So I just set up ComfyUI the other day. Downloaded Illustrious 2.0, got some western style cartoon Loras, and I just cannot make it produce anything but anime style pictures. It doesn't give a shit about positive prompts or negative prompts when it comes to that. Everything that pops out is anime.
Obviously there's a way to fix this. Is there a specific prompt I need to use or a specific Lora or what's your recommendation?

File: 01020-gobbo_goblin_girl,_(...).png (2.79 MB, 1296x1920)

2.79 MB PNG

Thank you for the bake, sir.

File: 01025-1girl,_orange_eyes,(...).png (2.26 MB, 1296x1920)

2.26 MB PNG

>>9065803
Mostly artist tags and style loras. I recommend perusing civitai for style loras as inspiration. Download the ones that stand out as a unique style, mix and match them.
Get the stealth png info extension/node and take a look at what other anons are using.

>>9065844
>Mostly artist tags
bro, you're a lifesaver. thank you

File: 1762817432416375.jpg (161 KB, 1664x2432)

161 KB JPG

https://civitai.com/models/530111/diana-of-themyscira-wonder-woman-2009

If anyone wants to make gens with this model

>>9066682
saar you have your own thread

>>9066686
sorry, 0 idea where to post anything anymore. Every thread has their own rules these days.

File: IMG_7219.png (513 KB, 832x1216)

513 KB PNG

>>9065878

File: alex toilet.png (3.02 MB, 1664x2432)

3.02 MB PNG

She belongs to the public

File: Mia Rodriguez.png (3.52 MB, 2003x2408)

3.52 MB PNG

Requesting this (https://files.catbox.moe/9ljkr8.png) with Mia Rodriguez (Baskup).

File: IMG_7211.png (503 KB, 832x832)

503 KB PNG

>>9066761

File: IMG_7212.png (696 KB, 832x1216)

696 KB PNG

>>9067208

File: IMG_7217.png (352 KB, 554x810)

352 KB PNG

>>9067223
ok last one

Dunno if that magazine anon is around

>>9066466
>>9066593
very good.

File: 1762664287919209m.jpg (95 KB, 500x1024)

95 KB JPG

>>9065702
Has anyone saved these compilation images used in the OPs?

>>9067717
https://desuarchive.org/aco/thread/9026702/

>>9067851
Can you do some ass shots of Alex please?

File: PK9YX1BZ4F6FYDQ8RY1J7A5AE0.jpg (154 KB, 832x1216)

154 KB JPG

>"Wait, what? You're talking to me?"
>"Filthy goblin."

File: G8WPX2DYJBRBSSDDHH77NK1A70.jpg (150 KB, 832x1216)

150 KB JPG

>>9067900
>"Just because you got me pregnant and e're married doesn't mean I like you."
>"Filthy goblin."

File: alex selfie rotated.png (2.81 MB, 1664x2432)

2.81 MB PNG

>>9067852

File: alex beach.png (3.64 MB, 2048x2432)

3.64 MB PNG

>>9067852
>>9068240
here's one more I never posted here

>>9068240
>>9068242
I like that stuff! Thank you!

>>9068242
always liked her and her mom ... the old design of her mom and the others

>>9068240
>>9068242
Can you do ones with Brittney but full frontal?

She never gets any love desu

>>9067900
>>9067922
Love the implicit story!
Do you have a Twitter or somewhere you post these?

>>9065702
>AI image upscaler model database
1xDeJPG_realplksr_otf -> fast, sharpens image, maybe too much sharpening without blending in an image editor.

x1_ITF_SkinDiffDDS_v1 -> brightens skin tone, fixes some compression artifacts

File: same selfie rotated.png (3.27 MB, 1664x2432)

3.27 MB PNG

>>9068240

File: 9T5FSQ4NWWYXAMEXVTXZDJCBS0.jpg (151 KB, 832x1216)

151 KB JPG

>>9068572
Thank you. I don't post anywhere, but I've been thinking about creating some sort of conquest game where you'd capture and breed heroines. I can't decide if it should have darkest dungeon style rpg combat or something more like capturing settlements on a map though.

>>9065702
How do you guys get good backgrounds? Or is it all just in the styles? It feels like every time I try, it's either just empty or nonsensical

>>9069537
straight txt2img gens suck on just about every model, closed or open source, although models can hide jank with natural or simple backgrounds and effects.
i leave "simple background" on every gen.

File: 01036-gobbo_goblin_girl,_(...).png (2.79 MB, 1920x1296)

2.79 MB PNG

>>9069537
Just tell it what you want. Use scenery loras or artist loras that have a lot of backgrounds in them.
Simple tags like "indoors", "outdoors", and "scenery" work as well as any location you can imagine. Try a different model if you have to fight too hard.

File: instagram_dreamscape_aist(...).jpg (453 KB, 1636x2048)

453 KB JPG

Can anyone help me find what style/lora this uses?

File: sam sinister.jpg (1.77 MB, 2432x2048)

1.77 MB JPG

Sam looks downright sinister when she gets dicked down

File: sam bouncing.png (3.27 MB, 1664x2432)

3.27 MB PNG

I am addicted to Sam

File: ComfyUI_00970_.png (1022 KB, 1024x1024)

1022 KB PNG

slowly but surely getting better at this. the dataset is japanese stuff so it simply does not understand what a weed pipe looks like - every pipe is a kiseru. this isn't 100% there, but it's miles ahead of what it's been spitting out and now it's roughly in the realm of 'good enough'. I used unClipConditioning to get this. is there another method you'd recommend or is this basically my best bet?

File: 1000026508.png (1.55 MB, 832x1216)

1.55 MB PNG

Im back from a ban!

Requesting some in-style/on-model Lola Bunny.

What models/loras are you using to get western comic book artstyles done well using SDXL? I've mainly done photorealism and anime in the past and have been looking at switching it up, but it's harder finding good sources at this.

File: 2loras_test__00384_.png (763 KB, 768x768)

763 KB PNG

any obscure characters you really want to see more of? Thira here has only about 30 images to her name.

File: maggieride1.jpg (133 KB, 1024x1024)

133 KB JPG

>>9074055
Most characters I want are too obscure for models, but I feel that that Ava's Demon hardly got any. Especially Maggie.

>>9073377
what'd you get banned for?

File: 1753427060444207.jpg (314 KB, 1152x864)

314 KB JPG

Anyone wanna host a Christmas party with these ladies?

File: thirablows.png (946 KB, 768x768)

946 KB PNG

>>9074062
I get that, had to train most of this one myself, Still very jank.It gets there sooner or later

File: thirablows2.png (777 KB, 768x768)

777 KB PNG

>>9074084
Another, for your troubles

File: catbox_23zyhq.png (1.73 MB, 1200x1600)

1.73 MB PNG

>>9073955
I mainly use https://civitai.com/models/1281214/oops-all-toons-illustrious-v10?modelVersionId=1990774 as my cartoony/western model. As for style, I just use loras of western comic book artist or tags if the artist is well known in the web.

>>9074113
cheers mate, I'll look into it

>>9074300
if you use the dmd2 speed lora at <50% strength with low CFG and steps, every active lora will be stronger.
but it'll overwhelm the base model outputs somewhat.

File: 1000026927.png (1.31 MB, 832x1216)

1.31 MB PNG

>>9074076
Long story short, on a male exposed thread, the guy asked me to post his pic with his information and all, all his coordinates being censored of course, but because I wrote his discord name, it was considered doxxing so :P I did appeal to it and I got accepted, so here I am back

>>9074477
Damn, that style is wonderful. What is it? I would love to use that!

File: catbox_jnti97.png (1.79 MB, 1280x960)

1.79 MB PNG

>>9074477
>>9074494
Its https://civitai.com/models/2070157/neo-pinup?modelVersionId=2342472 a pinup art style.

File: 1000026912.png (1.08 MB, 832x1216)

1.08 MB PNG

>>9074380

File: 1000026919.png (1.21 MB, 1216x832)

1.21 MB PNG

>>9074750

File: wonder_woman_by_eepol_dkn(...).jpg (71 KB, 768x1216)

71 KB JPG

>>9066217
>>9072482
Think its high time we get 2009 Wondy into her birthday suit

File: ComfyUI_02963_.png (1.01 MB, 1024x1024)

1.01 MB PNG

remade the picture above with >>9074113 model and it's really interesting how the different models have different personalities. this one is hornier than the Illustrious 2.0 I was using to start, but it's better about face/hairstyle consistency but not so great about clothes where I really need to reign it in. it also has a better innate understanding of pipes.

File: ComfyUI_02970_.png (1.02 MB, 1024x1024)

1.02 MB PNG

that said, neither model knows what a corncob pipe is

File: ComfyUI_00693_.png (1.48 MB, 1024x1024)

1.48 MB PNG

>>9069537
>>9069753 (nice gobbo btw)

'Simple background' and 'white background' in negatives helps nudge it away from that concept.
That said, it seems like there's only so much 'attention' to go around in every gen, so asking for a plain background can make the foreground subjects significantly better across the board (more detail, better adherence, less anatomy errors etc)

File: 00212-334025071.png (1.34 MB, 1440x1920)

1.34 MB PNG

>>9075427

File: 00228-2835382271.png (1.74 MB, 1440x1920)

1.74 MB PNG

>>9075428

File: file.png (33 KB, 1224x588)

33 KB PNG

Do you guys know how to use this to make a lot of images?

File: Molly Jo.png (2.23 MB, 2459x4546)

2.23 MB PNG

Requesting topless jeans pinups of Molly Jo (Amphibia).

File: af8c46a2-036e-481c-b75c-e(...).png (1.36 MB, 960x1088)

1.36 MB PNG

or bigger tits

>>9075456
>>9075460
Those are really good. Thanks!

File: 01125-gobbo_goblin_girl,_(...).png (3.15 MB, 1296x1920)

3.15 MB PNG

>>9075080
>there's only so much 'attention' to go around
That's why I upscale and inpaint. The models are capable of a decent amount of detail, but you have to focus it in blocks. Essentially, chunking the image up into 3 or 4 inpainted sections can make a huge difference.

File: 02985-gobbo_goblin_girl,_(...).png (2.7 MB, 1296x1920)

2.7 MB PNG

For instance, there's a night and day difference between the upscaled but still pre-inpainted image here and the final above.

>>9075469
looking at the two the tweaks you made really helped quite a lot. the cigarette, the face details, the weather effects on the floor and signs and her ears, cleaning up the signs.
when you say you upscale I figure your initial generation is a smaller resolution to start so you can generate a bunch of base images quicker so you can sift through to find a decent one to upscale and inpaint. But once inpainted do you keep your final image at that higher resolution or could you then reduce the size back down, or would doing that potentially distort some of the details?

>>9075469
How do you do the inpainting? Do you need to change the prompt, or can you just mask the area you want focused and leave the rest of the prompt the same?

>>9074114
I wonder what the next step is

File: 1752822054586257.jpg (178 KB, 768x1344)

178 KB JPG

>>9075457
Usagi Tsukino?

>>9075855
Are you the guy we sent doodles to

>>9075859
No

It's been a while since i've seen a Melusine or any Spirou Magazine characters.
Does someone still working on the Lora ?

>>9075456
Bigger resolution: https://files.catbox.moe/x1cv40.png

>>9075859
No, he's the guy that constantly spams sailer moon requests with other characters. Pretty sure he's also the wonder woman fag.

File: 01106-gobbo_goblin_girl,_(...).png (3.22 MB, 1296x1920)

3.22 MB PNG

>>9075487
>>9075489
SDXL works at around 1024x1024. You can change the aspect ratio as long as you keep the number of pixels roughly the same. It's not really a matter of quicker generation, though that helps. Things look wonky if you try to gen a 2048x2048 picture, the model's not designed to do it. Once I have an image I like, I upscale it by 1.5x.

I keep the image at the resolution you see, but set inpaint to only do img2img on the selection, not the whole image. Generally my inpainting resolution for a selection is 1280x1536 and I use around 0.4-0.6 denoise. Effectively this means Stable Diffusion is generating at a resolution higher than the selection and shrinking to fit.
Sometimes I can leave the prompt as-is, but if it's zoomed in on a region l have to adjust the prompt a bit. Like if the selection doesn't have a face I have to get rid of most of the facial feature tags or else it sometimes tries to gen little faces somewhere in the background or whatever. There's not much to it once you get used to it.

>>9074810
This would require that Korra be competent at something.

>>9074810
That skill would earn her millions

File: ComfyUI_00199_.png (1.01 MB, 1280x768)

1.01 MB PNG

Getting your kid a rig that can do local gen is gonna be a big filter for future generations. Brainrot isn't gonna be the half of it. 16 year olds smoking pot and slopping all day will vote too. Genuinely fried.

>>9076166
literally no different now. 16 year old kids are smoking pot and playing video games all day. these people don't generally vote, by the way.

File: bon ff.png (3.27 MB, 2048x2432)

3.27 MB PNG

For the Bonfags

Do any of you take requests? I'd like to turn an image that I created using Nano Banana into a more cartoon-like illustration.

>>9076418
more!

>>9075464
You are satisfied now
Do not spam that request again here or elsewhere
Learn how to use the same sites everyone else does from now on even if you're a phoneposter

>>9075861
Please tell us which base model you're using because it looks the same as his threads used and it might be useful

File: bon ff2.png (2.77 MB, 2048x2432)

2.77 MB PNG

>>9076418

File: 00188-334025071.png (1.27 MB, 960x1280)

1.27 MB PNG

>>9076766

File: 00165-334025072.png (883 KB, 960x1280)

883 KB PNG

>>9076767