Mirajboi201
Expert Member
- Joined
- Oct 17, 2017
- Posts
- 44
- Media
- 0
- Likes
- 109
- Points
- 68
- Location
- Miami, FL, USA
- Sexuality
- 80% Gay, 20% Straight
- Gender
- Male
Photos & Videos AI Tools: How to and Technical Discussion on Creating NSFW Gay Sex Stuff with Stable Diffusion (and others)
- Thread starter StudmeatFL
- Start date
-
- Tags
- bara civitai men stable diffusion
What types of things are you looking for with ControlNet?
So far I've experimented (and get much better results with) Depth. I'm also playing with OpenPose but find it really klunky and likewise with Canny.
I find that I either start with an Img2Img or I create an image I like and then pull it into Procreate or Figma to 'add' stuff. Then I pull that into ControlNet Depth and write a good prompt around it.
Img2Img seems to work better if I only want a starting point and play with the Denoising strength. But sometimes it just gets impossible to get what I want and I have no idea why. Especially with Inpaint.
Nice job, remember this is a more 'technical' discussion, so if you're posting here, try to also post the prompt. We have that other thread for dumping AI "Spank Bank" material into.
Granted, I'm just as bad and think I forgot to add my prompts for some of the stuff above. May go back and edit.
I've been playing with Inpaint and Img2Img over at Tensor.art and I can attest that it DOES work depending on your Model file.
Diffusion Bee seems to be missing any kind of Denoising Strength, though, so that might also be the issue.
Yeah I’m having lousy luck with Diffusion Bee. Tensor.Art with a SD XL model seems to do a much better job with inpainting. I do get some really weird results. I just keep playing with Denoising and the mask to see what I can come up with.
The other thing I noted with Diffusion Bee is that you couldn’t use Inpainting on any SD XL models, like Juggernaut XL, though you could download it. It says something about Control.net only supporting 1.5 models and you also can’t add any external SD XL models.
With those limitations, it sucks donkey dicks compared to what SD XL can do over at Tensor.art.
It looks like it was added back this morning. Just search for “Virile”.
I had good luck with Hard Muscle -Vanilla. This image was generated off that one.
Attachments
yeah bcs people were asshole to the original author so he nuked it lol
there's a reUP- Virile Stallion - v5.0 Photoreal | Stable Diffusion Checkpoint | Civitai
Just playing around, honestly... most of this is pretty easy if you start with the right model. I do agree with Fjuwhr above that if you want to play without costing anything, try it out at Tensor.art. They've even got an 'easy' mode where you basically just put in the prompt and see what you get. I'll keep trying to do more tutorials as I learn more stuff as well.
I don't feel good enough to do commissions, yet, and for me, the things I create have to match my own personal 'spank bank' or I won't enjoy making them, y'know what I mean? It's like asking someone who gets their rocks off taking videos and pics of hot muscle men jacking off and asking them to photograph ladies instead.
What
What did they do? Just curious as I didn't see anything over on Tensor.Art.
might be a stupid question but how did you get a perfect cumshot like that ? I’ve been trying and they all turn out horrible
I had an img2img with a pic where the guy was cumming so that helped with the reference. I also use the LoRA Bukkake which seems to work better with properly positioned cumshots.
I lot of it is the phrase in the prompt: explosive_cum, cum_shot.
Let me grab what I used in the prompt on that one and share.
The Civitai ones are much better. Much greater range of stuff as well. I get lucky at tensor.art and getting much better results since shifting to SDXL