Previously, on /asdg/ >>8840706
The cartoony side: >>>/aco/csdg

No DALL-E, Meta AI or hailuo AI generated stuff.
No futa.
>>>/aco/futasdg
Report trolls.

>Guides
More modern options:
https://github.com/comfyanonymous/ComfyUI
https://github.com/lllyasviel/stable-diffusion-webui-forge
Too old, might hold some value: github.com/AUTOMATIC1111/stable-diffusion-webui/wiki | rentry.org/sdg-link | rentry.org/rentrysd
Hires Fix: rentry.org/hiresfixjan23
Inpainting/Outpainting: rentry.org/drfar | rentry.org/inpainting-guide-SD
Animation: rentry.org/AnimAnon
Create LoRAs, TIs: rentry.org/59xed3 | rentry.org/simplified-embed-training
>Auto1111 extensions
ControlNet: github.com/Mikubill/sd-webui-controlnet | rentry.org/dummycontrolnet
Regional Prompter: github.com/hako-mikan/sd-webui-regional-prompter
Tag autocomplete: github.com/DominikDoom/a1111-sd-webui-tagcomplete

4chanX Catbox/NAI prompt userscript: https://rentry.org/hdgcb
4chan-x userscript from /h/ that lets you read generation metadata from images in-browser, and upload to catbox with one click when you want to share metadata.

>Wan Video 2.1
Setup Guide: https://rentry.org/wan21kjguide
Triton Guide: https://github.com/woct0rdho/triton-windows/blob/readme/README.md
Easy Sage Attention Install: https://github.com/woct0rdho/SageAttention/releases

>Related
>>>/b/degen
>>>/h/hdg
>>>/e/edg
>>>/d/ddg
>>>/g/sdg
>>>/trash/sdg
>>>/trash/slop
>>>/vg/aids
>>>/vt/vtai

Close this thread you dumb fucking asshole.
Didn't you read the message? Now I have to do it early because you're a retard.

You couldn't even wait for the thread to fall off? The old thread isn't even close to the image limit.

>>8866073
>>8866073

>>8866084
It´s on low page 10, so per usual a new one is made.
Do whatever you want.
But the baking call was made with enough time.

>>8866088
God you're so fucking dumb and predictable. Get a life other than molesting undead threads.

Get therapy.

>>8866096
You've been molesting threads for months. You literally search for AI porn on reddit to post here. Your worse than any AI bugman.

I love inpainting!

>>8866465
Liquify tool in photoshop.

>>8867215
is this sdxl? if so what model?

>>8866465
What girl/girls did you put in the prompt, or with a lora, to get her face?

>>8866088

hot and rare. indigenas are my favorites.

>>8866195

and preggers too

>>8867463
I used a real pic of Stefanie Joosten and inpaint tits on her

>>8868656
nice try, but it's weird how the hoodie sags down to below her hips

>>8868765
Yeah fair. I thought it was an interesting pose since it looked to me like she's stretching her hoodie out with her hands but I can see why it could look weird.

>>8868656
How well does Chroma work with existing Flux models for doing slutty things?
https://civitai.com/models/331335?modelVersionId=1004306

>>8869610
I've heard they still kinda work but I haven't tried because I didn't expect them to work well. I only used some big titty, BBW and similar loras with Flux though and that kind doesn't seem necessary with Chroma.

>>8869750
is it chroma? prompt? can't get this hot size!

Why do people say Flux can't do porn but there are a bunch of porn loras for it

>>8866083
need more violet

>>8870303
Can't do pony by default, the way SDXL trained models can, or Chroma. Loras are very limited and it's a pain having to find one for every pose or composition you want to try.

>>8870526
*porn by default

>>8870285
https://files.catbox.moe/nmsuxk.png
Yeah it's Chroma v28, and it's a super long schizo/horny prompt that reaches the token limit, I'm afraid. I don't think it works that well since most seeds with that prompt have really big women with balloon tits and it's rare to get that amount of sag and softness. I'm pretty new to this, but I know through testing that Chroma understands terms like "gigantomastia" and I'm pretty sure it responds well to strong adjectives with regards to breast size, so maybe that's a good place to start.

>>8871295
Does it respond to e621 tags at all? I mean the Chroma guy is a furry, photorealism is just a byproduct for him. Should be "hyper breasts".

>>8871325
I've heard people say Chroma should be prompted like <description> <tags> but I haven't tried that much yet. Worth a shot.

I will now view your streams

Requesting-

-1- a redhaired muscle babe jogging down a country road , nude other than her shoes, please.

For bonus points do shots of her both coming and going.

-2- a brunette , muscle MILF working out in a gym either completely nude or top-free, with the presidential seal somewhere in the background.

>>8871871
>>8871887
Love the boobs.

>>8871411
Pony originally described its prompt structure as “anything you like + tags” but if you experiment and find something that works consistently then that works too.

requesting a series of pics featuring a mousey petite college coed with glasses going about university campus nude other than a backpack.

-her in a lecture hall with several other nude students

-her studying in the library

-her in the cafeteria

-her running late to class

-your ideas here

>>8872706
>requesting a series of pics
Pass.

>>8872706
best i can do for you is this

>>8873622
>>8872706
just as an FYI though this shit is easy enough to make (not that you should as it is very addictive) that you shouldn't need to post requests here

>>8866074 (OP)
a bit of a dumb question, but can you do nsfw art on stable diffusion's online model?

>>8873637
how easy is it?

>>8874066
i generated this and then animated it on another site on one of them fandangled smart phones and the artificial robots that live inside

>>8874475
Nice bounce. It´s tiresome to watch bouncing that would make the people that work on the DoA games ask for moderation.

>>8874281
If you have a reasonably good gaming PC with an Nvidia card, its almost point and click these days. Most complicated thing you'll do is install python and maybe have to fiddle with versions of pytorch. There's a guide /gif/ and probably lurking around here somewhere.

>>8866195

Excellent can you share prompt for this one! Love the gloss and the realistic transitions and overall plumpness.

>>8867431. Also great, ive seen many prompts tat utilize a open crotch but none that were consistently good like this. CAre to share the prompt?

Is inpainting different than facedetailer? Is it better?

>>8875426
adetailers are essentially automated inpainting. The face detailer detects faces and runs a separate diffusion step on the designated area during image generation. I use forge and if you look at the adetailer tab, it has inpainting settings similar to what you'd see in the img2img inpaint tab. They are basically the same, just deployed differently.

>>8875426
same thing except it figures out the inpaint area on its own

>>8875451
What model/lora is this? I can't get bellies to go that big with Juggernaut and it's really annoying.

>>8876334
anything with pony in the mix should give you a bigger pregnant belly

>>8876522
I'll keep investigating then, thanks anon.

>>8876470
>>8876466
Very hot.
>>8876046
>>8874475
Mmm. Disney girls usually end up looking too weird when they’re realistic. These are nice.

>>8876934
It's mostly the face though, right? My trick is to run adetailer with a non-pony/illustrious model that won't recognize her.

>>8877423
bro did a (thin waist:2.5)

I'm playing about with TensorArt, but the tiddies are getting progressively worse! Can any ai artisans recognise what I'm doing wrong? I never asked for steel nipples..

>>8878456
Tensor.art hosts a large number of different models, and each may behave differently. You have to provide more info.

Generated images have metadata, as long as you don't edit them in Photoshop or similar. You can put the image up on https://litterbox.catbox.moe/ and link. This lets people reproduce it on their own and figure out what you did wrong.

>>8878456
>>8878476
Cheers for the response! I didn't know where to ask such a question, so I dropped it here. I'm still at a tinkering stage and trying to get my head around how things work, I used Flux.1 and a LORA called Grini that replicates a famous cosplayer, and this was the result. I've heard phrases such as overbaked, and wondered if that was what I was doing-
https://litter.catbox.moe/hovd4bmzqakelzfh.png

>>8878502
First, check for compatibility, as >>8878476
said, there are a lot of possible combinations and not all play nice with each other.
Also, Flux is notoriously censored, so for some explicit content, porn LoRAs will help.

>"ckpt_name": "EMS-597859-EMS.safetensors"
>lora:EMS-559942-EMS.safetensors:0.800000
Are these hashes? I didn't realize Tensor hid the filenames, would be the only online service to do so. Still, it looks like a Flux variant with a nsfw lora.

>>8878541
I was using Flux.1 (https://tensor.art/models/757279507095956705), and the following Lora (https://tensor.art/models/791727975193327830)

>>8878541
No, just a weird name; and then it seem it is just the LoRA name and the weight being used.
It might be an internal hash that gets used as a name.

>>8878546
in that case >>8878535 is right
Flux is "supposed" to be SFW, so this is probably its attempt at censorship. Use a lora.

You can also try Chroma https://tensor.art/models/867658986558633399/Chroma-V32
which is some guy training Flux further in an attempt to add NSFW knowledge.

>>8878456
If you are just getting into this, don't. It's weirdly addictive.

Quick Q: how do I put ''prompt weights'' in SD Forge?

>>8879768
Isn´t like in Automatic 1111?, (token:weight), with values 0<x<1 decreasing weight (emphasis) and values 1<x<2 increasing it.

>>8879618
I think I'm in too deep desu. I've started prompting and also using engines online to streamline or fix some of my prompts. Found myself at work today questioning whether I could improve results by changing the order of my prompt last night. I'm done for.

>>8879877
It's not so bad with realism, you only get like three styles and few bodytypes to experiment with instead of thousands. So you'll get tired of it faster.

>>8879877
Wait until you find out about Infinite Worlds

>>8880174
I wish that were true.

What's everybody using to replicate character voices to say sexy shit? Is there a good website for it loaded with popculture voices, and a known good comfyui workflow?

>>8880174
I dunno, I'm busy trying to figure out how to use gel/danbooru-referenced prompts and to make an anime-esque image into a realistic image. I just want to see if hot cosplayers and egirls can coexist with great-looking tentacle sex rather than that rubbery shite you see in low budget JAV films.

>>8880181
Down the rabbit hole I gooo..

>>8874475
She's unbelievably beautiful. This is awesome!
If anyone could make a version with bigger titties, it would be even better.

>>8881034
nice

>>8881038
Holy shit.

What interface and models let you do this?

What's the necessary VRAM?

>>8881071
All here https://rentry.org/wan21kjguide
Especially the lightx2v section

The attached video generated in 106s on a 4080 (16 GB VRAM)

>>8881076
Hmmmm, not hopeful about my 8gb VRAM :-/

>>8880809
>see if hot cosplayers can coexist with tentacle sex
It's more like, you spend too long staring at the failures you forget what real photographs are supposed to look like. But everyone else remembers.

There's just no training data for real non-plastic tentacles.

>>8881086
>>8881086
Holy FUCK. That looks sensational.
Way better than what I've been getting!
I must admit I'd wondered if it had been down to my choice of words in prompt, or just that the models hadn't been introduced to the concept.

>>8881219
Used https://civitai.com/models/1539137/erotophilia-evolved
>masterpiece, photorealistic, 1girl, suspended, restrained, tentacle sex, vaginal, huge breasts, matsumoto rangiku, black kimono, necklace
No negatives. To be fair, I just ran the same test prompt on a half dozen checkpoints I had saved and thought this one's tentacles looked the best, no "skill" involved.

>or just that the models hadn't been introduced to the concept
In general there are two categories of realistic models. Ones that have never seen anime/cartoons like bigLust or Flux, these have no trouble looking like photographs but they have zero tag knowledge and require natural language prompts. And ones that were originally Pony, illustrious or NoobAI then merged with other stuff to try and push them into a realistic style. These use Danbooru or e621 tags for prompting. But the more tag knowledge they retain the less realistic they look, dipping into uncanny valley.

>>8876466
catbox?

>>8881241
It's not in Tensor yet but hopefully it shows up there. Your prompt is simple but clearly effective as hell, I think I have too much specifics in mind when I write one so it probably gets super clunky for a machine to understand and then output. I'll work more on making mine like yours.

Has anyone done photorealistic on anime locally? I'm experimenting with flux kontext, but it has real trouble making one character be a statue, while the other character is a person. Looks like it's not going to be possible without inpainting and two separate checkpoints.
I made the pic by photoshopping two generations, but it's still a base model that probably doesn't know that the stuff on the right are supposed to be the lower part of the man.

>>8881076
what are you prompting for the cum drip? If I put cum anywhere near my prompt i get a splash of water like from a bucket

>>8881320
french impressionist painting of a cute milkmaid chained to the wall in the palace dungeon, blonde, cleavage, dress, 18 years old, cum on tits, shot on leica, detailed photo, arms up, handcuffed

Negative prompt: oversaturated, overexposed, ugly, mutilated, extra limbs, extra fingers, deformed, disfigured, mangled, disorder, malformed limbs, fused fingers, twisted body, washed out, poorly textured, plasticized, render, matte, shiny, doll, 3d, 2d, drawing, anime, cartoon, child, clones, camera, lopsided, walking, airbrushed, mangled, broken, accident, amputee, lopsided

Chroma v 38 25 steps euler beta

>>8881506
thanks

>>8881459
I managed to get this far with regional prompter, a lot of emphasis on (photorealistic) on one side and (harada takehito) on the other. Problem is that only works on those uncanny valley models capable of doing both styles, which also means they're not good at either one.

Would make the inpaint job way easier, but then you need three checkpoints instead two.

>>8881608
catbox has failed me once again https://litter.catbox.moe/s4e1q7x31oqqmpa7.png

good job flux

>>8881647

Flux fucking sucks.

Requesting a cute , wholesome looking brown haired muscle babe nude other than sandals walking away from the viewer swishing a strand of anal beads behind her.

>>8881647
HAHAHA so if I'd asked for an ass on some chick's chest I could have got nips instead of botched flesh, problem solved.

>>8881700
Sadly most of the models I want to use are on Flux, so I'm going to have to learn myself somehow. Got any recommendations for models for flexible, realistic content?

>>8882037

I've heard that Chroma is a more NSFW and anatomy friendly version of Flux, but I've not managed to get anything out of it yet.

It's frustrating as fuck that tensor is just jam packed with so many god damn flux models :-/

do loras work for inpainting? I used a character lora (autismmix checkpoint) when I tried inpaint mrs. incredible's tits and it was just a mess.

What site please? (Asking for my friend)

>>8882107
It should, but inpaint on itself can be frustrating. Remember to check it´s on for the masked area, and that the area is large enough.

>>8882267
and that your prompt describes what's in the inpaint area, not the whole image
and that your denoise level is not too high

I don't know if it's the model I'm using but I cannot get images to be dark, and I cannot get the girl to do anything but stare directly into the camera

>>8882332
It is the model you're using, I don't even need to ask which. Nsfw finetunes rely on porn shoots and onlyfans for their dataset, which means studio lighting and making faces at the camera.

>>8882332
politely asked her to stare past the viewer/camera, avert eyes, etc?

>>8882358
I'm using CyberReal Pony, what other photoreal models should I be using? Obviously flux is better but it can't do porn

>>8866074 (OP)
I know it's a slim chance to even get a reply but I've been living under a rock recently. Is there any way to get celeb LORAs like
>>8871530
was looking for Nicky Minaj I see that civitai got nuked (along with alcohol kek)

>>8882670
In his case it's not a lora, just an edit of an existing photo. But Tensor.art hadn't nuked celebs, yet.

>>8882679
>tensor
curious, I previously searched some stuff and it all linked to 404 links. Thanks :D

>>8874475
Very nice.
Which site did you use to animate the image? Kling and Pixverse are heavily censored

>>8882670
In the AI celeb thread on /b/ someone compiled this https://archived.moe/b/thread/936056331/#q936057829 As usual, caveat emptor.

>>8882958
it is pixverse. theres a firefox script on simpcity which bypasses the nsfw filter for uploading and refunds your credits if it detects a nsfw image (but not ones without nudity plainly visible). even unlocks the 720p option.

>>8880922
i could try a bigger version but since it is pixverse, the jiggle physics on bigger breasts really sucks compared to wan.

waits on tensor suck, I'm probably gonna sign up and pay soon

Honest question. Why do anons always ask for a catbox link? Everytime I check it's the same file with the same resolution.

>>8883665
4chins removes metadata (ie the embedded workflow). catbox does not.
so if you want someone to share their prompt/workflow, ask for a catbox
not everyone will provide one, and some anons will catbox a file without embedded workflow just to fuck with you

>>8883666
I hadn't even thought about that. I always thought it was about getting a higher quality image. Now it makes sense.

>>8883505
Thanks! Pixverse is really good, I will try again.

Do you write your prompts long sentences describing what you want, or do you separate out each thing like "white, woman, big tits" etc

>>8884265
I think some models work better with one kind and others with the others. You have to check. Either way though, you can type what you want however you're comfortable with and have chatgpt process it. Or even work an llm into your workflow to do it automatically.

>>8884265
As >>8884332 said you should check the description the author provided. If they are based on Pony/illustrious/noobAI then they use exact tags from the Danbooru tag wiki. So "big tits" becomes https://danbooru.donmai.us/wiki_pages/large_breasts

For all others, it's up to how the author captioned their dataset. Which only they know, so hopefully they provided a prompting guide. It's common practice on Civitai at least, and if not you can look at metadata of the model's preview images, or the gallery.


That said, I always feel unsure about whether the model will understand my natural language prompt. And writing out an essay for one pic or having AI generate my prompt feel cringe. So I tend to just use tags everywhere, even if the don't work as well. Couldn't get anything good out of Chroma so far.

>>8884265
Depends. SD 1.5 is strictly token speak, but SDXL/pony/illustrious give you some moderate flexibility. Full purple prose it´s for Flux.

>>8884918
Cheers for explaining that, it's good to know the difference

>>8884265
this is supposed to depend on the model and how it was trained. 1.5 models generally were trained on images that may have had a few tags or were otherwise labeled certain ways in the LAION dataset. XL Models were a mix of images captioned that way and with longer captions, some generated with AI (Albedobase is my favorite example of this - you can get some excellent images with 5 tags and with long prompts, but they do look very different aesthetically). With Flux, the images are described by an AI, which then becomes the image caption.

While this can theoretically offer the proompters more accuracy, the tradeoff is twofold, the larger one being that AI language models are very verbose and no one wants to type a five paragraph essay into forge to get Asuka pegging Shinzo Abe. The second is that each small letter combination is reduced to a "token" and given the purple prose in AI language models, there are a lot of tokens which don't actually mean anything in terms of content but which become necessary for image quality. This is where a niche arises for a model like Chroma - beyond NSFW, it is basically a Flux model that has had a lot of the "padding tokens" removed and as such it is easier to prompt for.

>>8884265
>>8885327
it should also be stressed that what I have described as "padding tokens" in Flux are functionally similar to the "award-winning photography, trending on artstation" tags in 1.5 in that they don't mean anything content-wise, but they help steer the model to a set of images of high quality. Score_9 and its friends in Pony models do the same thing. Flux's equivalent would be the long descriptive strings that talk about photo lighting or background or whatever.

I spent a couple months fucking around on MAGE with a paid sub (before it went SFW) and I've already learnt more in the span of this thread.

>>8885859
Well yeah, most interactions tend to be between genners and their fans/subscribers, not each other.

>>8885983

There are no words for how much I want to fuck this non-existent woman

>>8886224
The speed at which this shit progresses is insane. Not two years ago it couldn't even make hands or backgrounds that made sense.

>>8870229
Cat box?

>>8886224
Mon dieu!

>>8881459
Did you manage to get anywhere?

I think the easiest way is to gen both in one style, then use heavy controlnet to re-generate the pic in the other style. Then splice them together, and pick either model to inpaint the edge where the characters meet.

>>8887638
img2vid that shit. I need to see how an AI makes an anime character and RL person interact with each other.

>>8887638
>>8887793
Could probably look good with more trial and error

>>8887802
Thanks for trying anon. I don't have the hardware for that.

>>8887638
>>8887793
managed it with img2vid with an anime illustration, then transport the character into real life.

Posted the result at /gif/, not sure how cross-board links work
>>29073055
Done with pixverse, no idea how wan would handle it, and pc isn't powerful enough to try.

>>8888269
Alright, I'll post it here too.

>>8888269
it's three arrows, then the board name with slashes
>>>/gif/29073055

>>8887638
>>8887802
oh great
new fetish unlocked

>>8887802
What did you animate that with? Local or public?

>>8888596
lovely

>>8886384
She looks like I imagine a character in a series of erotic novellas and short stories on Amazon and Kindle. If you like to read, the first book in the series is her as a Milf but the second is her as a schoolgirl losing her virginity to a teacher. "Laura 2 A Young Woman's First Time and further adventures an erotic novella."

The question is, how do I convince normies to pay me for this?

>>8889788
Just make a Patreon. The amount of no-effort shit that I´ve seen with the damned logo is unbelievable. That pic is better than two-thirds of the1girl crap I´ve seen.

>>8889800
I made a subscribestar but they say I need another repository. Im making a twitter so maybe that will work. I will make one, thanks anon.

anyone have a simple comfy workflow using facedetailer and sd upscale? i dont get any errors but it gets to the end and just starts repeating the last step and wont finish.

>>8890286
My wife likes your stuff. I may try some oil paints too.

>>8890520
The last step being Ultimate SD Upscale? It hasn't been the "ultimate" in over a year when it comes to tiled upscaling, use Mixture of Diffusers from https://github.com/shiimizu/ComfyUI-TiledDiffusion

...or just upscale without tiles like a normal person.

Requesting a pic like a babe like this but with way bigger curves

>>8891399
requesting you learn how to do this yourself

>>8890423
catbox this, I have a number of ideas but don't want to bug you for prompts and settings

>>8891648
https://files.catbox.moe/0c4ijs.png

>>8891202
i was under the impression that if you had a picture with many people in it, a tiled upscaler was the way to go

>>8891646
>>8891646
that's the only good request here

>>8891824
It just reduces peak vram usage when upscaling, at the cost of speed.

If you have a picture with many people it usually means their faces are further away from the camera and thus smaller. AI needs a lot of resolution to get fine details right, so if you don't want deformed faces you either have to upscale the whole image larger than usual, or inpaint the faces separate; manually or using Adetailer.

>>8892352
Very nice

>>8876467
>>8878049
>>8879683

What model did you use for those?

Realistically speaking, can you run Flux on a 3050 with 8GB of VRAM?
I heard the Schnell version is considerably lighter

>>8892682
You can run Chroma's gguf quantized versions. They are finetuned off of Flux, but not the same thing anymore.

>>8892690
Chroma seems interesting, I like that it incorporates booru style tags (or so I heard). Is it better or worse than Flux?
And also, is it compatible with reforge?

>>8892731
>>8892734
still would

>>8892731
>>8892795
>>8892867
Nice.

>>8893113
>Trade Offer
>I Receive: Big Humie Cock
>You Receive: Fat Gobbo Ass

Hello. I used to casually using Forge as I was first inspired by the /r/ request thread Realifying so a few guys shared a few basic details of how they set up Forge, and got myself from civitai the Pony v9 model. A bare minimum basic setup and then I just convert drawings to more 3d type of images, which I liked for a few days.

However, I tried including nipple clamps in the prompt in Forge but it didn't seem to gen properly. It looked more like nipple rings and nipple piercings.

Is clover nipple clamps or general nipple clamps too advanced for the AI? Any advice or any idea if anything can be added or need to be installed within the Forge directories perhaps? Or there are generic nipple clamps but it can work if typed into the prompt? Thanks in advance for any feedback/suggestions.

(Unlike AI clips/pics, for real life or live action examples there are various depictions of usage of clover clamps, either in non bondage submissive videos where the subject kinda challenge themselves to the pain, or those bondage clips where a bound girl is helpless against her torturer applying the clover clamps and it looks intense on her)

>>8893423
Forge is the user interface, not the AI model.

>>8893423
Regularly, if a token isn´t enough (in this case, nipple clamps), you look for a LoRA (a kind of specialised add-on). In your case, you can try something like this https://civitai.com/models/743123/nipple-clamps

Perfect Breasts flux will answer

>>8894282
gobbos cute
orc women with tusks cuter

>>8886384
Welcome to Pygmalion

I wish there was a way to filter images by face

>>8894667
Look I don't post >>8894610 THAT often

>>8894667
We seriously need an ai that lets you flag images from a particular poster and the ai automatically detects and removes their posts based on their style.

survey
https://strawpoll.com/XOgOVDj1Gn3

>>8895267
wait, haven't I already answered this one yesterday

>>8892681
chroma

>>8892225
True, but IMO there is something with images with 3+ faces where the faces look worse the more people there are, beyond what happens due to lack of pixels.

>>8893423
As mentioned you should use a LORA. Once you load the UI there'll be a folder under models where you save them.

...

>>8895448
>>8895449
>>8895459
>>8895460
>>8895461
>Live updates: Danish Prime Minister reveals the discovery of Rembrandt's secret goon stash

>>8895476
>>8895473
ok i've probably spammed this thread enough so here's my lazy attempt at a certain character

>>8895484
nah yours is easily the best content itt, good theme and good execution

>>8896108
you might be in the wrong place then, if you like drawings over photorealism

>>8886224
bravo

Is there an archive for previous threads?

>>8896419
you can search desuarchive for /asdg/ on /aco/

>>8896108
thank you

>tensor is kill
are there any alternatives?

>>8896837
You have these https://datahoarding.org/archives.html#CivitAI
Essentially, no. Hosting costs a mint, that requires either a very rich fool or charging your users, and with credit card processors being two steps beyond wahhabism on the puritan scale, there´s very little to do, except hoping p2p sharing takes off.

Using Forge i am trying to make the girl give a guy a blowjob while showing both full bodies, it ends up in both sucking a cock that connect their mouths together, how to fix? up to learn any extensions needed

>>8897056
But who's gonna pay for tensor without porn?

>>8897067
As an image, or as a video? What model are you using? Any image where an accurate depiction of a relationship between two objects (here, dick must be partially in mouth) is going to have a lower hit rate than your standard 1girl. I'd go on civit or wherever, find an image that's close to what you're looking for, and put that into the PNG info tab so you can copy its prompt.

>>8897182
yes let me give my credit card details to the obvious honeypot site

>>8896837
colab lol

>>8897420
u wot? I thought they killed that

>>8896837
afraid so. I was hemming and hawing over paying for a sub, was about to start but fuck that now.

>>8897955
These days, paying for a on-line generation site is pretty much a loosing proposal. The way they are getting straight out extorted is appalling.

>>8898169
*losing
compare: to lose/to be loose

>>8897774
not on the paid tier unless something changed in the past few months. i've mostly been using tensor but colab was my main one for a while

>>8886384
>>8899685
Which model and loras for these?

>>8899988
checked

do you ever add social media overlay to increase goonability of a lewd scene?
instagram story and snapchat snap make the most sense
but instagram post, reels, and snapchat spotlight are also sexy because they have likes and comments
any other ideas bros

>>8900489
here's a good idea, you can get your desired colors/lighting by putting a fractal or gradient through img2img

>>8900490
>tie-dye backdrop
definitively not a fan of this

>>8900779
It's not for everyone, that much is certain.

>>8900780
this ain't the 70s
it's not for anyone anymore

>>8901113
Adetailer is your friend. Helps with the damaged eye.

>>8900931
good gawd I want to believe

Baking.

New thread:
>>8901996
>>8901996
>>8901996

so as a guy that has had fuck all experience in terms of technical expertise outside of fucking about with MAGE/TensorArt, am I basically out of the game until 1) a new uncensored service debuts, 2) I buy a high end computer and learn how to run it all?