- Feb 21, 2018
- 2,013
- 20,855
- Feb 21, 2018
- 2,013
- 20,855
- Feb 21, 2018
- 2,013
- 20,855
- Mar 1, 2020
- 1,824
- 5,871
Brenda making workout clothes look hot.
- Feb 21, 2018
- 2,013
- 20,855
- Mar 1, 2020
- 1,824
- 5,871
- Sep 19, 2018
- 548
- 2,095
- Sep 19, 2018
- 548
- 2,095
You just enter the room and Alice waits you like this.....
- Jun 3, 2018
- 73
- 70
Been spending some time on a side project would this be a good place to drop a HI mod for Beat Banger? Or is crossing between games frowned upon when it's a fan mod and not the dev?
Also curious if I missed any opinions shared about the AI posts, saw the question posed, but wasn't sure if it was answered.
Also curious if I missed any opinions shared about the AI posts, saw the question posed, but wasn't sure if it was answered.
- Sep 15, 2021
- 853
- 3,179
I've not had any reply from darkhound1, but I would certainly not recommend applying AI (i.e. img2img) to someone else's fanart without prior approval.
As to pure AI-generated images (i.e. txt2img), I guess it depends how close it is to the DAZ stuff.
Personally, provided it's reasonably faithful to the game, I'd say go ahead.
But "Here's a photo-realistic 'Alice' with black hair, mecha armour, a face more like Emily Ratakowski and a flat chest" is likely to be frowned upon!
- Sep 19, 2018
- 548
- 2,095
There was someone here who created a pretty good AI Alice. Look back in this thread. BTW I say it too, DAZ works and AI works are two different things.
- Jun 3, 2018
- 73
- 70
Yeah, I mostly lurk around here and post to help people troubleshoot DAZ assets. Saw the LoRAs mentioned earlier and figured I may be able to do a few tweaks to this side project, and was curious if there were thoughts posted.
Jimwalrus wasn't doing much with the existing fan art or any significant changes other than style. Now that I think about it though, giving Yvette some cat ears would fit BB's genre. I was just using img2img to adjust the style a bit to see if I could make it fit the UI of the mod a bit better, and test out workflow for animations. Which was the underlying goal of working on this mod, to see if I could improve my workflow on animations. And of course that means I'm on my second re-work...
Maybe I'll post a clip once I finish the current update pass.
- Aug 21, 2018
- 951
- 1,776
We've had Eva as a vampire and the girls as zombies without much fan fair. However, having the girls wear sexy costumes has been suggested on the main page many times. I think seeing Yvette in a cat costume, Natasha as Uhuru, or Yumiko as a geisha would be very well received.
I say try it and see.
I say try it and see.
- Aug 21, 2018
- 951
- 1,776
- Jun 3, 2018
- 73
- 70
First few attempts. Still having some trouble with animation workflows.
You don't have permission to view the spoiler content.
Log in or register now.
- Jun 3, 2018
- 73
- 70
Running stable diffusion locally with Automatic1111's webui. Here's the metadata from the first image.Hooooly crap! Which AI image gen did you use?
WOW
<lora:Y3v3tt3:1>, zPDXL, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up
Negative prompt: zPDXL-neg, score_4, score_5, score_6, futanari, censored, multiple limbs, multiple fingers, worst quality, merged, messy, blurry, multiple frames, bad faces, bad hands, bad proportions, horror, weird
Steps: 30, Sampler: DPM++ 2M, Schedule type: Karras, CFG scale: 7, Seed: 4094614932, Size: 1280x720, Model hash: 67ab2fd8ec, Model: ponyDiffusionV6XL_v6StartWithThisOne, Clip skip: 2, Lora hashes: "Y3v3tt3: fe3eab740f66", TI hashes: "zPDXL: c1fcf5eef414, zPDXL: c1fcf5eef414, zPDXL-neg: 31a5da9018b9, zPDXL-neg: 31a5da9018b9", Version: 1.9.3
I had given up on trying to fix img2img importing description text into the prompts, for some reason the option to use PNG info from the folder doesn't appear to actually do anything. So instead of trying to work in bulk for animations, I decided to try out creating some training.
The first image's prompt was literally my first one with simply my attempt at a Yvette LoRA as the prompt, also included my usual pre-saved positive and negative prompts as well as the zPDXL positive and negative embedding. I need to re-run the training as I messed up trying to also include the swim2 outfit, which I think resulted in that batch of images being included in the normal part of the model. Checking that folder, I definitely messed it up, so this model is overfit, and will put Yvette's hair on any male or female head included in the image.
The rest of the images also use a 3danimation style lora, I've found that seems to get rid of some of the uncanny valley issues with attempts at a photoreal style, and prefer the brightness of that style personally.
- Jun 3, 2018
- 73
- 70
Lol, incidental "cat girl" image... seems like a good profile photo selection.
