These Women Do Not Exist ..... The NSFW A.I. Art Open Thread

[smoovejazz]

Two things...

First, since winter is coming, how about some AI featuring some redbone, Indian, and Latina BBWs?

Also, is there a place that has a tutorial on how to make these AI images? I want to get in on this, and I've got a few ideas I want to create and post. Thanks.

 

[D24OHA]

MSI GeForce RTX 3060 Ventus 3X 12G OC

Bruh a 3060 is no slouch. I'm getting some decent output from a 2070 laptop gpu.....

You might want to check your settings and make some tweaks. I doubt that gpu is the problem.

 

[Mixd]

Two things...

First, since winter is coming, how about some AI featuring some redbone, Indian, and Latina BBWs?

Also, is there a place that has a tutorial on how to make these AI images? I want to get in on this, and I've got a few ideas I want to create and post. Thanks.

At what point do you need help?
I posted where to begin, putting Automatic1111 on your PC. There's some tweaks you may need to do to your batch file, read the comments on the tutorial link. It's very helpful.

From there, go to Civitai.com and look at images you like, look at the prompts and can copy into your setup. Grab the Checkpoint they used and any LoRA. Or can look at the most popular checkpoint, right now it's Juggernaut. Look at the images it has made and if you like the scene and look, just alter it to what you want to see, like Indian or Latina or lightskin black girl (redbone, I doubt it will understand)

All the images I posted today, I made from other images I saw of white chicks and altered the prompts.

Also understand, the newer SDXL models require less prompts and less steps to create great images, and hands are handled a lot better.

 

[smoovejazz]

At what point do you need help?
I posted where to begin, putting Automatic1111 on your PC. There's some tweaks you may need to do to your batch file, read the comments on the tutorial link. It's very helpful.

From there, go to Civitai.com and look at images you like, look at the prompts and can copy into your setup. Grab the Checkpoint they used and any LoRA. Or can look at the most popular checkpoint, right now it's Juggernaut. Look at the images it has made and if you like the scene and look, just alter it to what you want to see, like Indian or Latina or lightskin black girl (redbone, I doubt it will understand)

All the images I posted today, I made from other images I saw of white chicks and altered the prompts.

Also understand, the newer SDXL models require less prompts and less steps to create great images, and hands are handled a lot better.

Thank you. I'll play around with it, and come back with further questions if needed.

 

[Duece]

 

[Mixd]

 

[Duece]

 

[Duece]

 

[Duece]

 

[Duece]

 

[Mixd]

some more of outputs I made

 

[Mixd]

​

 

[Mixd]

 

[Mixd]

 

[Mixd]

Eve in the garden?

 

[Mixd]

Chicks in a cave

 

[Mixd]

variation on Eve

 

[TheBigOne]

There's a lot of prototypes up in here. Just dreaming

​

 

[Jake Reed]

Chicks in a cave

​

Did you do this?

prompt: a black woman, large breasts, (black hair:1.2), nude, dark and mysterious cave ,
Porta 160 color, shot on ARRI ALEXA 65, bokeh, sharp focus on subject, shot by Don McCullin
Negative prompt: BraV4Neg, bradhands, BadDream, EasyNegative, Brav5Neg, FastNegativeV2, UnrealisticDream, badhandv4
Steps: 28, Sampler: DPM++ 3M SDE Exponential, CFG scale: 5, Seed: 751731076, Face restoration: CodeFormer, Size: 640x960, Model hash: 07919b495d, Model: picxReal_10, VAE hash: 15e96204c9, VAE: vae-ft-mse-840000-ema-pruned.ckpt.vae.pt, Denoising strength: 0.37, Hires upscale: 2, Hires steps: 17, Hires upscaler: 4x_NMKD-Siax_200k, Version: v1.7.0
Template: a woman, small breasts, (blonde hair:1.2), nude, dark and mysterious cave , pussy cat
Porta 160 color, shot on ARRI ALEXA 65, bokeh, sharp focus on subject, shot by Don McCullin
Steps: 40, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1672963291, Face restoration: CodeFormer, Size: 512x768, Model hash: 0724518c6b, Model: juggernautXL_v7Rundiffusion, Denoising strength: 0.29, Hires upscale: 2, Hires steps: 20, Hires upscaler: 4x_foolhardy_Remacri, Version: 1.7.0

Thank you for this. I'm going to try out the NMKD-Siax upscaler and the pixreal model. I was getting subpar results with the original NMKD superscale as my final upscaler. I'm shocked at how that turned out with just 17 high res upscale steps. Do you normally use so little steps?

 

[Mixd]

Did you do this?



Thank you for this. I'm going to try out the NMKD-Siax upscaler and the pixreal model. I was getting subpar results with the original NMKD superscale as my final upscaler. I'm shocked at how that turned out with just 17 high res upscale steps. Do you normally use so little steps?

Yes, I just copied another image I found on Civitai and changed the prompt for "a black woman". Didn't change anything else. It's probably because it's SDXL is why so little steps.
I see that it jumped to NMKD, not sure why, the original used Remacri, it's what I usually use.

 

[Mixd]

Did you do this?



Thank you for this. I'm going to try out the NMKD-Siax upscaler and the pixreal model. I was getting subpar results with the original NMKD superscale as my final upscaler. I'm shocked at how that turned out with just 17 high res upscale steps. Do you normally use so little steps?

I redid these with 3 different checkpoints, I see the original prompt data "junk" got copied into the Negative prompt, I didn't catch that, my bad. I'll mess with the NMKD, just downloaded it

This is the original

 

[SmaLLz]

Bruh a 3060 is no slouch. I'm getting some decent output from a 2070 laptop gpu.....

You might want to check your settings and make some tweaks. I doubt that gpu is the problem.

I didn't necessarily have an issue with the gpu, I think it was more an issue with my expectations. I tried to train an image and it didn't come out right and some of the other stuff I made was trash. I don't think I have the bandwidth right now add something new like this to my plate, but I am going to revisit. I am trying to get nice with Excel right now and Stable Diffusion requires a little more time than I have to dedicate at the moment.

 

[OmariAmaru]

​

Now this some straight b.s. "it was real to me, damn it!"

 

[Jake Reed]

I didn't necessarily have an issue with the gpu, I think it was more an issue with my expectations. I tried to train an image and it didn't come out right and some of the other stuff I made was trash. I don't think I have the bandwidth right now add something new like this to my plate, but I am going to revisit. I am trying to get nice with Excel right now and Stable Diffusion requires a little more time than I have to dedicate at the moment.

You tried to train a model?

12GB is fine for SD 1.5 based models (using them not training them), but not enough for SD2.1-768.

What site made these? These are some of the more realistic ones that I've seen. Most of the others look like anime or video game characters.

Great coherency in those images. I was using Lifelike v3 because it is more responsive to body type LORAs, but the images are coming out somewhat blurry.

 

[Duece]

 

[Duece]

 

[Duece]

 

[D24OHA]

Wow

 

[Duece]

 

[Duece]

 

[Duece]

 

[D24OHA]

Bruh......that looks realistic....

Prompts? And negs

 

[Duece]

Bruh......that looks realistic....

Prompts? And negs

aerial shot, wide lens, curvy black woman in a glass boat, she is laying down in a pose, wearing a swimsuit, in the ocean,the boat is see through clear, the water is clear, blue tinted water, location Grand Turk, hyperdetailed, hypermaximalist, fujifilm --v 6.0

3d, cartoon, frame, text, low quality, long neck, plastic, 2 heads, 2 faces, ugly, tiling, poorly drawn hands, poorly drawn, crossed eyes, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, blurry, bad anatomy, blurred, watermark, grainy, signature, cut off, draft, duplicate, distortion of proportions, anatomy, coppy, multi, two faces, disfigured, kitsch, oversaturated, grain, low-res, mutation, mutated, extra limb, missing limb, loating limbs, disconnected limbs, malformed hands, blur, out of focus, mutilated, mangled, dots, bad quality, draft, text, logo, watermark, blurry, bad anatomy, extra limbs, poorly drawn face, poorly drawn hands, missing fingers

 

[Duece]

 

[D24OHA]

aerial shot, wide lens, curvy black woman in a glass boat, she is laying down in a pose, wearing a swimsuit, in the ocean,the boat is see through clear, the water is clear, blue tinted water, location Grand Turk, hyperdetailed, hypermaximalist, fujifilm --v 6.0

3d, cartoon, frame, text, low quality, long neck, plastic, 2 heads, 2 faces, ugly, tiling, poorly drawn hands, poorly drawn, crossed eyes, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, blurry, bad anatomy, blurred, watermark, grainy, signature, cut off, draft, duplicate, distortion of proportions, anatomy, coppy, multi, two faces, disfigured, kitsch, oversaturated, grain, low-res, mutation, mutated, extra limb, missing limb, loating limbs, disconnected limbs, malformed hands, blur, out of focus, mutilated, mangled, dots, bad quality, draft, text, logo, watermark, blurry, bad anatomy, extra limbs, poorly drawn face, poorly drawn hands, missing fingers

My man......

I'm learning.....damn that's a good one