Previous Thread: >>3045583Request Thread: >>>/y/airt>LOCAL FORGEUI **NEW**https://github.com/lllyasviel/stable-diffusion-webui-forge>LOCAL WEBUI SETUP *OLD*https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki>LOCAL COMFYUI *OBSOLETE*https://github.com/comfyanonymous/ComfyUIhttps://comfyanonymous.github.io/ComfyUI_examples/https://openart.ai/workflows/templates>LOCAL MODELShttps://civitai.com/models/257749/PonyDiffusion-v6https://civitai.com/models/288584/autismmixhttps://civitai.com/models/184813/animeconfettitunexlhttps://rentry.org/5exa3 (hll+fluff) *OLD*https://gitgud.io/gayshit/makesomefuckingporn (use wayback machine) *OLD*>PAIDNovelAI: https://rentry.org/hdg-nai-v3>RESOURCESFeatures: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/FeaturesTraining: https://rentry.org/59xed3 | https://github.com/derrian-distro/LoRA_Easy_Training_Scripts | https://civitai.com/articles/3105/essential-to-advanced-guide-to-training-a-loraTags: https://danbooru.donmai.us/wiki_pages/tag_groups | https://danbooru.donmai.us/related_tagControlNet: https://rentry.org/dummycontrolnetIOPaint (LamaCleaner): https://huggingface.co/spaces/Sanster/iopaint-lama | https://www.iopaint.com/installAnimation: https://rentry.org/AnimAnonWildcards: https://rentry.org/NAIwildcardsBooru: https://aibooru.online4chanX Catbox/NAI prompt userscript: https://rentry.org/hdgcbNAI API/prompting in auto1111: https://github.com/Metachs/sdwebui-nai-apiPony-related: https://rentry.org/ponyxl_loras_n_stuff | https://rentry.org/ponyxl_lora_previews | https://lite.framacalc.org/4ttgzvd0rx-a6jfEtc tutorials: https://stable-diffusion-art.com/tutorials/
does anyone know how to make novelai not shoot out the ugliest images imaginable
>>3047697Have you tried the link in the OP?
>>3047701didn’t notice it, i’ll check it out ty
>>3047690Lmao, that hand and the testicle.>>3047695Was this supposed to be a thong? A fundoshi?
>>3047718there are 2 guys and 4 penises in that
>>3047718your guy is taking a big shit wtf
>>3047718Is he pooping out dicks? Lmao that looks so bad.
SDXL
AnimeboysXL V3
Stablediffusion
I really like this one, despite the black strip. Such an interesting pose.
You should announce yourself when entering a god's temple
requesting more beefcake bara daddy bums
>>3047732>>3047733>>3047736Is it difficult to get only one set of ears?
>>3047758It's hit or miss, depending on the theme and the hairstyle it goes with. ADetailer can also come in and ruin it if you're not warding it there too. score_9, score_8_up, score_7_up, score_6_up, male only, perfect face, 18 years old, young, full body, (catboy) only cat ears, slim, thin, twink, pubic hair, foreskin, ((ancient egyptian theme)), egyptian god, <lora:1dkXLP:0.6>, <lora:add-detail-xl:0.6>, <lora:xl_more_art-full_v1:0.6>, <lora:Expressive_H:0.5>,Negative prompt: makeup, old, facial hair, watermarks, signature, human ears, earsSteps: 60, Sampler: DPM++ 3M SDE Karras, CFG scale: 7.5, Seed: 3137101854, Size: 1024x1024, Model hash: ac006fdd7e, Model: autismmixSDXL_autismmixConfetti, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer negative prompt: "penis, foreskin, precum, cock, exposed cock, leaking precum, excessive precum, precum, cumming, orgasm", ADetailer confidence: 0.55, ADetailer dilate erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.27, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer version: 24.4.2, Lora hashes: "1dkXLP: b02ab01e2c98, add-detail-xl: 9c783c8ce46c, xl_more_art-full_v1: fe3b4816be83, Expressive_H: 5671f20a9a6b", Version: f0.0.17v1.8.0rc-latest-276-g29be1da7
>>3047758
>>3047758This is the same prompt as >>3047760 and >>3047761 without the Egyptian theme, as an example
You ever get it so good your grip crumples stone?
Can someone re-do this one but from behind, so it shows the top's & bottom's asses?
>>3047859I made this one, but I cleared out a lot of my old ones so I don't remember what I used to make it. I recall the prompt included something about a collared hero fucked by a demon king along with "forced orgasm." I possibly used Confetti with Eclipse style (and several other) loras. No clue what weights.
Receive the blessing of your god
Trying out AnimeBoyxlV3
>>3047760>>3047761>>3047878>>3047880>>3047882saved to the spankbank, good stuff
>>3047758It's not that hard if you can do some photoshop. Just spot heal over the ear a few times to remove it and then run it trough img2img with low denoising strength to fix the mess photoshop might have made.
>>3047757What prompts do you use for this art style ?
>>3048018Autismmix:confetti checkpoint with Vixon's Pony Styles "Detailed Painting 1.0" at 0.7 strength.https://civitai.com/models/352581?modelVersionId=399443Picture not related
>>3048029
>>3048032It's really hot but it's lacking a lot of body hair for me, love hairy big dudes
>>3047951SEX>>3048007Cute kemonomimi boy
>>3048028Thanks. I wish I could do that on NovelAI.
AnimeboysXL v3.0
Stable Diffusion
Positive: score_9, score_8_up, score_7_up, score_6_up, score_5_up, rating explicit, full body shot, 1 boy, solo, focus male, young, final fantasy xiii,hope, bedroom location , blond hair,brown eyes, erection under clothes, tank top, amazing quality, best aesthetic, <lora:xl_more_art-full_v1:0.5> <lora:NAI Smooth Boys Style SDXL_LoRA_20r_20e_8i_nr32_a16_Pony Diffusion V6 XL:0.7>Negative prompt: nude,precum, old, facial hair, makeup, women, girl, abs, muscular, fat, heavy, text,lowres,bad,text,error,censored,bar censor,worst quality,bad quality,watermark,bad aesthetic,simple backgroundSteps: 50, Sampler: DPM++ 3M SDE Karras, CFG scale: 7.5, Seed: 2044389829, Size: 864x1152, Model hash: 80cc8f54fc, Model: animeboysxlV3_v30, Denoising strength: 0.4, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer version: 24.4.2, Hires upscale: 1.5, Hires steps: 35, Hires upscaler: 4xUltrasharp_4xUltrasharpV10, Lora hashes: "xl_more_art-full_v1: fe3b4816be83, NAI Smooth Boys Style SDXL_LoRA_20r_20e_8i_nr32_a16_Pony Diffusion V6 XL: d3ffbd6136e2", Version: f0.0.17v1.8.0rc-latest-276-g29be1da7
>>3048286don't look too close at the feet
>>3048296Yeah, I saw it when it was too late lol. I tried to fix it
https://m.youtube.com/watch?v=GU1o2blfeO0&pp=ygUJTmZsIHRoZW1l
Recently discovered that Tensor has a model for my boy, Ranulf. I've been able to make some hot stuff with him, but I can't manage decent looking bondage or shibari. Any good prompts to keep his hands down behind his back, and maybe get some rope shibari on him?
>>3048373
>>3048374Dark Sun sexo
Ass blasted to kingdom come
>>3048377
>>3048324I have little experience with that but you can just search the tag on Danbooru and it will show you similar tags, also if you clic on the "?" Symbol next to the tag it will send you to a page with info about it and other tags
>>3048360brilliant. more!
>>3047798>>3047801Love these, where can I see more of your work?
>>3048406
>>3047935>>3047936>>3047938I love these three. Rape, gangbang and school uniform are my favourite kinks! I would love to see more if you get the time.
>>3048377>>3048378>>3048381>>3048382based boyraper
>>3048420>>3048421Model?
I am doing Saint Seiya characters
>>3048442Model: AnimeBoysXL V3Lora: Zoroj Style [Pony/SDXL]
How does one learn to harness this power? Ideally locally on one’s PC. Steps unclear, stuck dick in blender
>>3048476Very quick tutorialStep 1: Have a GPU from the last decade, that has at least a few gigabytes of vram(the more the merrier, ideally 12gb+)Step 2: Go to automatic1111 github and follow instructions thereStep 3: Once that is set up go to civitai dot com, register and download "Pony diffusion V6", hands down the best model for porn(Don't worry it is not just Brony shit despite the name). Put it in "/StableDiffusion/stable-diffusion-webui/models/Stable-diffusion/"Step 4: Launch stable diffusion, select pony diffusion from top if it is not automatically loaded. Follow the instructions on pony diffusion's page. Set sampling method to Euler a, set sampling steps to 25 steps, set resolution of 1024 width and height, set CFG to 7. Batch size should be how much you can fit into your vram, leave it at 1 for now.Step 5: Use this prompting format, don't question too much, just do:"score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, just describe what you want, tag1, tag2"Tag parts are what you see in hentai/rule34 sites. Example prompt:"score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, A soldier giving BJ, 2 boys, looking at viewer, yaoi, big penis, oral sex, fellatio, andava style"Add these to negative prompt sections, prevent furry and brony stuff from seeping in:"text, watermark, ugly, source_pony, source_furry"Refer to pic related to see what kind of stuff comes outThere is a decent bit more to it, but this should get you started. Have fun anon.
>>3048476https://www.reddit.com/r/StableDiffusion/wiki/local/It is a lot to learn, but good luck and don't be afraid to ask questions
I discovered that animeboysxl kinda recognizes this dude and I'm loving it
>>3048496
Many thanks for your benevolent wisdoms!
>>3048386Not too shabby. Model gets a bit weird with the ropes, and the prompt certainly doesn't know how 'gags' of any kind work, but they certainly lead to fapable content, at least.male focus, 1boys, solo, ranulf, facial mark, tape_gag, pectorals, shibari_over_clothes, arms_behind_back, red_rope, kneeling, rope_bondage, shibari, angry, heterochromia, purple eyes, green eyes,
>>3048516Nice, out of curiosity, which model is it? Is a 1.5 model? I can't find the Lora on tensorart
>>3048486not the same anon, but I also recently started.wondering if there are any decent impainting guides around. for example, I tried sketch impainting over some hands, or getting rid of extra arms etc. but sd keeps redrawing the stuff I don't want (i.e. too many fingers, and stuff like that)So I guess I'm doing something wrong.My idea was to sketch the rough shapes and then impaint with lowish denoising with mask over the area I want to redraw and use original source or w/e it's calledbut like I said it either redraw wrong things or was still noisy, so are there any hints what to do here?
>>3048539>Learn to inpainthttps://rentry.org/drfar | https://stable-diffusion-art.com/inpainting_basics/From /airt/, not very keen on AI myself but i hope this'll help.
>>3048541ah,ty I already stumbled upoin the second one, but the rentry one was new, I'll give it a try
>>3048538Yeah, 1.5.https://tensor.art/models/710302310158324494
Wish it could get the eyes right...
Last one. Looks kinda crappy though.
>>3048516Getting a lot better. Gags still get wonky with this model. Possibly due to his facial markings. Otherwise, I'm still seeing if I can manage 'fellatio', 'unzipping his shirt', 'lifting his shirt up', and the dreaded 'managing a second figure'.That said, all this is purely to give myself porn. I'm no artist with this.
>>3048644It's really good, nice improvement!
>>3048748rintaro?
>>3048861Ryoji Kaji (Evangelion)
Where can I find gut punch Loras? I love gut punching.
Heero Yuy captured
AnimeboysXL V3.0
>>3049077is this a lora or does the animeboys model recognizes tentakun?
>>3049111lora
Does anyone know of a lora that has Haikyu characters, specifically Oikawa, Kuroo and/or Bokuto? I know there’s ones for Hinata and Kageyama but I’m not too interested in them.
SD
>>3049667The ones you mentioned are the only ones I've seen. But you can try your luck with animeboysxl, if they have 200+ entries it should be an easy gen prob
>>3049510isn't he supposed to be a white dude
>>3049404I like the retro style characters.
AutismmixSDXL Confetti
>>3049744He looked better before he was unmasked desu
>>3049975I so want to see the "during" and "after" version of this! Could be epic!
>>3050390so clean, do you mind to share your prompt? and which high-res sampler do you use? tysm
>>3050404test
>>3050403score_9, score_8_up, source_anime, rating_explicit, big penis, orange hair, gloryhole, blowjob, 2boys, duo, public restroom, public bathroom, (deepthroat, deep throat:1.3), wide-eyed, crying, (ridiculous fit), huge penis, drool, drooling, throat bulge, deep penetration, thick penis, huge balls, sagging balls, fondling, fondling balls, ball grab, ball fondling, cum in mouth, cum in nose, cum from mouth, excessive cum, overflowHires upscaler: R-ESRGAN 4x+ Anime6Bthere's 2 loras too which you might be seeing as what is clean
my horny ass could never work in the labrynth
>>3050458Mind telling us what are those 2 loras?
>>3050605it's mostly just a watered down rokudenashi (picture related)it's not the one I'm using but there's one on civitai: https://civitai.com/models/296797/rokudenashi-style-for-pony-v6-xl
I miss Rokudenashi stuff so much
>>3050826he's still alive and making art but only sfw irl art gallery stuff sadly, under the name heyshiro
Still new at this so i decided to mess around with img2img for the lulz, not sure how to feel...
I have decided to bless you all with this lora https://files.catbox.moe/1q6u12.safetensorsI'm pretty bad at prompting so it'd be interesting to see anything good someone could make with it
>>3051101Nice thank you, im curious about what is it. Tonight in gonna test it
>>3051101nice, i'm trying it now and it works way better in autismix than animeboysxl. I had one problem tho, I can't use it in local for some reason, the lora is not recognized. I had to upload it to tensorart.
>>3051119Another test, kinda loving the style.
>>3048032>>3048029Post more, these are fucking hot
>>3051119you can use it on local but you have to manually write it in the prompt for some reason
>>3051132Whoa that's true, and the quality it's amazing there. Sorry for the mess, I tried once before going to bed and local is really slow on my computer, it ended making that random cock inside his tanktop because yes, probably a fault in my prompt lol. If you use local, do you have any tips to make the proccess quicker? I've already used the nvidia trick and it worked, but still kinda slow.
>>305113420 steps + use forge ui + lower resolutions are faster but worse for qualityalso it doesn't really seem like the style is coming out much, which is weird cause it seemed to with only a few tags for metry adding male focus and male, if they aren't already in your prompt; I know "score_9, score_8_up, source_anime, rating_explicit, BREAK, male, male focus, muscular male, penis," makes it clear as day for me at least with autismconfetti
>>3047690Say, what are y'all general go-to?Models and LORAs (basically, things that you prefer no matter the situation/gen)
>>3051134>>3051120>>3051119Do you do requests?
>>3051135I have to try next time I'm free to use the computer, thanks. Maybe that was because because I put the weights in like 0.75 instead of 1? I think I used confetti too
>>3051135Not on my computer, but this was my first gen testing the Lora in tensorart with your prompt and weight 1, the only other Lora I used was the pony quality one that I've always use. I don't know if it's the result expected because I don't know the artist, but I loved the result.>>3051307 oh, ask if you want and I'll try for fun, but they weren't anything special, pretty messy prompts, it was more the result of the Lora anon uploaded.
i've been having fun generating these images.
>>3051351Thank you. Requesting a fairy barbarian. He has short spiky dark blond hair, pale skin, grey eyes, tribal war paint on his body, and wears nothing. He has a chinstrap beard. If you can, he wields a sword with both hands since it's a regular size sword to others.
\
>>3050826Close enough. Welcome back Rokudenashi.
>>3047690
>>3051543
>>3051554
>>3051543>>3051554May i ask for prompt on these?Love the lighting here, most of my gens' lighting are awkward to say the least...
>>3051101What was it supposed to do? It didn't change anything.Pic not related.
So is there just no controlling for style with SDXL/Pony? I've seen plenty of good stuff by a lot of different people, but everyone I see never has a consistent style, it completely changes based on the subject.The style Loras on CivitAI all seem to either be very general or overtuned to the point where they essentially just make one character.
>>3051654civitai loras are trash but the style is trained on tags, so obviously the style is going to change based on the tags you use or don't use
>>3051656Alright, where does one get good loras, then?
>>3051654The original Pony V6 is famous for having issues with style types. The way they tried (and failed) to design Pony V6 is an interesting read https://civitai.com/articles/5069/towards-pony-diffusion-v7I would like to point out here that real people volunteering their time is how this stuff is getting created. Some mistakes are bound to happen, but lets all be just a little grateful it is happening at all.Most people either use a style Lora, or switched to a different finetune, such as AutismMix. You can call specific Artists in the prompt, it has been made to recognize many. Many people don't bother, but that is one of the reasons that Pony can seem off. It is expecting you to ask for a style, and when you don't it swings around without guidance. https://docs.google.com/spreadsheets/d/1m2W-pZEvHuEpfHcNHrxCSr-Aw1mgtUUYho6sz9LChEA/edit?gid=0#gid=0Also, remember that Pony, and SDXL in general, is expecting you to describe the person you want to see with details. If you are prompting just for "1 male, male focus", with no actual description of that man, then yes they will all look the same. Usually black hair, dark eyes, very basic.
>>3051660Not OP but thank you for the tip!Never knew about this feature for PONY, I really need to update myself more on AI in general...
>>3051660more lingerie please
>>3051654styles are honestly not too versatile on pony, especially base pony. with 1.5 you could just mix any artists you want.imo the only good pony model that is also the most versatile is pony real magic. it's tensor-exclusive though. it can absorb style loras pretty well (like the paper mario or family guy LORAs, for example)
>>3051712I love it, it reminds me of an sprite of the remakes of the games Phoenix wright
>>3051782
>>3051783
>>3051784>>3051783May i ask for the model & prompt?
didnt see much content of 17
>>3051817I wish I could figure out how to make the robot or the man's hands rubbing oil all over 17's body, holding his hips, grabbing his nips. idk why no matter what i prompts it just does nothing.
>>3051818I added the Kirsi Engine Lora. Interesting effect.
>>3051782>>3051783Oh god yes, more Bolin please. Or anything with Avatar men.
>>3051798Yes! The checkpoint was AAAutismhttps://civitai.com/models/353543/aaaautism-or-pony-finetune-mix?modelVersionId=514466bolin's model: https://civitai.com/models/341134/bolin?modelVersionId=381923trafalgar law's model: https://civitai.com/models/361307?modelVersionId=4037621man,score_9,score_8_up,score_7_up,score_6_up, naked, nude, male focus, moaning, blushing, sweating, solo, penis, pink glans, testicles, low-angle pov, masturbating for viewer, ejaculating on viewer, low-angle view, cumshot on viewer, cumming on viewer,negative: EasyNegativeV2https://civitai.com/models/100191/ti-easynegativev2-textual-inversion-embedding>>3051822avatar guys are so fucking hot, so glad you liked it!
>>3051823here's another version of that bolin pic too.
>>3051654>but everyone I see never has a consistent styleBecause people like change and vary, only a few autists pick a style lora and use that forever, most people want to try different stuff, it has to do with what people do not what is possible, you can pick a style lora and use that forever and you will always have that style on pony
>>3051864Nah, I don't buy it. Take a look at this guy as just one example:https://civitai.com/user/MuscleGong/imagesEvery batch of images is one character in one style, never the same style, never any style changes in the same character batch. I've seen multiple profiles on Civit that are the same situation.
>>3051864>>3051869Not to be insulting, but people who use Stable Diffusion like to think of themselves as artists. And just like Bludwing or Sakimichan, they like to figure out a specific combination of Loras that gives a distinctive style. Thus, giving them recognition with viewers.That isn't really a good or bad point of A.I generatio, just the way people live.
>>3051823Thanks!May i ask what's the "score" is about? I'm new so i'm kinda confused about all of the "score" tags...Does it helps make the gen better?
>>3051895it's the prompting style for that specific checkpoint (or in this case, for all pony-style checkpoints). it's how the model was trained. every checkpoint usually has some information about prompting.and yes, it's to get better pictures.
>>3051917thank you!I'll def remember this when using pony related stuffs, will be very useful
>>3051895>>3051922also, you could NOT add the scores and replace them with a good embed.Like thishttps://civitai.com/models/332646?modelVersionId=380277 (this one provides both positives and negatives, use both.)I replace the scores for "zPDXLxxx" in the positive prompt and, in the negative prompt, "zPDXLxxx-neg". Getting really cool results with that.
>>3051956It is a very good thing to be skeptical of what people are telling you must be prompted. I have personally found that even the recommended prompts from the checkpoint trainers on Civitai must be toyed with for best results.I did NOT like what PDXL Embeddings did to the images I created. That doesn't mean you are wrong, or I am right. This is (essentially) art, after all, it is good for people to view it differently.I would suggest that you (and everyone) not feel any Lora or Embedding is a "must have". Thinking like that will stunt your creations at some point.You should always set time aside to experiment. Challenge your own preconceptions.
>>3051956>>3051960Thanks for the insight!Deffinitely going to try and see what i like best.
>>3051960agreed, in the end it all boils down to trial and error, the better your hardware, the more "toying around" you can afford, I guessbut yeah, I love trying out lora combinations, to see what that does to the image, etc.also checking out prompts of pictures you like can get you ideas for your own prompts
May i request something? Link standing under a waterfall and washing himself, with a low and tight cut flaccid cock. Hes kinda embarrassed but doesnt hide his dick
>>3052019>>>/y/airt is more suitable for request in general*It's not exactly forbidden to request here, you'll just get a better chance at /airt/, since the thread itself is about requests(pic not related)
>>3052019while it's not the highest detailed one, I thought it had an interesting style and it's kinda cute
>>3048131Not an algorithm but if you have a story idea maybe I could write a short piece for you.What is your kink?