[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


Janitor acceptance emails will be sent out over the coming weeks. Make sure to check your spam folder!


[Advertise on 4chan]


File: 1758076705142671.jpg (1.16 MB, 2394x1501)
1.16 MB JPG
Previous: >>109294915
NEW: https://huggingface.co/circlestone-labs/Anima
Anima1.0 is out

>UIs
ComfyUI:https://github.com/comfyanonymous/ComfyUI
SwarmUI:https://github.com/mcmonkeyprojects/SwarmUI
KritaAI: https://github.com/Acly/krita-ai-diffusion
Forge Classic/Neo: https://github.com/Haoming02/sd-webui-forge-classic

>How to generate anime:
image : https://rentry.org/comfyui_guide_1girl
video: https://rentry.org/wan22ldgguide

>Style Explorers
Anima: https://anima.mooshieblob.com/ | https://animadex.net/
Illustrious: https://tagexplorer.github.io

>Checkpoints, LoRAs, Upscalers, & Workflows
https://civitai.com
https://civitaiarchive.com
https://tensor.art
https://openmodeldb.info
https://www.seaart.ai
https://www.liblib.art/

>Resources:
Share Metadata: https://litterbox.catbox.moe
Img2Prompt: https://huggingface.co/spaces/fancyfeast/joy-caption-beta-one
Online metadata viewer SD/NovelAI: https://spell.novelai.dev
Catbox/Metadata Userscript: https://gist.github.com/catboxanon/ca46eb79ce55e3216aecab49d5c7a3fb
Output cleanup: https://rentry.org/RemovingDiffusionGunk , https://www.mediafire.com/file/vipr23exc5htmnt (batch processing python script)
Inpainting Guide from an Anon: https://files.catbox.moe/fbzsxb.jpg >>106520607

>/adt/ logo:
https://files.catbox.moe/pakfgt.png
>>
>>109321765
>UIs:
AniStudio (WIP): https://github.com/FizzleDorf/AniStudio

>Maintain Thread Quality:
https://rentry.org/ranfaggot

We hate Catjack here
>>
>>109321805
>Ani
Great, a schizobake
>>
>>
File: 1769330451734286.jpg (441 KB, 1792x2560)
441 KB JPG
>>
File: 00276-3930223392.png (3.1 MB, 1792x2304)
3.1 MB PNG
>>
File: ComfyUI_01091_.png (1.01 MB, 1024x1024)
1.01 MB PNG
>>
File: 1776119407642458.png (1.7 MB, 1024x1024)
1.7 MB PNG
>>
File: ComfyUI_temp_axypj_00003_.png (998 KB, 1024x1024)
998 KB PNG
>>109322653
>>
File: 00277-577394211.png (3.12 MB, 1792x2304)
3.12 MB PNG
>>109323421
the cute!!
>>
File: 00278-3860124210.png (3.1 MB, 1792x2304)
3.1 MB PNG
>>
File: 1784565884676668.png (126 KB, 208x206)
126 KB PNG
>>109322701
>that left hand
One shot, oekaki. You actually used "that tag" for hands in positive? Say the tag's name right now or you're a certified low attention effort slopper for life.
>>
File: ComfyUI_01044_.png (1.25 MB, 1024x1024)
1.25 MB PNG
>>109324384
what tag? also yeah now i see the thumb on the wrong side, happened a lot with this prompt but didnt see that one. i caught the wrong thumb in this image but i guess i should have just inpainted this one instead of keep rolling
>>
>>109323421
cuite!
>>
File: file.png (1.6 MB, 1120x1440)
1.6 MB PNG
>>
>>109324433
https://danbooru.donmai.us/wiki_pages/wrong_hand
https://danbooru.donmai.us/wiki_pages/artistic_error
>>
File: 1girl sitting.jpg (570 KB, 1152x1536)
570 KB JPG
>>
What model should I use if I want amazing prompt adherence/ability to handle complex prompts and situations/multiple figures/etc. but I don't care about artist tags/loras/styles?
>>
>>109326081
anima
>>
>>109326081
novelai
>>
>>109326081
NovelAI is a decent middle ground. If you don't care about styles or NSFW, GPT Image 2 or Nano Banana Pro are also options. Local is kind of a dead end at this point after the Anima flop.
>>
You left the kot out again.
>>
>>109326867
kith her forehead
>>
>>109319377
Sure
>>
>>
File: 1760872244707275.jpg (343 KB, 832x1216)
343 KB JPG
>inpainting out some random red thing
>does this
that's kinda funny how the ass works both ways, the original girl is the one on her back
>>
>>
>>109326686
>Anima flop
wdym flop anima is great
>>
File: ComfyUI_05697_.png (885 KB, 1280x720)
885 KB PNG
>>
>>
>>
File: CUNFY_00622_.jpg (2.51 MB, 1664x2432)
2.51 MB JPG
>>
File: ComfyUI_05543_.png (898 KB, 1024x1024)
898 KB PNG
>>
>>
>want to slop up a fan comic
>artstyle drift over multiple images, doesn't look consistent

This shit is harder than it looks. Is IPAdapter the best way to keep a character looking the same over multiple gens or is it old tech?
>>
File: novelaireferenceimage.png (1.15 MB, 832x1216)
1.15 MB PNG
>>
How do you feel about nochekaiser?
>>
>>109327999
no
>>
>>109327999
che
>>
>>109327999
i think he just got 'hide content from this user'ed
>>
>>109327087
dozo
>>
File: gtaH.png (1.24 MB, 2000x2447)
1.24 MB PNG
https://www.youtube.com/watch?v=zA0QsWj-mUE
>>
File: h1.png (1.86 MB, 2600x2600)
1.86 MB PNG
>>
File: h2.png (1.02 MB, 2600x2600)
1.02 MB PNG
>>
File: h3.png (1.08 MB, 2600x2600)
1.08 MB PNG
>>
File: h5.png (1.43 MB, 2600x2600)
1.43 MB PNG
>>
File: h6.png (1.24 MB, 1734x2000)
1.24 MB PNG
>>
holy fucking slopfest
>>
File: bun.jpg (3.13 MB, 2752x1536)
3.13 MB JPG
>>
File: ComfyUI_05623_.png (1.69 MB, 1920x1080)
1.69 MB PNG
>>
>>109329072
prompt for godzilla beach scene?
>>
>>109329129
Anime movie, 1993. Scene in action. Widescreen, black bars. letterboxing, 16:9. Sunny, clouds, blue sky, city. Japan. Young adult Sailor Moon in bikini. Tanning. Ocean beach. Sleeping. Lying on sunlounger. Godzilla far in distance. Blonde hair.
>>
File: 1769770171992672.jpg (963 KB, 1920x1088)
963 KB JPG
>>109329209
nice
>>
File: ComfyUI_04530_.png (1.51 MB, 1200x1200)
1.51 MB PNG
Hello, /aicg/ here. I'd like to generate images of my OCs with slightly better resolution than usual this time. Can I get an Anima preset with built-in upscaling?
I barely know what I'm doing with this tool.
>>
File: ad.png (738 KB, 1280x720)
738 KB PNG
had to try multiple times to get the text right
>>
File: 00050-2904499034.png (2.62 MB, 1280x1856)
2.62 MB PNG
>>
File: ComfyUI_05804_.png (1.31 MB, 1280x1280)
1.31 MB PNG
>>
>>109329288
kawaii
>>
>>109329268
/g/ldg/ more fast and technical reply
>>
>>109329268
Are you already using anima? Just up the base res.
>>
>>109329536
Isn't that bad?

>>109329520
I will always put my faith behind dweebs who like anime.
>>
>>109329561
>will always put my faith behind dweebs who like anime
I thought the same way once. But eventually you will walk and see things with your own eyes
>>
File: 1784088247467169.jpg (352 KB, 1248x1824)
352 KB JPG
>>109329561
>Isn't that bad?
NTA but anima does up to 1536x base res pretty well
>>
>>109329288
myon myon :3
>>
>>109327668
yes
>>
>>109329268 >>109329561 Notice how they refuse to help you, but they will troon you with irrelevant off topic comments >>109329870 >>109330119. They were never born to be a great general.
>>
>>109330272
>>109329268
Ok but share the wip already
>>
File: nbp-the.jpg (291 KB, 1600x1195)
291 KB JPG
>>
>gptslop
>>
File: ComfyUI_05469_.png (1.53 MB, 1920x1080)
1.53 MB PNG
>>
he mad
>>
File: 1781421256117065.jpg (190 KB, 1024x1024)
190 KB JPG
>>
File: hostage.png (1.08 MB, 1280x720)
1.08 MB PNG
>>
>>109330466
Wip deilvery sirs
(WIP): https://github.com/FizzleDorf/AniStudio
>>
>>109331311
Oh cmon try at least
>>
File: pinecat bed.png (2.65 MB, 1856x1536)
2.65 MB PNG
i don't upscale my anima gens
>>
>>109331792
Now I do it using SD1.5 base resolution with Hiresfix 2x
>>
how come krea is better than anima?
>>
>>109332530
krea has a better text encoder and the dit model is much bigger. It's also trained on actual anime and not temu fanart
>>
>>109332530
where do you see that?
>>
>>109331792
Anima can't hiresfix without altering the composition
>>
I miss Illustrious tho. If Anima never happened we'd have a Chenkin 0.5 IP-Adapter by now (he said so himself, but he dropped the project the moment Anima came out)
>>
>>109327999
Better than him https://civitai.red/user/fluxis/models
->slop tier style
->the same recycled anime characters everyone already has a million loras of
->somehow the #1 new creator
>>
>>109332645
Anima can't outrank the whole illustrious ecosystem that's grown up around it. And it won't be able to outrank the upcoming local models that already come with known characters baked in like Krea 2.
Anima is the middle child, the least loved child of local diffusion.
>>
>>109332645
im still genning my porn with ill, i love it
>>
>>109333016
anima is the best local model available for anime in my fact tbqh
>>
File: Shigure Frisbee.jpg (502 KB, 1536x1536)
502 KB JPG
>>
>>109329298
thats really cool
>>
The downside of Illustrious/NoobAI/Anima is that it's being gimped by outdated dataset
>>
>>109333098
Give it some time, Krea 2. People are already baking anime styles and character for it, so maybe we'll end up using LoRAs instead of artist tags since nobody's going to finetune that monster. But if it ends up supporting regional prompting with LoRAs, LoRA scheduling, and people start making ControlNets for it, it could turn out to be a great model.
>>
>>109333134
if krea2 gets a finetune ill switch over but using character loras are a pain, especially when you want multiple characters in an image. plus loras take up space which starts to add up
>>
>>109333098
Not even remotely true lol
>>
Have there been any Illustrious/noob shitmixes released since anima with updated datasets that are any good?
>>
>>109333218
>updated datasets
just use a lora?
>>
>>109333218
MUGEN, using the flux2 vae
>>
>>109333218
why use iluslop
>>
Nobody helped him yet? >>109329268 , why do you have a general if you wont help anyone?
I wont help him tho, i'm just a lurker
>>
why does this schizo faggot exist?
>>
illustrious is all you need btw
>>
>>109331792
more importantly, who turned pine into a cat?
>>
Weird tiddy skindentation thing from Krea Turbo but I can't find irl photos for comparison of BIG tiddies with off-shoulder neckline any lower than the armpit, which I suppose would be too impractical, though I assume the "impractical" nature would be alleviated if straps were part of the top. But such a strapless top existed would they use extra pressure leading to skindentation? "Just use tape"?

Models are now on OpenRouter but isn't taking images to use them as style transfers, so it's still pretty much txt2img (i.e. useless) on OR.
>>
File: Anima_01557_.jpg (404 KB, 1536x1056)
404 KB JPG
>>
>>109326686
>Anima flop
What's wrong with it?
Why is local dead?
>>
File: Untitled(1).jpg (2.43 MB, 4645x4106)
2.43 MB JPG
>>109334451
Tell me, has anything improved? These Anima shitmerges, this Anima base model everyone worships has it given you anything real? Because from where I stand, I see nothing, not one shred of progress, go on, prove me wrong, I dare you faggot.

You want my proof? Fine, here is an /adt/ collage from almost a year ago, sitting woth today's collage, look at them, tell me what separates them, because I cannot find it. I see no difference between Anima and Illustrious. I see no improvement. I see nothing but a new model with no ControlNet and less Lora versatiliy and no CLIP.
>>
File: comic painyan 2.jpg (3.9 MB, 2779x4096)
3.9 MB JPG
>>109333581
it's been a thing for years. the artist liked it so they made it canon in 2024
>>
>>109329268
Sampling Method ER SDE
Schedule Type Beta
32 steps
Shift 3
CFG 4
>upscaling
Enable hi-res fix
Upscaler ESRGAN
Denoising str 0.4 or w/e you want
Otherwise same settings as above
>resolution and upscale factor
Whatever you want, I have been doing 1024x1024 or equivalent base gen and 1.5x upscale.
>>
>>109334887
forgot to say 10 steps for hi-res
>>
>>109334901
>>109334887
When you say “1024×1024 or equivalent base gen,” does “equivalent” mean the same total pixel count for non-square images?
>>
>>109334887
If I change the base resolution, do I need to change the 32 steps too?
>>
What compels fags to log onto 4chan every day and argue about AI generated anime girls instead of just generating AI anime girls?
>>
>>109333134
Isn't Krea2 too beefy for midrange computers?
>>
>>109334887
And Beta schedule vs Karras, is there a real difference?
>>
>>109335109
we are more philosophers than genners
>>
>>109334901
So just to make sure I’m understanding this correctly, when you say “otherwise same settings as above” for the hi-res fix, do you mean the same 32 steps, CFG 4, Shift 3, and ER SDE sampler settings, or are the settings supposed to be changed specifically for the upscaling pass? I feel like this is probably obvious, but I’d rather ask before accidentally doing it.
>>
>>109335037
>When you say “1024×1024 or equivalent base gen,” does “equivalent” mean the same total pixel count for non-square images?
correct
>>109335077
no
>>109335113
I haven't tried karras, don't know if it is recommended. ER SDE Beta is the standard/default for Anima I believe. The Anima base model page goes over a few options https://civitai.red/models/2458426/anima?modelVersionId=3126581 ctrl+f samplers
>>
>>109335151
Yes that's what I meant except for hi res steps should be 10.
Unless, you know, you want more. That's all a lot of this really is anon. You can do just do things.
>>
SFW Geminislop I made in Google Flow, hence no watermark on the bottom right
>>
>>109334767
It's not the models, it's the people. What difference does it make when people just gen the same slop over and over again on different models?
>>
>>109335238
AI general, use em dash
>>
>>109334767
>I see no difference between Anima and Illustrious.
if your only use case for anime gens is looking at them thumbnail-sized while squinting I have SD1.5 to sell you
>>
File: 635653.png (29 KB, 1200x237)
29 KB PNG
How much stardust have you contributed today?
>>
>>
>>
File: ComfyUI_temp_jxyno_00154_.png (1.6 MB, 1024x1536)
1.6 MB PNG
>>
File: 1779976278697820.jpg (2 MB, 4200x2400)
2 MB JPG
>>109332625
Yes it can, if you mean img2img. It's weird though sometimes I can't get it working.
>>
>>109335670
I didn't know there were shoe cults among vampires.
>>
File: 1760148299690820.png (2.28 MB, 1024x1024)
2.28 MB PNG
>>
File: 00283-810482899.png (3.41 MB, 1792x2304)
3.41 MB PNG
>>
>>
A Krea2 rank 128 style lora trained on HD stills from the anime Nichijou:
https://huggingface.co/quarterturn/krea2_nichijou

Checkpoints from 10000 to 20000 are provided. I stopped at 20000. I think 18500 is about right. YMMV. Enjoy.
>>
>>109335983
catbox?
>>
File: ComfyUI_01755_.png (1.26 MB, 1024x1024)
1.26 MB PNG
>>
>>109336230
Gay
>>
>>109336253
its female
>>
>>
>>109334767
Ignoring the size differences the other anon mentioned, I have not tried anima enough yet but is the natural language not an improvement?
Or is it too random/untrustworthy to be considered one?
>>
>>109336278
Me? Yes. Also have cute feet.
>>
File: ComfyUI_temp_jxyno_00040_.png (1.73 MB, 1536x1024)
1.73 MB PNG
>>109336352
Speaking as a fan of Illu who wasn't overly excited about Anima, I'm finding some pretty fun uses for it. Like, forget about asking Illu to give you a pixelated old school JRPG dialogue box, let alone the text. Not that a lot of those ideas couldn't have been achieved better by feeding the image into CGPT2, but it's something
>>
>>109336352
so far the best or most reliable way ive seen to use NL in anima is to prompt for direction. like X on the right, or Y in the center background etc. also I think its best to basically use short comma-separated phrases instead of full on sentences and paragraphs. but tags should still be your default option assuming what you want does have a tag
>>
>>109336141
Oh nice, this one looks more accurate to the style than id4, thank you for sharing



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.