File: jyrsdw.png (356 KB, 512x512)

356 KB PNG

>>91122029
>>91122008
I do not know what XYZ grid or Prompt S/R are

Literally was away for a month, what happened

Can someone remind me please name of a 19th century artist that was drawing various bugs and plants. Very pretty pictures. Some mythology as well, if I remember correctly.
I forgot his name, unfortunately.

>>91122259
Maria Sibylla Merian

File: AI can't make this.png (406 KB, 600x600)

406 KB PNG

AI CAN'T MAKE THIS.

>>91122277
Thank you, I'll add her to my list, but that's a wrong person. It was a man and from the 19th century.

File: 5ej7ywhr.png (327 KB, 512x704)

327 KB PNG

>>91122293
gem

>>91122295
nakadashi

>>91122293
this is cool, was there a specific artist you mentioned to get this?

>>91122267
is that the right archive? I always thought it was archiveofsins, they're probably the same though. yeah it's hard to find it... try searching for pastebin links maybe

every xy grid script and extension is broken atm except for the most recently modified xy -> xyz script commits from yesterday. alternatively try the xy-infinity script, that also works, though to get it working with hypernets and loras you need my patches and i don't know how to make PRs lol.

File: 788dy8.jpg (50 KB, 572x436)

50 KB JPG

>>91122348
Sarah Andersen Kelly McKernan Karla Ortiz Lee Unkrich Leon Keer

were the artists i used. those first three are actually the three women involved with the SD lawsuit outlined here https://stablediffusionlitigation.com/ (scroll to the bottom)
so it's extra spicy

File: i67rjytd.png (284 KB, 512x512)

284 KB PNG

why is nai so much more creative than "fancier" models

File: 00187-833693357.png (1.31 MB, 832x1024)

1.31 MB PNG

>>91122336
kegarawashi。

File: 00605-202238145.png (2.03 MB, 1024x1024)

2.03 MB PNG

The horror ... the horror ...

File: catbox_aeudy3.png (2.72 MB, 1600x1088)

2.72 MB PNG

>>91122369
hmm, found thsi so far on the /hdg/ main, but ya it's short on detail:
- https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb

must keep looking:
- https://archived.moe/h/search/text/gist.github/

>>91122393
>he says as he posts an easily achievable image

>>91122378
thanks anon, from all their work, it seems like Leon Keer kicked in the hardest, interesting

File: like this.png (35 KB, 944x506)

35 KB PNG

>>91122198
...X/Y grid and Prompt S/R were both around a month ago

File: file.png (415 KB, 772x386)

415 KB PNG

what should I do if I'm trying to generate a character portrait? I'm attempting to generate a man in a suit with a Super 8 camera for a head, but everything I get is basically just a crude swapping of the head with a DSLR, a camera lens, or just a guy holding a camera instead. probably doesn't help that there are a million pictures of a person with a camera obscuring their face

File: 01699-590236100-Butterfly(...).png (1.1 MB, 768x1152)

1.1 MB PNG

>>91120745
>>91120763
/g/ pls...

File: 00003-2760224696.png (2.48 MB, 1328x1328)

2.48 MB PNG

>>91122498
>>91122498
ask chatgpt

>>91122295
Love these

File: 4.4.model3_00001_.png (2.59 MB, 1600x1088)

2.59 MB PNG

>>91122521
oh no its an ayylmao

this is gonna sound dumb, but I put (ayylmao:1.25) in my negatives and fix the alien-ification issue that occurs when upscaling some prompts.

>>91122541
just an img2img i'm goofing with

>>91122448
I will no longer ask. Give it up.

File: 00015-999552691-(1boy, 1g(...).png (1.18 MB, 1248x880)

1.18 MB PNG

Anyone have a princess peach LORA? This style would be really good for her

>>91122369
> is that the right archive?
Dunno, that's why i'm asking

> every xy grid script and extension is broken atm except for the most recently modified xy -> xyz script commits from yesterday
Well that's sad af, again need to have different versions of everything, so it could work correctly.

>>91122423
U sure about those script was compatible with dollchan? Just is it worth to spend a time to search?
Even if i find one, i have no idea how to make them work, if it's not gonna work out of the box.

File: 01659-3355975968-best qua(...).png (1.26 MB, 768x1152)

1.26 MB PNG

>>91122521
I wouldn't even know what to ask it, and if I did, I'd have Google'd a way to do it. Found plenty of ways to search/replace text in PowerShell or Python, not really too many ways to..."store", "replace", and "shuffle around" bits of text to match a desired syntax.

>>91122599
guess you're shit out of luck

File: h54t.png (160 KB, 512x512)

160 KB PNG

>>91122458
why does it sound so hard

any info on 4070 Ti performance in SD etc?

let's try to be more more social people
nobody is commenting on nobody else's works
we're all just dumping out our own brand of autism without any involvement, we might as well be blogposting but with images

File: 4.4.model3_00010_.png (2.37 MB, 1600x1088)

2.37 MB PNG

>>91122598
ya, it was for "users with no scripts" so presumably it will not be compatible with dollchan. oh well. even the new userscript on /hdg/ seems to not work with me, even with 4chan-x. hmm....

File: n5t.png (1.4 MB, 1024x1024)

1.4 MB PNG

>>91122628
because you're a dum dum
>>91122650
shut the fuck up

File: 77173-413944823-portrait,(...).png (2.72 MB, 1024x1600)

2.72 MB PNG

You don't remember the 1920s cyber vampire baseball team?

the degen thread on /b/ is the friendliest one, but....

>>91122717
who are you calling unfriendly, fag

>>91122717
but but what?

>>91122717
yeah....

and the /h/ one has the most tips and know how.

File: 00018-760500353-(1boy, 1g(...).png (1.24 MB, 1024x1024)

1.24 MB PNG

>>91122577
Alternatively, do you think I could train one on a 2060?

>>91122676
juwt very sleepy mkay
>>91122717
do not go to this wasteland

>>91122747
but...90% proud pedos over there, would be happy to see a bit more variety

File: 00091-20230125183439.png (1.22 MB, 896x1536)

1.22 MB PNG

>>91122628
Prompt search and replace is a very simple concept anon. What are you having problems with?

File: 5г76ло.png (318 KB, 512x512)

318 KB PNG

so sleepy didnt even attach an image

File: 4110135827-444fc829f5-229(...).png (1.57 MB, 1280x1024)

1.57 MB PNG

>>91122667
Noticed that it's not including boards.4channel.org domain and don't appear in tamper on the page as a compatible script

>>91122782
i'd say kill all pedos

File: wthbtr.png (352 KB, 512x512)

352 KB PNG

>>91122786
yes why is it expressed with complex math terms

File: image.png (176 KB, 803x463)

176 KB PNG

>>91122183
so auto1111 supports instruct pix2pix now
was playing around a bit.

File: shahsgfhsfg,.png (373 KB, 512x462)

373 KB PNG

Ok, after many times, I got an output that looks a lot like my SO.
Is there a way for me to build on it and do more with it? It took dozens of attempts to get this out of the prompt.

File: 9h5.png (1.4 MB, 1408x1536)

1.4 MB PNG

>>91122768
>>91122788

File: 00642-3119009293.png (1.99 MB, 1024x1024)

1.99 MB PNG

Holy shit the Dream Diffusion model sucks ass.
It's supposed to be good at dreamlike environments and shit but it just would not understand that I wanted Ancient Egypt in its prime and kept giving me ruins no matter what prompted. The moment I switched to another model it instantly knew what I wanted.

File: 29897-1241653265-looking (...).png (1.97 MB, 1344x864)

1.97 MB PNG

>>91122809
>replace X in your prompt with Y,Z,...

prompt:
(sad expression)
s/r
sad,happy,mad

will generate you three images - one with (sad expression), one with (happy expression) and one with (mad expression)

File: WeirdBloom.png (500 KB, 640x640)

500 KB PNG

stop talking about fucking pedos

how do I fix this bloom effect?

>>91122393
Not throwing shade but that's not exactly a unique image

>>91122765
>Alternatively, do you think I could train one on a 2060?
I think so yeah? Don't you just need 4gb or something? Good luck if you do. Also I tried that prompt you gave me the other day but it didn't come out as nice for some reason. What model are you using?

>>91122855
>Neuschwanstein
Not a real castle

File: j5yeh.png (256 KB, 512x576)

256 KB PNG

>>91122828
he will never

>>91122393
(Masterpiece)(best quality) 1girl nigga catgirl (fisheye lens)

File: Screenshot 2023-01-25 161822.png (522 KB, 1878x880)

522 KB PNG

>>91122817
I just get completely different images? Is there a special model I need?

File: Wolverton Fusion 01.jpg (1.4 MB, 3112x1392)

1.4 MB JPG

As I expected, Wolverton's style is so unique and the features are so distorted that the AI struggles to blend features together. There's plenty of strange bubbles and frills of flesh.

That ain't to say this was a complete waste - who knows, maybe these bizarre head shapes could be the foundation of a design for a monster.

>>91122809
You let things too easily intimidate you. If you recognize it as something that may be a math term then you already recognize it's functionality.

>>91122452
yeah, those graffiti artists and trompe l'oeil artists shouldn't be underrated!

>>91122866
AOM2 hard, but yeah they usually don't come out very good on the first pass. I think the LORA strength was too high. This batch as been better, especially after I upscale

((masterpiece)),(high quality),(insaneres), perfect face, detailed eyelashes, from above, (extreme closeup, zoomed in, leaning forward:1.2), (pov hands, pov hands holding hair:1.4), kissing face, head tilted up, finger in mouth, (arms spread, arms raised:1.2), (eyes closed:1.2), (face shot), puckered lips, [boku no hero academia],dramatic lighting, short hair, <lora:vae_compatible_afrobull_64_epoch-000005:.4>, <lora:kronya:.6>, kronya
Negative prompt: (worst quality, low quality:1.4), body shot, legs, midriff, breasts, cleavage, thighs, legs, (eyes open:1.2)
Steps: 24, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 2408456726, Size: 512x512, Model hash: 0fc198c490, Model: AbyssOrangeMix2_hard, Clip skip: 2, ENSD: 31337
(this was actually with no upscale, very cute)

File: Wolverton Fusion 02.jpg (1.66 MB, 3090x1531)

1.66 MB JPG

Basil Wolverton, I oughta say, he was a well known cartoonist in the 20th century.

File: 5hye.png (333 KB, 512x512)

333 KB PNG

>>91122440
>>91122865
okay so called free thinkers but I meant NAI outputs have more variety in terms of poses, emotions etc
>>91122854
this is such a simple concept, why would you need to explain it with grids or other weird math stuff

File: 00466-2408456726-((master(...).png (237 KB, 512x512)

237 KB PNG

>>91122918
forgot image, LOL

File: 04864-2617388706-masterpi(...).png (1.18 MB, 896x1280)

1.18 MB PNG

Is NAI still the best model for posing two girls? It kills me how much better solo pictures come out.

>>91122825
you could use variants from the same seed and prompt but it's not guaranteed to work
the best option is to train a lora on photos of her

File: 1674692582573.png (98 KB, 1200x675)

98 KB PNG

>>91122637
Slower than a 3080 10gb without xformers, faster than a 3090ti with xformers

>>91122881
yeah you need the instructpix2pix model

for me, turning it into a safetensors didnt work, and apparently the LDSR extension needs to be turned on if you disabled it (its on by default, built in)

I got the model by cloning this repo:
github /timothybrooks/instruct-pix2pix

then just manually running the
\instruct-pix2pix\scripts\download_checkpoints.sh

Use .\ in Powershell to call the script and run it. It will open whatever bash terminal you have for me it was git for whatever fuckin reason.

then just moving that over to Auto1111 from
\instruct-pix2pix\checkpoints\

it takes a while to download.

File: 1664649479640-639982253-a(...).png (399 KB, 512x512)

399 KB PNG

Ellie kinda cute tho

File: Wolverton Fusion 04.jpg (1.12 MB, 2035x1409)

1.12 MB JPG

And the last of the four.

File: 00653-592569963.png (686 KB, 512x960)

686 KB PNG

You ever try out stuff on models that are not intended for it just to see what would happen?
For instance, this is the Uber Realistic Porn Merge.

File: 4whtea.png (178 KB, 512x512)

178 KB PNG

see? this >>91122940 is what I was talking about, you're better off using AI and then funneling the output thru other models
>>91122919
>>91122937
new jak variants are getting wild
make some racist caricatures

>>91122930
Because it is a grid anon. It is a grid that uses x and y variables.

Your pics are also not selling NAI over something else any time soon.

File: 00398-789051771-((masterp(...).png (322 KB, 512x512)

322 KB PNG

>>91122949
Thanks!

>>91122930
yes
i agree on that part, its why i still love anythingv3

>>91122650
sometimes there just isn't much to say

File: 1664649479645-889125386-a(...).png (375 KB, 512x512)

375 KB PNG

Joel looking gruff

>>91122998
i feel
fantastic

File: erherh.png (212 KB, 512x640)

212 KB PNG

>>91123009
>anythingv3
sucks ass in comparison to nai variety-wise
>>91122965
>AI
n-naI I Mean

btw I made a mix, deleted it and now stable diffusion keeps crashing cuz
>RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper__index_select)

File: 08183-900.png (1.95 MB, 1024x1536)

1.95 MB PNG

so i just did a big batch
i had (thigh_high_boots:1.1) (fishnet_stockings:1.3) in the prompt and it was the only thing affecting the legs or anything below the waist honestly
how come i struggled so hard to get actual fishnets? this is the only prompt that had the fishnets i was after, but its not on all her legs

File: cowgirl8.png (326 KB, 400x400)

326 KB PNG

Getting a cartoon face that isn't...off...is quite the feat. I don't remember if I posted this before, but either way, it's probably one of the few good-looking ones I have as far as the face goes

>children's book, 12-year old centaur cowgirl, black beady eyes, elf ears, cute, hands on hips, light smile, (jet black hair), cowboy hat, white shirt, vest, sky

File: Hatching a Scheme.png (657 KB, 1024x1024)

657 KB PNG

Donald Trump + Nutty Professor + Mr. Bean

>>91123035
He looks cleaner than in the game.

File: catbox_u6fey4.png (2.47 MB, 1600x1088)

2.47 MB PNG

>>91122791
Ah, thanks, and it seems to drag-and-drop upload, the user must hold CTRL while doing so.
Modifications, only two lines, the last one with 'sdg' and the match include for "4channel...g":

// ==UserScript==
// @name        4chan /hdg/ catbox.moe userscript
// @namespace   4chanhdgcatbox
// @match       https://boards.4chan.org/h/thread/*
// @match       https://boards.4channel.org/g/thread/*
// @match       https://boards.4chan.org/e/thread/*
// @match       https://boards.4channe.org/vt/thread/*
// @grant       GM.xmlHttpRequest
// @grant       GM_xmlhttpRequest
// @version     2.4.2
// @author      Anonymous
// @description Upload image directly to catbox.moe from 4chan
// @updateURL   https://gist.github.com/raw/ca46eb79ce55e3216aecab49d5c7a3fb/catbox.user.js
// @downloadURL https://gist.github.com/raw/ca46eb79ce55e3216aecab49d5c7a3fb/catbox.user.js
// ==/UserScript==

(async function () {
  const TITLE_CHECK = ['/hdg/', '>NAI LEAK SPEEDRUN', '/sdg/', '/vtai/'];

File: 1664649479646-889125387-a(...).png (287 KB, 512x512)

287 KB PNG

>>91123073
He has sad eyes though.

>>91123068
jesus christ, amazing

File: 04907-1325665781-1girl, [(...).png (1.19 MB, 824x1032)

1.19 MB PNG

Testing a non-comfyui, regular webui image to seee if right-click on the image download link in 4chanx shows metadata without needing to download the image, because it seems it won't find comfyui's metadata automatically

so many steam games left untouched....

File: 08169-902.png (2.15 MB, 1024x1536)

2.15 MB PNG

this is a pirate btw

>>91119523
prömpts?

File: catbox_lfy9rs.png (792 KB, 576x768)

792 KB PNG

>>91123100
okay it didn't even upload that one, how weird.

File: 00104-20230125191412.png (1.52 MB, 896x1536)

1.52 MB PNG

>>91123107
This is a cashier

>>91123056
give me a prompt and lets do a grid to compare
same samplers/res all that

>>91123113
https://files.catbox.moe/08yxfk.png

File: Screenshot 2023-01-25 at (...).png (296 KB, 1944x887)

296 KB PNG

>>91123114
Okay, it uploaded that one just fine. It seems the new catbox upload script:
1. expects you to hold ctrl when dragging and dropping to upload to catbox. otherwise, normal upload occurs.
2. doesn't upload certain filenames to catbox (it's not filenames with spaces in them, i can't figure it out).

>>91123126
after hours maybe

>>91122183
How do I apply multiple LORA pt files to a model? is it literally just target a base model checkpoint, apply a LoRA pt file, and then for the next LorA pt file, target the checkpoint generated from the previous LoRA, rinse and repeat?

File: 00677-1619220891.png (640 KB, 512x960)

640 KB PNG

Look upon my works, ye Mighty, and despair

>>91123103
I don't want to play vidya anymore I want to generate random stuff.

>>91123131
(DSR_1.5) masterpiece, girl, (bold lines_0.5), cat ears, cat tail, (brown skin:1.5), black hair, short hair, flat chest, (plaid skirt), (hoodie:1.1), (solo:1.6), (smug)

add an angle/pose/situation of your choice, NAI will have less coherent but more creative results

File: 1665974058054.png (1.15 MB, 768x1280)

1.15 MB PNG

>>91122918
I'll give it a shot.

File: 01731-3554272054-Gordan f(...).png (2.14 MB, 1200x1600)

2.14 MB PNG

Ok so I kinda finally found out how to train a dreambooth model how do you extract the LORA out of for the smaller file size and to be used on other models is it inside the Webui or is it done in another script separate from automatic1111?

>>91123237
You would need to have a baseline definition of what constitutes good concurrency and creativity.

>>91123202
Giant tits hidden under the sand

Do I have to keep prompts under 75 tokens? What's the info on that?

File: w4th4t.png (318 KB, 512x704)

318 KB PNG

>Giant tits hidden under the sand

>>91123268
https://www.reddit.com/r/StableDiffusion/comments/10kuzmh/how_to_extract_small_lora_file_from_custom/

>>91123294
do you want your pc to melt?

>>91123114
b-but I want the other one

File: 4.4.model3_00011_.png (3.16 MB, 1600x1088)

3.16 MB PNG

>>91122964
how did you add the numbers in the top-left of each sub-grid?

>>91123313
I'm already running gens all day. What's the information?

File: grid-0093.png (631 KB, 1024x512)

631 KB PNG

>>91122941
>train a lora
But anon, the Lora how-to guide is completely useless because the faggots that made it expect everyone to be using Linux and know exactly what they mean when they use shorthand and jargon.

>>91123294
Bro you DO want to keep your GPU alive r-right?

File: 01728-3753226282-Gordan f(...).png (285 KB, 512x512)

285 KB PNG

>>91123311
Thank you anon I love you.

File: 34524-2864363506.jpg (498 KB, 832x1280)

498 KB JPG

>>91123294
no, and there are some technical implications about 75 token breakpoints but honestly you don't have to worry about it

>>91123325
i manually did it in mspaint lol, i was going to ask which model produced the most soul but stopped caring right before i went to write the post

File: 00695-2520562245.png (496 KB, 512x960)

496 KB PNG

SINK INTO THE DARKNESS!

File: 1664649479649-2854829881-(...).png (583 KB, 512x512)

583 KB PNG

what would happen if i kissed a zombie

>random seed
>early novelAI tier if I'm lucky
>loot a seed from /g/
>something good
Why?

File: 09330-2676946700-(DSR_1.5(...).png (224 KB, 512x512)

224 KB PNG

>>91123353
>which model produced the most soul
I think 3 is the most versatile of all
>>91123399
early NAI is gold

mixes will never make this VVV

File: 1651627103679.png (742 KB, 664x992)

742 KB PNG

2400048824
here u go

>>91123422
What you posted isn't the kind of early AI I'm talking about.

>>91123342
Seriously, where the hell do you put the files and where do you run the scripts? There are two script folders, so I put the files in both, but no matter what, they won't run properly.

>>91123385
You would probably die of ligma.

File: 29986-233480711-masterpie(...).jpg (306 KB, 768x960)

306 KB JPG

>>91123422
you do like NAI huh, that's 50:50 nai/cafe i used to use that mix a lot, it's nice

File: 00049-1042096867-masterpiece, .png (368 KB, 512x512)

368 KB PNG

This got real meta, real fast.

>>91123467
Who's Steve Jobs?

File: 05659-48938188-1girl,(mas(...).png (2.3 MB, 1288x1536)

2.3 MB PNG

>>91123349
Pls post your gordon lora.

>>91123488
Ligma balls

>>91123473
Now generate an image of a guy generating AI art.

File: compositing_test.jpg (3.09 MB, 1280x6336)

3.09 MB JPG

Applying prompts only to specific parts of the image while sampling seems to work pretty well now that I got the math right.

File: 00048-1042096867-masterpiece, .png (306 KB, 512x512)

306 KB PNG

>>91123510
I was actually trying to generate an image like this but the AI read my thoughts of wanting to become an artist even now and pulled a fucking /ic/ on me. I didn't flee that place to have an algorithm lecture me.

>>91123473
prompt?

>>91123580
The AI is mocking us because it can draw better than us.

File: catbox_79zo0p.png (2.39 MB, 1536x1024)

2.39 MB PNG

>>91123080
It works too, but not with dollchan too as expected

i argue the bigger the boob one is into the lower the iq

>>91123456
Like, what the fuck?
I have the latest ss.
I've put the fucking files in every single "scripts" folder in the install, but no matter where I execute the commands they tell you to, I get errors. It won't fucking run.
How fucking hard is it to just tell the user exactly where to put shit and exactly where to run shit instead of lazily saying "throw it in the scripts folder" like a fucking retard?

>>91123592
masterpiece, best quality, high quality, absurdres, large breasts:1.2 inconvenient_breasts, ((drawing)), pencil, sketchbook, desk,
Negative prompt: lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name
Steps: 30, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 1042096867, Size: 512x512, Model hash: fbcf965a62, Model: anything-v4.5, Clip skip: 2

>>91123602
I know, right? It's like those faggots saying "hurr you can draw if you sacrifice ten hours a day and do nothing else", sure thing buddy I can do some garbage realism in three months if I do nothing else but draw.

File: Screenshot_20230125-200817-080.png (169 KB, 1080x1126)

169 KB PNG

for fucksake, I've heen shadow banned. I was wondering why all of a sudden my pics were getting almost no views or likes on twitter

File: 00002-3628080963.png (3.49 MB, 1152x1728)

3.49 MB PNG

>>91123655
at least not perm banned like me, twitter doesn't like when you tell a woman that you hope her cancer returns

File: 08222-2515463375.png (819 KB, 1024x1024)

819 KB PNG

>>91123422
i tried
gonna delete those outputs now..

idea for Voldy:
a message box appears that says "warning: you are using an inpainting model. would you like to switch to [non-inpainting equivalent]?" when you try to start a batch in txt2img with an inpainting model

it could have a "do not display this message again" checkbox

I have ruined so many batches by forgetting to switch models

>>91123636
Bad opinion

File: 05668-2139448839-1girl,(m(...).png (2.21 MB, 1136x1360)

2.21 MB PNG

It has been really hard to get a long skirt.

>>91123734
actually I've changed my mind, to simplify it should just say "Start batch anyway?", forget about the needless bloat of "switch to x?", I don't think anyone minds switching manually

>>91123755
try short dress instead

File: 77746-1777870819-((master(...).png (2.5 MB, 1024x1600)

2.5 MB PNG

>>91122918

File: 108061-3554394729.jpg (896 KB, 1024x1536)

896 KB JPG

hm labeled color swatches nice

File: 1672848935180149.webm (960 KB, 512x512)

960 KB WEBM

>>91122944
oh wow fersh test, nice

File: catbox_pu4zwu.png (2.65 MB, 1600x1088)

2.65 MB PNG

>>91123631
pity, but it is good it at least works with the modifications on another's machine. how is dollchan, why do you like it more than 4chanx? Does it work on firefox on the newer userscript extension called "FireMonkey"?

File: 00019-1914637813-(1boy, 1(...).png (1.56 MB, 1280x1024)

1.56 MB PNG

>>91123263
Cute! The hand feet are great lol

File: catbox_ui0fh9.png (455 KB, 512x512)

455 KB PNG

>>91122855
Insufficient attention (some concept in stable diffusion) has been allocated to the trees, the building, etc. - or, you are doing img2img with a very low denoise. A long time ago someone wrote you could add things like "arboreal tinge" to fix the blurring on trees. Alternatively try specifying a style but with low weight like (john william waterhouse:0.8), i find that to add sharpness, pic related, a low-res with a bit of fine detail, although you can see that the detail is almost "over-fine".

File: 00020-1180614051-(1boy, 1(...).png (1.21 MB, 1280x1024)

1.21 MB PNG

>>91123522
Can you elaborate, is that different than inpainting?

File: 01705-4081437461-1girl, l(...).png (410 KB, 440x768)

410 KB PNG

why does NAI suck so much?
it's literally a model that you have to PAY FOR and it is still inferior to the anythingv4 you can download for FREE
why?

File: 31077-563492582.png (1.56 MB, 1024x1024)

1.56 MB PNG

I miss my landscape pals...

File: 08230-50.png (2.32 MB, 1280x1280)

2.32 MB PNG

>>91123896
>anythingv4

lol
lmao even

File: 234.png (1.26 MB, 1024x1024)

1.26 MB PNG

Made a youtube channel for AI stuff...wish me luck bros. The trend is only half-month old.

>>91123916
what's wrong with it?

>>91123896
because they can. Tons of people wanna generate images but have lack either knowhow, or the computer specs

File: 106533-3395353238.jpg (556 KB, 1792x672)

556 KB JPG

>>91123914

File: 00022-141659808-((masterp(...).png (1.2 MB, 1024x1024)

1.2 MB PNG

>>91123942
Anythingv3 was the last serious one, v4 and v4.5 were made as memes.

Wildcards are super fun. Anyone got ideas for more cool poses/outifts?

>>91123964
>Anythingv3 was the last serious one, v4 and v4.5 were made as memes.
elaborate pls, it gives me good results

>>91123830
> how is dollchan, why do you like it more than 4chanx?
Not so well at the last time AFAIK, but was using it for so long, cuz it works on many boards and got too familiar with it.

> Does it work on firefox on the newer userscript extension called "FireMonkey"?
Have no clue, try it if you want to

>>91123964
>Anythingv3 was the last serious one, v4 and v4.5 were made as memes.
Wouldn't say that myself. But AOM2 and 7th and plus derivatives like grapefruit arguably just are better rn.

File: 31075-2467188547.png (1.66 MB, 1024x1024)

1.66 MB PNG

>>91123953
I wanna live in that lighthouse

File: 4wthr.png (323 KB, 512x512)

323 KB PNG

>>91123696
caught lackin

File: 77749-1312795325-portrait(...).png (2.46 MB, 1024x1600)

2.46 MB PNG

>>91123964
rename to things.txt

https://files.catbox.moe/er1tyh.txt

put __things__, __things__, __things__, __things__ into your wildcard prompt

Anyone know why paperspace shows a black screen whenever I try to verify my phone number?

File: 1673581851617356.png (321 KB, 696x623)

321 KB PNG

Did dreambooth update their bitch-ass extension yet?

https://rentry.org/clipfix

https://github.com/arenatemp/stable-diffusion-webui-model-toolkit

Anyone smarter than me had a go at this? The broken clip seems interesting

File: ComfyUI_08261_.png (3.25 MB, 1920x1088)

3.25 MB PNG

>>91123894
It's a bit like using AND but instead of applying prompts to the whole image they are applied only to parts of it.
In those images I have 3 prompts, one for the whole image, one that describes the fox girl applied to the center left of the image and another one that describes the pink catgirl applied to the center right of the image.

Not sure how useful it is yet but it's fun to play with.

someone prompt "chinese characters"
i want to see what it spits out

>>91124050
>__things__, __things__, __things__, __things__
Isn't that just 4$$__things__

>>91124120
it depends on the model and the anon who does this for you

File: catbox_j9ntqr.png (2.84 MB, 1600x1088)

2.84 MB PNG

someone please add /hdg/'s catbox-metadata extension to the /sdg/ header, it now supports all boards and multiple webui interfaces' metadata formats: https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb

>>91123353
model 2 or model 7. these are crazy for model merges if they're diverging to much in detail, wow!

>>91122944
hahah there's amd oh man... why didn't i get a 3060 oh right nvidia drivers...

>>91123522
note, be aware that the anon at /hdg/ responsible to maintaining the newly-updated-to-support-all-boards catbox metadata extension (ctrl-click to click-to-drag upload in 4chanx) added support for comfyui's "prompt" and "workflow" metadata a few minutes ago: https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb

File: 00025-2654770025-((master(...).png (1.45 MB, 1280x1024)

1.45 MB PNG

>>91123981
It's essentially just AnythingV3 with some extra stuff mixed in that doesn't actually make it more coherent or better, just different. So it's less of a v4 and more of a v3a or something

>>91124050
I'm already using UMI for wildcards, it's really cool

>>91123522
>specific parts of the image
You can do the same trick nvidia eDiffi can do?

>>91124140
>someone please add /hdg/'s catbox-metadata extension to the /sdg/ header, it now supports all boards and multiple webui interfaces' metadata formats: https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb

Why?

File: t4hwt4.png (271 KB, 512x512)

271 KB PNG

>>91124120
okay

File: 109082-3068283928.jpg (1.05 MB, 1920x1080)

1.05 MB JPG

>>91124089
i fixed the clip in my jamix merge and after comparing a bunch of outputs back and forth i tended to like the results from the original broken version better lol
it's possible, maybe even probable, that comparing equivalent prompts/seeds is an unfair comparison in this case because the fixed version should need less emphasis and what looks good in the broken one might not directly translate

>>91124140
nah they're mostly completely different models, just what i had laying in my models folder. 2 is jamix and 7 is just anythingv3

>>91124009
word

>>91122577
Do you have the d-art one?

File: 71925-4005047186-dreamlik(...).png (2.4 MB, 1600x1024)

2.4 MB PNG

>>91124120

>>91124183
So it might be worth biting the bullet and fixing ones that need it and not worrying about existing prompts?

>>91124183
Prompt?

>>91124199
last one looks a bit like 布
they all look like 绷

>>91124172
It adds:
- Adds upload-via-catbox in the 4chanx interface: file is uploaded to both 4chan and catbox, for reasons that follow.
- Images uploaded via catbox in the 4chanx upload interface are named with the catbox ID for the benefit of folks with neither the extension nor 4chanx.
- Right-click on the download link for an image to show only metadata without downloading the entire image. Metadata looks like this and can be easily copy-pasted by clicking on the little blue C button: >>91123139
- Actually downloading the file downloads the file with pnginfo or exifinfo as appropriate.

File: 31095-2062998162.png (1.64 MB, 1024x1024)

1.64 MB PNG

>>91124183
Very impressive work

File: 109074-1060092725.jpg (956 KB, 1920x1080)

956 KB JPG

>>91124200
well let me not overstate it, the results weren't dramatically different, the outputs were obviously slight variations on each other. but in most cases i just liked the old version better. i want to do more tests later because i want to like the fixed version more but i'm not using it yet lol

>>91124220
let the meta tags flow:

masterpiece, best quality, highest quality, cinematic lighting, (volumetric lighting), CG unity 8k wallpaper, focused, 8k wallpaper, 4k wallpaper, (ultra realistic), (photorealistic), sharp focus, absurdres, (HDR:1.2), [high contrast], (photograph), [instagram], ((photography on film)), (32mm), detailed and intricate, ((movie still)), octane render, film, kodak, [kodachrome portrait], ((subsurface scattering)), cgsociety, [grainy photograph], ambient occlusion, subtle bloom effect, hyperrealistic, highres,

(((beautiful landscape)))

(by Alois Arnegger, by Andreas Achenbach)

>>91124255
Inferno Cop got a girlfriend?

File: 768d5f316f083e8ef3f41f06e(...).jpg (71 KB, 736x890)

71 KB JPG

I have a fetish for pretty women reading Vogue magazine. Can someone prompt some so I know what it can do? Thanks.

File: ComfyUI_08279_.png (3.4 MB, 1920x1088)

3.4 MB PNG

>>91124140
>added support for comfyui's "prompt" and "workflow" metadata a few minutes ago
Nice.

>>91124171
Right now it's limited to rectangles but yes.

>>91124269
Sure, why not. It was an attempt at hell rider from smt etc. the girl was a unasked for bonus.

>>91124306
Do Trumpeter next.

File: 00026-92720389-((masterpi(...).png (1.46 MB, 1280x1024)

1.46 MB PNG

>>91124109
That sounds great for multiple different people in a pic! I've never had success with AND, it always ends up looking really bad

>>91124292
no

>>91124336
noooo ;___;

File: tmpfz7tevy9.png (417 KB, 512x704)

417 KB PNG

>>91124109
I feel if you could capitalize on that type of logic, you eventually can end up with environments being generated almost in a like, game engine esque sort of setup. One to paint the environment, and then descriptions and prompts for each of the actors.

>>91124292
Very specific fetish. I can kind of see the appeal. Is it because it makes them look simple-minded?

>>91124295
c.f. >>>/h/7144075
Suggestion: make it possible for the user to paste text into the UI rather than loading a workflow.json or file.png, with the expectation that the metadata is copied to the user's clipboard from the userscript and looks like this (so it is the entire below string):

prompt: {prompt-json-here...}
BLANK LINE
workflow: {workflow-json-here...}

i.e. some regex like /^workflow: (.*)$/ extracting the json in capture group 1 (but my regex-fu is rusty so i dunno how to write it).

File: tmpt_rpkpv9.png (1.72 MB, 1280x1024)

1.72 MB PNG

>>91124323
>Trumpeter

File: ana-cosmo-spain.jpg (1.08 MB, 1536x2048)

1.08 MB JPG

>>91124380
No, I started fapping to Vogue when I was 8 and I like women too so I want both in the pic. Vogue gives a posh vibe, if I want them looking simple-minded then they gotta be reading Cosmopolitan.

File: 01725-1484532775-1girl, l(...).png (526 KB, 576x768)

526 KB PNG

>>91124345
i thought you were trolling anon
it can't generate images of the vogue magazine if it hasn't seen yet and i'm pretty sure sd hasn't seen enough of it to look good, but some other anon may try it
what I can do for you is an amnime girl reading a book

>>91124380
Vogue is for classy people, what are you saying?

File: 1672935648593.png (1.02 MB, 1072x712)

1.02 MB PNG

>>91124170
>UMI
What's that?

File: tmpcf9t2c38.png (898 KB, 1024x1024)

898 KB PNG

For me is anything-v4.

File: ddzpy7sedbu71.jpg (93 KB, 640x1138)

93 KB JPG

Ill make some food, please prompt some pretty ladies reading Vogue in this thread or in the next.

>>91124451
>>91124451
>>91124451
new thread

File: 103270-1361708713-2022. r(...).jpg (1.61 MB, 2048x1408)

1.61 MB JPG

has anyone experimented with the instruct2pixel stuff?

>>91124424
http://magazinefetish.com/

Here's a community of coomers who masturbate to Vogue

File: 00387-4027660221-(masterp(...).png (342 KB, 512x512)

342 KB PNG

>>91124460
wut dat

>>91124463
first post i see on the page
>Kent to manchester I have drawn a blank to buying magazine where i can go to the houses an personally collet but around my home area of wythenshawe theres hundreds from 50p a mag to bundles for £10 i get very aroused when collecting them

it takes all kinds indeed

File: file.jpg (110 KB, 466x486)

110 KB JPG

>>91124400

>>91124516
Is that... him...?

>>91124525
lol

File: 02347-1745528113-portrait(...).png (2.13 MB, 1088x1472)

2.13 MB PNG

testing

"exit code: 1
stderr:
C: No module named pip"

Why am I randomly getting this after its been working for weeks

Is there a booru like website where we can see all the prompts used to generate an image? Or just a prompt website in general to get samples from? Trying to prompt from nothing is really tiring.

>>91124467
Fuxking mad lad prompt

>>91124810
the galleries on civitai have an icon to display the prompt and copy gen details

>>91124810
lexica.ai, the ai boorus don't actually give you the prompt itself. however there is an ai booru where you can directly search tags somewhere, go check the sdgoldmine rentry.

more useful would the the lexica-style clip-guided tag search. since you don't actually feed sd a tag but a text embedding vector, you can search "a house on a hill overlooking a lake" which lexica turns into a text embedding, then feeds to a database of things that are "semantically" (literally geometrically) close to the embedding vector -- this process works quite well, and though the results' prompts might not have any of the words in your search string, if you plug them into even the anime models, you'll get good results.

Why the fuck is it so hard to rename every file in windows just 1, 2, 3, 4? Why the fuck is it so hard to filter the brackets windows tacks on?
And I swear to god, if I don't have to actually name every file in my training set from 1 up and the retarded guide writer didn't make that clear I'm gonna shit in their mother's mouth.

>>91124434
https://www.patreon.com/posts/75914044
It's an Auto111 extension. It's free, the patreon is just for fundraising I think?