Thread #9170354
File: csdg.jpg (1.4 MB)
1.4 MB JPG
Previously on /csdg/:
>>9154460
https://desuarchive.org/aco/thread/9154460
>Pick one and install Stable Diffusion
https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#installin g
(new Forge) https://github.com/Haoming02/sd-webui-forge-classic/tree/neo
(deprecated) https://github.com/lllyasviel/stable-diffusion-webui-forge?tab=readme- ov-file#installing-forge
ComfyUI image/video model guides
https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#examples
>Recommended XL models
NoobAI: https://civitai.com/models/833294/noobai-xl-nai-xl
DHXL: https://civitai.com/models/1291404/dhxlv10
Illustrious: https://civitai.com/models/795765/illustrious-xl
ConfettiTuneXL: https://civitai.com/models/184813/animeconfettitunexl
-
DMD2 speed LoRA: https://civitai.com/models/1608870/dmd2-speed-lora-sdxl-pony-illustrio us
>Uncensored Flux-based model
Chroma: https://civitai.com/models/1330309/chroma | https://civitai.com/models/1807633/ chroma-unlocked-q4ks (low VRAM)
>New models to try out
Z Image: https://civitai.com/models/2342797/z-image-base
Flux.2 Klein: https://civitai.com/models/2322332/flux2-klein
Anima: https://civitai.com/models/2458426/anima-official
>AI image upscaler model database
https://github.com/OpenModelDB/open-model-database
>LoRA metadata viewer in HTML form
https://github.com/Xypher7/lora-metadata-viewer
>Optional extensions
https://github.com/DominikDoom/a1111-sd-webui-tagcomplete
https://github.com/hako-mikan/sd-webui-regional-prompter
For ComfyUI:
https://github.com/ltdrdata/ComfyUI-Manager
>IS MY GPU GOOD ENOUGH?
https://www.tomshardware.com/pc-components/gpus/stable-diffusion-bench marks
>3DPD, Furry, Futa
>>>/aco/asdg
>>>/trash/sdg
>>>/b/AI
>Related threads
>>>/g/sdg
>>>/h/hdg
>>>/e/edg
>>>/d/ddg
233 RepliesView Thread
>>
File: ComfyUI_temp_kyafe_00001_.png (1.6 MB)
1.6 MB PNG
>>9165584
Finally managed to make it work with the portable version, and it's actually fast. do you have any guide on how to add controlnet?
>>
File: 02896-3351308740.png (1.9 MB)
1.9 MB PNG
>>
File: 02848-3808222829.png (1.3 MB)
1.3 MB PNG
>>9170616
>>
>>
File: HiResGwynGoldEyeshdow_c_00013_.png (2.3 MB)
2.3 MB PNG
>>
File: HiResGwynGoldEyeshdow_c_00008_.png (2.3 MB)
2.3 MB PNG
>>
File: HiResILLPandaCreamed_00001_.png (2.6 MB)
2.6 MB PNG
>>
File: 1773954934774854.png (1.4 MB)
1.4 MB PNG
I am once again requesting Apricot from Apricot Cookies in her magical girl uniform but with a thick mommy body
>>
>>
File: ComfyUI_04215_.jpg (1010.2 KB)
1010.2 KB JPG
>>
File: 233247_00001.mp4 (750.8 KB)
750.8 KB MP4
>>9170622
>>
File: 234519_00001.mp4 (1.1 MB)
1.1 MB MP4
>>9170616
>>
File: 001709_00001.mp4 (1.3 MB)
1.3 MB MP4
>>9170683
>>
File: Mrs. Boonchuy.png (3.3 MB)
3.3 MB PNG
Requesting sling bikini be changed into panties.
>>
File: IMG_8322.png (578.7 KB)
578.7 KB PNG
I was gonna ask animatoranon if he could try this one
it doesn’t have shadows under her breasts, so i think it’ll work better
>>
File: IMG_0459.gif (3.9 MB)
3.9 MB GIF
>>
>>
>>
File: ComfyUI_25708_.png (895.7 KB)
895.7 KB PNG
>>
>>
>>
File: ComfyUI_25724_.png (1.1 MB)
1.1 MB PNG
>>9171219
There are no models (that I know of) that are great for comics, at the moment. The new Anima model can be coerced into making comic book like art if you use the right artist names and style prompts, but I'm still trying to figure out how far that can go.
>>
File: ComfyUI_25739_.png (1.3 MB)
1.3 MB PNG
>>9171219
Yeah, it knows at least dan mora, jhony caballero, dima ivanov and stanley lau that have a bias towards comic books. There's also the tag western comics \(style\) that you can use. There's a lot of room to improve here, but I'd say this is your best bet atm.
>>
File: ComfyUI_25756_.png (1.1 MB)
1.1 MB PNG
>>
File: ComfyUI_25762_.png (1.7 MB)
1.7 MB PNG
>>
>>
File: 112212_00001.mp4 (1.4 MB)
1.4 MB MP4
>>
File: 130153_00001.mp4 (1.3 MB)
1.3 MB MP4
>>9171175
>>
>>9170599
I takes around 40 secs or less, not counting the model loading time. the only gripe is that the model isn't very flexible or good with loras, but alas that's the price of not having to buy a new pc.
Controlnet is mainly for posing the character
>>
>>
>>
File: ComfyUI_04219_.jpg (531.1 KB)
531.1 KB JPG
>>9171646
>>9171742
Don't know how to write prompts?
Head to the one place where everyone copies.
>https://danbooru.donmai.us/
click on image you think you like.
(fuck it whatever) >https://danbooru.donmai.us/posts/11127772
click on bottom left where it says "History" click [TAGS]
copy that.
>noyemik original blonde_hair blue_eyes hair_tie magazine_(object) navel_focus outie_navel slingshot_swimsuit swimsuit tall_female text_background tongue tongue_out
paste in your prompt in the workflow that you gened your thumbs up girl.
see what you get.
Freeze seed (where it says "randomize/trash icon ######")
modify prompt to saying something different.
>Ginger_hair red_eyes navel_focus outie_navel slingshot_swimsuit swimsuit tall_female white_background HUGE_BREASTS
gen image.
Tada!
once you're done with it, go back to copy/pasta site and get new set of tags until you only go there for a reference on artist or something obscure like "split"
Note: the crap with [_] is not needed if you're using [,] as your word seperator.
Eventually you'll know what model can do and if it needs seperator at all.
>>
>>
File: Kara2.png (1.3 MB)
1.3 MB PNG
>>9171886
And here's one trained on the same dataset for Preview 3 (not 2), but with a larger capacity, trained at 64 dim instead of 32. I actually prefer the smaller one; it's more idiosyncratic. This one is more saturated and more stylised, mainstream comics style. Less painterly. So you can see, you don't need much size or many training steps to capture a style
>>
File: ComfyUI_25980_.png (1.3 MB)
1.3 MB PNG
>>9171893
Nice. Good to know this is easy. What trainer is the best for anima at the moment? This is just @dan mora as the artist:
>>
File: 10998-959900980.png (1.4 MB)
1.4 MB PNG
>>9171219
Mora than models, I just use Lora if I'm looking for someone in specific. Here is one I'm using Adam Hughes.
>>
File: ComfyUI_04226_.jpg (963.1 KB)
963.1 KB JPG
>>9171219
Jim Lee
Ivan Reis
Gary Frank
Alex Ross
Jason Fabok
Joe Quesada
David Finch
Steve McNiven
Olivier Coipel
Leinil Francis Yu
Karl Kofoed
Simon Bisley
Alfonso Azpiri
artists will get you that style with anima or you can look up loras for them.
>>
File: ComfyUI_04227_.jpg (921.1 KB)
921.1 KB JPG
>>9172095
@Olivier Coipel
>>
File: 02976-1113456494.png (2.3 MB)
2.3 MB PNG
>>9172061
Frank Cho
>>
File: f0e39619-ddcf-45e8-ae38-c7bfc1ea6c48.jpg (158.7 KB)
158.7 KB JPG
Anyone order a Batgimp Wendy Corduroy? Well I made her anyways
>>
>>
>>
>>
>>
File: ComfyUI_26322_.png (1.9 MB)
1.9 MB PNG
>>
>>
>>
>>
>>
>>
File: 03012-735541055.png (1.6 MB)
1.6 MB PNG
>>9172138
>>9172487
>>
File: 03013-2801776713.png (1.3 MB)
1.3 MB PNG
>>9173032
>>
File: HiResGwynXRi_00000_r_.png (2.7 MB)
2.7 MB PNG
>>
>>
File: 65b0574d-5140-4b31-afca-387569510374.png (1.8 MB)
1.8 MB PNG
>>
File: ComfyUI_04234_.jpg (593 KB)
593 KB JPG
>>
File: 03021-3844692137.png (1.2 MB)
1.2 MB PNG
>>9173035
>>9173192
>>
File: 02899-3704209051.png (1.7 MB)
1.7 MB PNG
>>9170622
>>9171131
Looking good, like the skirt is animated.
>>9171135
Came out really cute.
>>
>>
>>
>>
>>
>>
>>9173373
I've had some very good results with certain tags but mostly related to outfits and sex acts. Backgrounds and more complex or obscure fetish stuff are a pain in the ass to make work.
I was wondering if there's a lamen term for underboob. Despite not being trained on danbooru tags, BL recognizes things like cleavage, piercings, blowjob, lactation, etc.. so I thought I might have just been using the wrong term for it.
Thanks for your help! I'll keep experimenting.
>>
File: 130108_00001.mp4 (887.2 KB)
887.2 KB MP4
>>9173282
>>
File: 121335_00001.mp4 (787 KB)
787 KB MP4
>>9173328
>>
>>
File: Wan2.2_image_to_video_00003_.mp4 (227.2 KB)
227.2 KB MP4
>>9173439
I really need a link to a tutorial or a workflow for video. I got myself started but comfy feels very overwhelming compared to forge and I dont know where to go from here
>>
File: 03074-1341679371.png (1.2 MB)
1.2 MB PNG
>>9173328
>>
>>
File: f89c4905bf867cd62d87e3da2597c9ad.png (967.2 KB)
967.2 KB PNG
>>9173487
Requesting an animation of this pic, if you would be so kind.
>>
>>
File: 03067-575059947.png (1.1 MB)
1.1 MB PNG
>>9173530
>>9173532
Suzie Chan, from Hanna-Barbera The Amazing Chan and the Chan Clan.
>>
>>
>>
File: ComfyUI_26371_.png (1.9 MB)
1.9 MB PNG
>>
>>
>>
>>
>>
File: 03109-3867920283.png (1.1 MB)
1.1 MB PNG
>>9173552
>>9173715
YW.
>>
File: 2026fifi.png (873.3 KB)
873.3 KB PNG
Any general advice? Just figuring out regional prompts
>>
lads, I haven't genned anything for a long time now. I'm trying to figure out what the easiest way to get 2 characters into a video of them fucking. I figure I get some images of both of them and use some image edit (qwen?) model to frame the scene and then wan to animate it?
or do I still have to go make loras to have a chance at making this work?
>>
>>
>>9173820
sick, thanks. I didn't want to waste a few hours setting all that up if it wasn't going to work. I'll be honest the thing that got me back was because I was playing a multiplayer video game that I will not name and I could not help but lusting after another player's avatar so I will have to make some shit here. will report back on success or failure
>>
File: ComfyUI_04239_.jpg (978.2 KB)
978.2 KB JPG
>>
File: animaOfficial_preview3.jpg (601.3 KB)
601.3 KB JPG
>>9173754
Anima is nearly 3x slower than FP8 XL on my GPU.
>>
>>
File: 014942_00002.mp4 (1.4 MB)
1.4 MB MP4
>>9173485
>Nice and smooth animation. Are you using Comfy?
Yeah. Comfy, mixing up checkpoints now and then, currently running with version 1 of smoothmix as my high noise model and the 'synthseduction' dasiwa wan as my low.
>>9173487
It's definitely a bit to get accustomed to. You seem to have at least the basics there, but you'll probably want some frame interpolation going on, 16 fps is rough. And maybe a different checkpoint, if that's the regular wan 2.2, some of the merged ones are better for dirty stuff.
I put up a 1 day litterbox of the same video I'm attaching here if you want to try using the workflow embedded. Though it has a good few custom nodes in it, and I'm sure it'd look a bit scary too, it's a substantially modified version of the one recommended by the creator of smoothmix.
>>9173537
>>
>>
File: HiResMajElD_00013_.png (2.5 MB)
2.5 MB PNG
>>
File: HiResMajElD_00018_.png (2.7 MB)
2.7 MB PNG
>>
File: HiResMajElD_00029_.png (2.4 MB)
2.4 MB PNG
>>
File: ComfyUI_04241_.jpg (806.2 KB)
806.2 KB JPG
>>
File: ComfyUI_04243_.jpg (672.1 KB)
672.1 KB JPG
>>
>>
File: ComfyUI_26418_.png (1.3 MB)
1.3 MB PNG
>>
>>
File: Anima_01718_.png (1.7 MB)
1.7 MB PNG
>>
File: Anima_01729_.png (2.2 MB)
2.2 MB PNG
>>
>>
File: Anima_01732_.png (1.5 MB)
1.5 MB PNG
The juxtaposition between comically large booba and shitty horror sloppa looks funny to me.
>>
File: Anima_01734_.png (1.3 MB)
1.3 MB PNG
>>
>>
>>
>>
File: Anima_01746_.png (2.1 MB)
2.1 MB PNG
Perky bewbs
>>
>>
File: Anima_01750_.png (1.5 MB)
1.5 MB PNG
>>
File: Anima_01758_.png (1.6 MB)
1.6 MB PNG
>>
File: d3fcad215f3e018ca2378b7b0bba8c1a.jpg (97.1 KB)
97.1 KB JPG
>>9171135
I've seen some really good jacket images floating around. AI generated but really accurate stylistically. I'm well versed in generating images regularly using models but I have absolutely no idea how this (the base image, not the animation) is being pulled off. Any ideas?
picrel. It even got the chest tube things.
>>
File: 03119-2888420387.png (948.9 KB)
948.9 KB PNG
>>9173803
>>9174062
Catbox is being nasty, so here is a litterbox of that image. https://litter.catbox.moe/3bhb3ydkvk4ufa03.png
>>
File: Anima_01763_.png (1.9 MB)
1.9 MB PNG
>>
File: mynigger.jpg (41.3 KB)
41.3 KB JPG
>>9174068
Jacketfag? I feel you. We are kindred spirits.
>>
File: 1471819820599.jpg (18.5 KB)
18.5 KB JPG
>>9173803
Tell me how this is done
>>
File: 591325049663014.png (1.1 MB)
1.1 MB PNG
>>
>>
File: 581690065337800.png (1.2 MB)
1.2 MB PNG
>>
>>
File: 848585397641947.png (1.1 MB)
1.1 MB PNG
>>
File: progress.png (1.9 MB)
1.9 MB PNG
>>9174120
I've been learning krita and its ai plugins. Been able to make this so far.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
File: 03127-3623907572.png (631.8 KB)
631.8 KB PNG
>>9174157
Ok, last one of her and I'll take a break.
>>
File: 1775928889900971.jpg (140.7 KB)
140.7 KB JPG
>>9173882
This is super cool! You think you could try your hand at animating this pic?
>>
>>
File: 233423_00001.mp4 (3.8 MB)
3.8 MB MP4
>>9174044
Was just a bit curious what it would do with this. Eyes are one thing that it seems like almost always get very heavily 'normalized' compared to reference images; it can be difficult to even keep cartoonish black-dot pupils instead of having full irises sprout out immediately.
Still, came out kind of neat overall. I think the shifting psychedelic colors were hard on encoding, though; it came out of the box at 6.5 MB where a quarter of that is more common, had to re-encode at a lower quality.
>>
File: 001030_00001.mp4 (718.3 KB)
718.3 KB MP4
>>9174071
Whoops, posted the wrong one. Though this should soon be the only one remaining.
>>
You can tell wan you want a particular style - cartoon, psychedelic, 8-bit graphics - and it can adhere a bit to that, but it seems like new elements are often much more realistic than existing ones, despite such instruction. See the hands of the girl in the first video, and of the man in the second, much closer to simple human realism.
>>
>>
>>
>>9174579
>>9173882
>currently running with version 1 of smoothmix as my high noise model and the 'synthseduction' dasiwa wan as my low.
Most of the time I'm also using a fairly simple lora I got from civitai which was just called 'naughty horny woman,' which makes for various seductive motions and behaviors, at varying strengths. Other ones where specific acts are called for. Unfortunately, that one was taken down at some point.
>>
File: ComfyUI_26524_.png (1.3 MB)
1.3 MB PNG
>>
File: 004807_00001.mp4 (941.3 KB)
941.3 KB MP4
One thing at whose seeming impossibility I am still surprised is having hips sway slowly from side to side. It seems like it ought to be so straightforward, but no amount of prompting will get it in most cases,
>>
File: 010441_00001.mp4 (1.6 MB)
1.6 MB MP4
>>9174587
>>
>>9174551
>Eyes are one thing that it seems like almost always get very heavily 'normalized' compared to reference images; it can be difficult to even keep cartoonish black-dot pupils instead of having full irises sprout out immediately.
Possibly caused by whatever step-distill lora you are using? Maybe other loras did not over-learn that.
>I think the shifting psychedelic colors were hard on encoding, though; it came out of the box at 6.5 MB where a quarter of that is more common, had to re-encode at a lower quality.
Yes, more shit going on the image = higher bitrate needed to hit the same crf target = larger file size
>>
File: IMG_4601.jpg (120.9 KB)
120.9 KB JPG
>>9174596
Oh! We can animate convincing cumshots now?
Someone try this one! maybe it’ll turn out good (then again maybe not)
>>
>>
File: 653987569329442.jpg (541.7 KB)
541.7 KB JPG
>>
File: 653987569329441.png (3.8 MB)
3.8 MB PNG
>>
>>
>>
File: Sharon McGee.png (3.8 MB)
3.8 MB PNG
Requesting an animation for this pic.
>>
File: imagem_2026-04-13_163751064.png (892.9 KB)
892.9 KB PNG
>>
File: 1775404872311448.png (2.3 MB)
2.3 MB PNG
>>9174295
requesting this one animated if possible
>>
>>
>>
>>
Can anyone recommend me some good genners to follow? For example Im following
https://x.com/StuffyAi
https://x.com/ShexyoArt
https://x.com/EroticNansensu
and want more high effort shit
>>
File: 151123_00002.mp4 (3.2 MB)
3.2 MB MP4
>>9174849
Gotta be careful what you go diving with.
>>
>>
>>
>>
File: IMG_0558.jpg (104.8 KB)
104.8 KB JPG
>>9174752
This isn’t the exact same image, but it’s close enough, if I manage to generate a better sample, I’ll send it over too
>>
File: IMG_0564.jpg (110.1 KB)
110.1 KB JPG
>>9175027
WAIT I GOT A BETTER RESULT
>>
File: miku 10.png (1.3 MB)
1.3 MB PNG
>>
File: miku 11.png (546.4 KB)
546.4 KB PNG
>>
File: miku 12.png (839.6 KB)
839.6 KB PNG
This one came out more normal than intended.
Enjoy I guess.
>>
>>
>>
File: 03190-4026848847.png (1.2 MB)
1.2 MB PNG
>>
File: 1750257521600642.png (3.3 MB)
3.3 MB PNG
>>9174157
nta who had asked, but thank you for the catboxes
I'm also a complete newfag and having a simple workflow for toons to start with has been a true blessing, helps in actually getting the basics done
>>
>>
>>
File: 1771752851975290.png (1.1 MB)
1.1 MB PNG
>>9175784
I'm ashamed to admit how many attempts this took, all I know is that I ended up with half a gig worth of outputs, but at least I kinda understand how inpainting works (altho I couldn't fix all the jank unfortunately)
>>
File: 4152026_157750018822244012_00032.mp4 (663.9 KB)
663.9 KB MP4
Great difficulty in removing eye highlights, none of the booru tags seem to work but tagging both eye and pupil with the same color makes it look like one solid color.
>>
File: 4152026_157750018822244012_00031.mp4 (815.2 KB)
815.2 KB MP4
>>
File: 157750018822244012_00015.mp4 (937.1 KB)
937.1 KB MP4
I did learn a lot from these cartoon gens though, I'd mostly stuck to anime/realistic or 3d stuff until this point. Anime styles/tags seem to preserve the look of cartoons better than using cartoon or realistic tags. Cum WAN models have a bad side effect of reintroducing pupils where they aren't wanted, even if you overload with explicit color tags:1.2+...
>>
File: 004642_00002.mp4 (808.1 KB)
808.1 KB MP4
Let's see if this works, 4chan has been totally uncooperative for me for the last day or so. Interesting that other people seem to be able to post, though.
>>
Hey, it actually went through this time, nice.
>>9175937
>none of the booru tags seem to work
Yeah, I assume you're also using wan, but it's not trained on booru tags at all, you want basically natural-language descriptions + any special keywords or phrases that your loras may have been trained upon.
>>
File: wanjizz.png (50.4 KB)
50.4 KB PNG
I find that turning off facesplash hi/lo together, or thiccum hi/lo separately improves the pupil issue. Not on all cartoon gens, some might work fine with all of those turned on but you have to experiment with it by turning off both or one side only. maleejac/handjob can be left turned on, but without maleejac you might get jizz with the wrong color, unless whatever you use as the service lora includes proper cum weights.
>>
File: 157732750245007188_00003.mp4 (749.9 KB)
749.9 KB MP4
And unless the video you want to add cumshots to is realistic the weights of both loras on high/low are rarely equal, leaving them both as 1, .5, or whatever for testing made the cum slide off the character too fast, I kept pushing the numbers around until I got something kinda gluey for the JLU video. With facesplash the cum might act more like a rubber band but it won't adhere where it splatters if both are equal. The numbers I have in the video I brought over from stuff I did for 3d characters, so everything has been lowered to avoid semen covered faces from morphing faces into realistic people. For 3d characters it's better to use first/last frame than allowing the model to add cum, I recommend using first frame without cum to put the characters face in a pose where it would be covered in jizz, mouth open tongue out, whatever, if first frame is simply them smiling and presenting their face to a penis. Then run the last frame through something that will paint the jizz on carefully, you don't need that much either, just some runny drops on the face and it will do a great job at preserving the face/head if its something really unique otherwise you'll get disney looking garbage if you only use WAN without any nodes to do pseudo face restoration or restrict how the cum affects whats underneath it.
Also, thanks as always to 1staidbag for all the comfyui stuff.
>>
>>
>>
>>
>>
>>
Recommend me a base model which has a more western style with less tendencies to use anime style, mostly because I need something a bit more realistic in proportions for character sheets but most of my models are either too rigid (very generic SD looking) or just outright super cartoony.
>>
>>
>>9175996
Cum/Facial lora: https://civitai.red/models/1962545?modelVersionId=2221382
That's on top of the checkpoints I mention in >>9173882
I'm using the original 1.0 version of the lora; I guess there's two newer ones now but I haven't tried them.
>>
File: 150049_00002.mp4 (841.4 KB)
841.4 KB MP4
>>9175854
Annoying little problems, it kept thinking that the exposed teeth in her initial, lip-biting expression were like a drop of saliva that would immediately drip down out of her mouth. And even without being told, absolutely every attempt resulted in her lifting her dress back up more or less into position.
>>
>>
File: 161830_00001.mp4 (1021.7 KB)
1021.7 KB MP4
>>9174772
It wasn't a very interesting starting point, but I realized that she's related to Rapunzel.
>>
File: 1774689943231065.png (1.4 MB)
1.4 MB PNG
>>
>>
>>
File: 201529_00001.mp4 (1.3 MB)
1.3 MB MP4
>>
File: 204919_00001.mp4 (1.5 MB)
1.5 MB MP4
>>
>>
>>
>>9176259
Yeah, but sometimes undressing goes for the fastest way, which results in stuff being taken off like skin peeling instead of unbuttoning, or even sliding it from top to bottom even tho it would be better if it was the other way around.
>>
File: coque5.png (1.1 MB)
1.1 MB PNG
I asked her if a white boy could get a little motion and she used telekinesis to turn my prostate into a black hole
>>
>>9176176
>How do these models handle undressing?
Poorly, as a rule. Maybe there's a lora for it, but I haven't seen one. Fundamental problem there is that 5/6 seconds isn't really long enough for a lot of realistic undressing, anyway.
>>
>>
>>
>>
>>
File: file.png (169.3 KB)
169.3 KB PNG
>>9176286
It is basically a manager which can also run local AI backends with webui and stuff, you can basically set up symlinks to use it as hub for all your models, loras and stuff, there is a browser extension which allows you to download from civitai.
>>
>>
File: file.png (523.8 KB)
523.8 KB PNG
>>9176289
Pretty straightforward honestly, you may still run into some issues with some workflows refusing to install some nodes for some reason but SM uses environments anyway so it wouldn't have to affect your whole system python setup, its inference tab is also a way to generate images in-app, however video generation never worked for me, sometimes it also gives you blank images for some reasons, probably its own workflow, for me at least it makes it easier to have model info and metadata when you can't remember trigger words and stuff, it pulls from civitai mostly.
>>
File: 03241-330906805.png (1.3 MB)
1.3 MB PNG
>>9176261
Ohhh, never mind, I'm just tired and being retarded. Didn't notice you were talking about animation.
The only way I can figure it working it would be doing like key frames. Making small animation with a first frame gen and an ending frame gen. Then continue as you want with another animation using the last gen as first and a new one for the ending and finally splicing all the animation together as a big one.
>>
>>
File: coque6.png (1.1 MB)
1.1 MB PNG
>>9176273
got distracted, took me a second
>>
A great tragedy is that the artist devilhs is well-recognized and his style even captured by the illustrious models, but it inevitably brings in the baggage of the simple "devil" tag as well, even when devil alone is put in the negative prompt.
>>
File: 003841_00001.mp4 (2 MB)
2 MB MP4
I largely stopped even trying to make images because mostly I don't even know what I would particularly want to see. Occasionally just throw a few almost-random tags in of things I vaguely like to see what comes out.
>>
File: coque7.png (1017.5 KB)
1017.5 KB PNG
>>9176300
Every time I think I'm done the "huge ass" tag dares to me to turn the weight up again like the fucking green goblin mask. Please enjoy and goodnight
>>
>>
>>
>>
>>
>>
>>
File: Odalia Blight.png (3.9 MB)
3.9 MB PNG
Bigger resolution: https://files.catbox.moe/jg6uzq.png
>>
>>
>>
>>
>>
>>
File: HiResPandaPotter_00022_.png (2.8 MB)
2.8 MB PNG
>>9176374
>>
File: 8K3DEVDXHP7MEKFGWQ83K92M90.jpg (95.5 KB)
95.5 KB JPG