into other shots, but still, wow:
So, after doing a whole bunch of reading and researching, it turns out you can train a thing called a character LoRA to get you a consistent character across AI-generated shots, which would solve both the SAME FACE and SAME BODY problems. (and perfectly matching nipples on the first try, yay!). The only catch? you need to start with 30+ varied pictures of your character, with very detailed caption text files, and then run a training job that takes... 126 hours on my machine (!!!). I have only 8gigs VRAM on my graphics card, but I'm told it would be a whole lot faster if I had 16.
Anyway, the good news is I have a whole lot of pictures of Denise, so I tried that but only ran it for a couple days with too few pictures, captioned wrong, and too many many similar backgrounds, and a mix of 3d render Denises and AI generated photoreal Denise headshots, then stopped and resumed it wrong to boot. You don't get a lot of chances for trial and error with this. I think I may have overtrained it too in some ways.
... It worked though! Kinda. Here are a few cherrypicked and heavily fixed up shots that it made for me. (clothes drawn-in afterward. Much, much re-in-drawing on the faces until they weren't somewhat mangled, but bodies pretty much right on the first try. Except the hands. Nightmare mutant hands, every time.
The good news: It's definitely Denise! And the body is spot on too, and taking poses that were not in the training data! It's Denise's nipples too! (you'll have to take my word for it, sorry) The bad news: aggressively having to fix the face did make her look a tad less like herself, the huge majority of what was coming out was difformed garbage beyond saving, and I had too many pictures in the same 3 rooms, so it also learnt to draw (semi mangled) red leather couches everywhere no matter how many times I asked for a beach shot.
It also never quite learnt to comingle the AI-photoreal and DAZ-render Denises, so I could get full body Daz-Denises, or closeup portraits of AI photoreal Denise. With the occasional uncanny photoreal not-quite-either-Denise in the mix. (her body before he underwear were added was phenomenal and yet still definitely Denise's:
Another almost-win here. This one started out with the dress, but I had to fight with the face quite a bit. Still, it's an almost great photoreal shot of Denise. If I can round up a couple dozen more of these, I might have enough to spend 6 days training a new LoRA the right way!
So, it's probably time to take a step back from this crazy adventure, but I learnt a lot. I'll leave this long post with a couple questions for you all:
- How do you feel about the photoreal AI models? Do you want to see them in the game, or do you prefer the DAZ renders?
- Does anyone have a 16g vRAM graphics card I can borrow?
Comments
Log in with itch.io to leave a comment.
Hi. First posting, but wanted to say this game is great. Lots of ways to customize it (voice, players, outfits etc.). I like that.
The issues with using AI concerning continuity (setting changes, clothes change) and the lady's body (weird anatomy, looking slightly different each image) are definitely there.
I generate the images I speak of online at dezgo. The results can still be pretty impressive, particularly that you can customize things to your preference, i.e. smiling, red cheeks, which makes AI images great.
It takes a lot of patience, but you can use a previous image with text prompts as input to create new images based on that, which results in more continuity.
AI images are definitely an option. With that recent AI short clip generated I think there will be a lot of changes and improvements in many, many ways in this field (not only positive, thinking of jobs and fakes).
BTW, and off-topic: Is there any way to change the amount of money the players start with and they get for pawning things? I would prefer the players had more money, so the game lasts longer.
You can give players a per-outfit starting cash bonus by adding this line to the outfit.properties files:
outfit.bonusStartingMoney=50There is no way to configure this game-wide or set the per-pawn cash amount though. These might be nice configurable settings to add, if I get around to it.
As for the AI stuff, I've had a lot of fun with it, and it can do some wonderful things DAZ can't, but the one thing I've come to grips with is it's definitely not a time-saver. Each amazing image is a long battle of fine tuning prompts and repeated in-painting until it's finally coming out right. Consistency is a big problem, too. There just aren't enough good LORAs out there to provide a good variety of useable characters, and especially clothing sets right now. A year from now we'll probably have non-mangled hands and a deep catalog of swimsuits and pants Loras to choose from. Until then, it's quite good for generating images of Denise taking a shower...
Well, as a contributor, I agree with the comments about the AI models, they do look fantastic but the poses end up be very samey from what I've seen. I like the way I can set expressions, poses, make those subtle clothing adjustments, etc, that all add into the atmosphere. I apologize for my lack of contribution recently but I've been working on my latest novel. The first draft will be done in a week or two and then I can get back to finishing up Aubrey in the coffee morning, and then onto the backyard.
I'm not down with AI art personally. But hey it's content.
imho I think AI would be more interesting in assisting with making new voice packs that sound like a real people.
Technically, the existing voice packs were made with AI. But you're right, they kinda suck and the technology's come a long way in the past two years. These could probably be redone much better (and easier) now. I just don't have the bandwidth to do everything, and hundreds of audio lines can be tedious work especially compared to rendering sexy naked ladies, so it hasn't quite risen to the top of my list yet. If only there were a bunch of enthusiastic fans around that had time to help... ;)
Putting my own kneejerk distaste for AI art in erotic games aside, I think part of the charm of California Strip Poker is the strong sense of space and layout, the continuity between shots. You're not just unlocking a series of pictures of an increasingly undressed woman; you're seeing Denise on the couch, her clothes discarded next to her one-by-one, her pose shifting as more of her's exposed. She doesn't look half bad in the AI art but the backgrounds look like they exist in a big void and that cuts out a lot of the appeal. And I'm no expert here but I don't think you can train the AI well enough to keep a sense of space in its head; the background is always going to be shifting a little.
If you want to use the AI art anywhere, I'd suggest making some kind of bonus gallery with one-off pictures rather than making a whole strippable character out of them. If you win with Maid Denise four times, you'd get to unlock some AI images of her in the same outfit, perhaps.
I'd have to second this. While the AI models look fantastic, you have an attention to detail that's lost with the AI models. They're perfect for a bonus gallery or as a new type of Clothing category (Something like "Spring Break" or "Mardi Gras" where each girl has a narrative that's not directly tied to stripping each piece of clothing), but I think you lose a lot of what makes the "Normal" CSP opponents appealing.
You're absolutely right about all that. If the AI renders can't meet the quality bar that was set by the 3d renders, and do it with less work on my part to boot, then they're not an improvement. The bar is different on games like, say, Striple, where poker cards in the hands aren't desirable, and varying art styles and locations for each costume set are actually a plus.
This might also be a pretty desirable approach for people who want to create their own opponents but don't have the money, hardware, and technical skills that go into doing 3d renders.
The good news is I'm getting a lot better at this AI stuff, and I finally have a properly trained LoRA for Denise, so the continuity part of things is getting a lot better as you can see below:
(I can't say this was less work than the equivalent 3d renders though, lol, but many lessons were learnt in the process)
Stay tuned for another update with even more progress!
Those don't look too bad, at least at this size, though the feet look a bit... off. I can't tell if it's just the angle the feet are pointed at, or if there's actually something wrong with the feet, but something looks wrong and I can't tell what at this size. Also, can't tell how the hands are, but everyone knows about AI and hands at this point.
It definitely ain't perfect, but it's all fixable with a bit more work. That was just me cutting corners because I was trying things out. Rear-Denise is flat out missing one of her feet, and at least half of the hands are problematic. In-painting the bad parts over and over until they turn out OK usually works, as does the classic approach of hiding some of the hands behind things.
I love the AI generated denise. she looks perfect
I'm guessing you're going to enjoy this one a lot too:
There's a whole set of this one up on Striple today! https://eldricus.itch.io/striple