/ldg/ - Local Diffusion General
Anonymous 01/12/25(Sun)17:36:23 | 341 comments | 118 images | 🔒 Locked
Discussion of Free and Open-Source Diffusion models.
Bobblehead Cheat Mode Edition
Last Time:>>103862817
>Newfrien Guide
Pick UI -> Download Model -> Slap Model into UI
Breathe In, Gen Out.
>UI
Metastable:https://metastable.studio
SwarmUI:https://github.com/mcmonkeyprojects/SwarmUI
Forge:https://github.com/lllyasviel/stable-diffusion-webui-forge
reForge:https://github.com/Panchovix/stable-diffusion-webui-reForge
ComfyUI:https://github.com/comfyanonymous/ComfyUI
InvokeAI:https://github.com/invoke-ai/InvokeAI
>Models, LoRAs, & Upscalers
https://civitai.com
https://tensor.art/
https://openmodeldb.info
>Training
https://github.com/Nerogar/OneTrainer
https://github.com/kohya-ss/sd-scripts/tree/sd3
https://github.com/derrian-distro/LoRA_Easy_Training_Scripts
>HunyuanVideo
Comfy:https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/
Windows:https://rentry.org/crhcqq54
Training:https://github.com/tdrussell/diffusion-pipe|https://github.com/kohya-ss/musubi-tuner
>Flux
Forge:https://github.com/lllyasviel/stable-diffusion-webui-forge/discussions/1050
ComfyUI:https://comfyanonymous.github.io/ComfyUI_examples/flux
DeDistilled:https://huggingface.co/TheYuriLover/flux-dev-de-distill-GGUF/tree/main
Pixelwave:https://civitai.com/models/141592/pixelwave
>Misc
Share Metadata:https://catbox.moe|https://litterbox.catbox.moe/
Img2Prompt:https://huggingface.co/spaces/fancyfeast/joy-caption-alpha-two
Archive:https://rentry.org/sdg-link
Samplers:https://stable-diffusion-art.com/samplers/
Digital Art Software:https://krita.org/en/
Txt2Img Plugin:https://kritaaidiffusion.com/
Collagebaker:https://www.befunky.com/create/collage/
Video Collagebaker:https://kdenlive.org/en/
>Related
>>>/aco/csdg
>>>/b/degen
>>>/c/kdg
>>>/d/ddg
>>>/e/edg
>>>/h/hdg
>>>/trash/slop
>>>/vt/vtai
>Local Sexting
>>>/g/lmg
>Maintain thread quality
https://rentry.org/debo
Bobblehead Cheat Mode Edition
Last Time:>>103862817
>Newfrien Guide
Pick UI -> Download Model -> Slap Model into UI
Breathe In, Gen Out.
>UI
Metastable:https://metastable.studi
SwarmUI:https://github.com/mcmonkey
Forge:https://github.com/lllyasviel
reForge:https://github.com/Panchovi
ComfyUI:https://github.com/comfyano
InvokeAI:https://github.com/invoke-
>Models, LoRAs, & Upscalers
https://civitai.com
https://tensor.art/
https://openmodeldb.info
>Training
https://github.com/Nerogar/OneTrain
https://github.com/kohya-ss/sd-scri
https://github.com/derrian-distro/L
>HunyuanVideo
Comfy:https://github.com/kijai/Comf
Windows:https://rentry.org/crhcqq54
Training:https://github.com/tdrusse
>Flux
Forge:https://github.com/lllyasviel
ComfyUI:https://comfyanonymous.gith
DeDistilled:https://huggingface.co/
Pixelwave:https://civitai.com/model
>Misc
Share Metadata:https://catbox.moe|https:/
Img2Prompt:https://huggingface.co/s
Archive:https://rentry.org/sdg-link
Samplers:https://stable-diffusion-a
Digital Art Software:https://krita.org/en/
Txt2Img Plugin:https://kritaaidiffusion.com
Collagebaker:https://www.befunky.co
Video Collagebaker:https://kdenlive.org/e
>Related
>>>/aco/csdg
>>>/b/degen
>>>/c/kdg
>>>/d/ddg
>>>/e/edg
>>>/h/hdg
>>>/trash/slop
>>>/vt/vtai
>Local Sexting
>>>/g/lmg
>Maintain thread quality
https://rentry.org/debo
Anonymous 01/12/25(Sun)17:38:18 No.103870558
>>103870474
Schedulers determine how much noise is removed from the latent at a given step. There are some comfy nodes which help you visualize it.
Schedulers determine how much noise is removed from the latent at a given step. There are some comfy nodes which help you visualize it.
Anonymous 01/12/25(Sun)17:38:20 No.103870559
Anonymous 01/12/25(Sun)17:38:40 No.103870562
Anonymous 01/12/25(Sun)17:38:48 No.103870563
Anonymous 01/12/25(Sun)17:39:40 No.103870575
>>103870559
Can you get Flux to do dismemberment or that kind of thing?
Like disfigured? Missing an arm, or with a deformed arm?
Can you get Flux to do dismemberment or that kind of thing?
Like disfigured? Missing an arm, or with a deformed arm?
Anonymous 01/12/25(Sun)17:40:41 No.103870586
>>103870575
dunno maybe, why?
dunno maybe, why?
Anonymous 01/12/25(Sun)17:40:42 No.103870587
>>103870564
It's just 4chan jeet employees. The human cancer.
It's just 4chan jeet employees. The human cancer.
Anonymous 01/12/25(Sun)17:40:52 No.103870592
Blessed thread of frenship
Anonymous 01/12/25(Sun)17:41:41 No.103870600
>>103870564
funny how you didn't attach anything to this post
surely you're not trolling
surely
funny how you didn't attach anything to this post
surely you're not trolling
surely
Anonymous 01/12/25(Sun)17:41:42 No.103870601
Anonymous 01/12/25(Sun)17:41:45 No.103870603
Flux Bigma any day now
Anonymous 01/12/25(Sun)17:42:03 No.103870606
Anonymous 01/12/25(Sun)17:42:40 No.103870618
>>103870600
>you didn't attach anything to this post
same thing can be said for you, where's your slop?
>you didn't attach anything to this post
same thing can be said for you, where's your slop?
Anonymous 01/12/25(Sun)17:42:45 No.103870621
>>103870601
cool it catjak
cool it catjak
Anonymous 01/12/25(Sun)17:43:06 No.103870626
>>103870601
As long as someone posts anime, there are no standards, anime is garbage.
My work is very important, and it's not possible for you to know or see this.
As long as someone posts anime, there are no standards, anime is garbage.
My work is very important, and it's not possible for you to know or see this.
Anonymous 01/12/25(Sun)17:43:14 No.103870630
Anonymous 01/12/25(Sun)17:43:51 No.103870641
Anonymous 01/12/25(Sun)17:44:32 No.103870646
>>103870559
Is this mayli?
Is this mayli?
Anonymous 01/12/25(Sun)17:45:45 No.103870665
imagine spending time seething in this blessed bread instead of improving your gens
very sad
very sad
Anonymous 01/12/25(Sun)17:46:03 No.103870671
>>103870641
Probably not, I have typed "haha" before though.
Probably not, I have typed "haha" before though.
Anonymous 01/12/25(Sun)17:46:08 No.103870672
Anonymous 01/12/25(Sun)17:46:12 No.103870673
I don't believe the collage is decided by a human consciousness? It's just images scrapped from the previous thread and fed into a canvas of a random size and arrangement.
Anonymous 01/12/25(Sun)17:46:14 No.103870675
>>103870665
says the nogen
says the nogen
Anonymous 01/12/25(Sun)17:46:22 No.103870678
>>103870665
good advice catjak, updooted
good advice catjak, updooted
Anonymous 01/12/25(Sun)17:47:16 No.103870684
>>103870673
>It's just images scrapped from the previous thread and fed into a canvas of a random size and arrangement.
you think it chooses randomly like 10% of the slops from the previous thread to make the collage?
>It's just images scrapped from the previous thread and fed into a canvas of a random size and arrangement.
you think it chooses randomly like 10% of the slops from the previous thread to make the collage?
Anonymous 01/12/25(Sun)17:47:16 No.103870686
Anonymous 01/12/25(Sun)17:47:29 No.103870688
>>103870630
I feel you. The only way anyone figures out which is best is by making a comparison grid. It never hurts to explore options.
I feel you. The only way anyone figures out which is best is by making a comparison grid. It never hurts to explore options.
Anonymous 01/12/25(Sun)17:47:33 No.103870689
Anonymous 01/12/25(Sun)17:48:16 No.103870696
>>103870684
That's my working theory.
That's my working theory.
Anonymous 01/12/25(Sun)17:48:54 No.103870699
>>103870688
Do computer scientists understand those terms and what they might imply? I just got pissed off at the one saying "normal", what the fuck does that mean!?
>ah the karrass scheduler, indeed
Do computer scientists understand those terms and what they might imply? I just got pissed off at the one saying "normal", what the fuck does that mean!?
>ah the karrass scheduler, indeed
Anonymous 01/12/25(Sun)17:49:32 No.103870711
Your only function is to seethe aimlessly
Do you ever think about that?
Do you ever think about that?
Anonymous 01/12/25(Sun)17:50:19 No.103870719
Anonymous 01/12/25(Sun)17:51:07 No.103870732
>>103870564
IMAGE BAD.
fucking hell this place gets worse every week, I swear. here, have a flat ass, you've earned it.
IMAGE BAD.
fucking hell this place gets worse every week, I swear. here, have a flat ass, you've earned it.
Anonymous 01/12/25(Sun)17:51:22 No.103870734
Anonymous 01/12/25(Sun)17:52:23 No.103870745
>>103870732
>fucking hell this place gets worse every week
if you hate that place you can leave and go back to /sdg/, that asylum welcomes slop spammers like you
>fucking hell this place gets worse every week
if you hate that place you can leave and go back to /sdg/, that asylum welcomes slop spammers like you
Anonymous 01/12/25(Sun)17:52:24 No.103870746
Anonymous 01/12/25(Sun)17:52:46 No.103870749
>>103870699
Euler is a regular mathematics thing, and most schedulers do pull from legit mathematical equations (don't ask me to elaborate tho I am retard).
Some are literal memes tho. It's important IMO to test all settings on a given scheduler to sus out if it's a meme or you're just using it incorrectly.
Most anon like me and you don't actually understand it at a low level desu and just to with what looks good.
Euler is a regular mathematics thing, and most schedulers do pull from legit mathematical equations (don't ask me to elaborate tho I am retard).
Some are literal memes tho. It's important IMO to test all settings on a given scheduler to sus out if it's a meme or you're just using it incorrectly.
Most anon like me and you don't actually understand it at a low level desu and just to with what looks good.
Anonymous 01/12/25(Sun)17:53:00 No.103870757
Anonymous 01/12/25(Sun)17:54:03 No.103870772
>>103870757
if you have enough skill you can write an ffmpeg script with auto cropping and aligning and shit
if you have enough skill you can write an ffmpeg script with auto cropping and aligning and shit
Anonymous 01/12/25(Sun)17:54:16 No.103870774
>>103870732
/ldg/ never valued images over substantive text. The whole concept of nogens is an /sdg/ism that has no place here.
/ldg/ never valued images over substantive text. The whole concept of nogens is an /sdg/ism that has no place here.
Anonymous 01/12/25(Sun)17:54:24 No.103870777
Anonymous 01/12/25(Sun)17:55:03 No.103870786
>>103870777
We are so back
We are so back
Anonymous 01/12/25(Sun)17:58:02 No.103870816
>>103870777
Flux gen good.
Flux gen good.
Anonymous 01/12/25(Sun)17:58:11 No.103870818
>>103870732
I don't give a fuck about your mediocre 1girl standing AI slop, I'm not lurking /ldg/ to see your turds, I'm here so that I can hear about the news, about the new tools, the new cool github nodes, that's what you need to understand, if you want to spam some shit no one give a fuck, there's /sdg/ for that
I don't give a fuck about your mediocre 1girl standing AI slop, I'm not lurking /ldg/ to see your turds, I'm here so that I can hear about the news, about the new tools, the new cool github nodes, that's what you need to understand, if you want to spam some shit no one give a fuck, there's /sdg/ for that
Anonymous 01/12/25(Sun)17:58:36 No.103870824
Anonymous 01/12/25(Sun)18:00:01 No.103870837
Anonymous 01/12/25(Sun)18:02:09 No.103870853
>>103870818
you sure bring up that other place a lot huh
you sure bring up that other place a lot huh
Anonymous 01/12/25(Sun)18:02:39 No.103870857
IMPORTANT gens.
Anonymous 01/12/25(Sun)18:03:39 No.103870873
>>103870837
become maylionke
become maylionke
Anonymous 01/12/25(Sun)18:03:41 No.103870875
>>103870853
I much prefer someone who contributes nothing than someone wasting thread spaces with AI slop turds, quality >>>> quantity, fuck you and fuck your spam slop
I much prefer someone who contributes nothing than someone wasting thread spaces with AI slop turds, quality >>>> quantity, fuck you and fuck your spam slop
Anonymous 01/12/25(Sun)18:05:29 No.103870903
>>103870732
They're upset because they don't get attention. You're right this past week they've been at it worse than before. Something must be bothering them IRL.
They're upset because they don't get attention. You're right this past week they've been at it worse than before. Something must be bothering them IRL.
Anonymous 01/12/25(Sun)18:05:37 No.103870906
Trying to gen a decapitated goose. I may have to inpaint.
Anonymous 01/12/25(Sun)18:05:58 No.103870908
Anonymous 01/12/25(Sun)18:07:36 No.103870931
>>103870903
>They're upset because they don't get attention
oh the irony, the ones who crave attention are the slop spammers, they're the one who begs to have some (You) and be included into the collage because they made a 5 sec SDXL turd on Forge
>They're upset because they don't get attention
oh the irony, the ones who crave attention are the slop spammers, they're the one who begs to have some (You) and be included into the collage because they made a 5 sec SDXL turd on Forge
Anonymous 01/12/25(Sun)18:07:39 No.103870932
>>103870533
Good collage
Good collage
Anonymous 01/12/25(Sun)18:07:54 No.103870938
Don't worry, I am working on a VIG: Very Important Gen. Should be posted soon and everything will be better after that. I guarantee its quality will rise to the level that /ldg/ demands for its valuable finite image slots.
Anonymous 01/12/25(Sun)18:08:26 No.103870946
Anonymous 01/12/25(Sun)18:08:51 No.103870955
>>103870906
based
based
Anonymous 01/12/25(Sun)18:09:30 No.103870959
>>103870903
I think the drug abuse and isolation is making them crack. All of them have some story of substance abuse and isolation.
I think the drug abuse and isolation is making them crack. All of them have some story of substance abuse and isolation.
Anonymous 01/12/25(Sun)18:09:32 No.103870961
Anonymous 01/12/25(Sun)18:09:43 No.103870965
>>103870946
Look how important this image is.
Look how important this image is.
Anonymous 01/12/25(Sun)18:09:51 No.103870967
>>103870946
Kim Kardashian?
Kim Kardashian?
Anonymous 01/12/25(Sun)18:10:01 No.103870970
Equality is important
Anonymous 01/12/25(Sun)18:10:59 No.103870981
Anonymous 01/12/25(Sun)18:11:03 No.103870983
>>103870961
the ones who crave attention are the ones who spam slop in an attempt to get some reactions out of it, some retard like you actually
the ones who crave attention are the ones who spam slop in an attempt to get some reactions out of it, some retard like you actually
Anonymous 01/12/25(Sun)18:11:47 No.103870989
>>103870983
So why are you cool with the known scitzos in /sdg/?
Why do you always project the same behavior others hate you for?
So why are you cool with the known scitzos in /sdg/?
Why do you always project the same behavior others hate you for?
Anonymous 01/12/25(Sun)18:12:26 No.103870997
>>103870981
Are you requesting a version without the buttchin, or are you simply excited to know some English words?
Are you requesting a version without the buttchin, or are you simply excited to know some English words?
Anonymous 01/12/25(Sun)18:12:44 No.103871002
>>103870989
>why are you cool with the known scitzos in /sdg/?
who says I'm cool with them? you can't make an argument without lying or something? my point is that if you like to spam trash, there's already a trash can that's called /sdg/, you need to go back
>why are you cool with the known scitzos in /sdg/?
who says I'm cool with them? you can't make an argument without lying or something? my point is that if you like to spam trash, there's already a trash can that's called /sdg/, you need to go back
Anonymous 01/12/25(Sun)18:12:45 No.103871003
>>103870959
We can only pity them. And laugh... but mostly pity.
>>103870989
Don't reply to trolling outside of /b/
We can only pity them. And laugh... but mostly pity.
>>103870989
Don't reply to trolling outside of /b/
Anonymous 01/12/25(Sun)18:13:30 No.103871012
>>103870903
I think imagebad doesn't understand why anyone would want to discuss anything with a rabid cunt. (And please excuse my language here but that guy belongs in a mental asylum. spammy, extremely aggressive, now he's hallucinating 'samefagging' nonsense (and he guesses wrong of course))
>>103870959
something ran out yeah.
enough of this crap I hate it when I get baited like a muppet.
I think imagebad doesn't understand why anyone would want to discuss anything with a rabid cunt. (And please excuse my language here but that guy belongs in a mental asylum. spammy, extremely aggressive, now he's hallucinating 'samefagging' nonsense (and he guesses wrong of course))
>>103870959
something ran out yeah.
enough of this crap I hate it when I get baited like a muppet.
Anonymous 01/12/25(Sun)18:14:00 No.103871019
what a shitehole
Anonymous 01/12/25(Sun)18:14:08 No.103871022
>>103870997
learn to read buttchinigger
learn to read buttchinigger
Anonymous 01/12/25(Sun)18:15:45 No.103871042
>>103871022
Buttchins are very hard to gen on sdxl. On flux you can turn them off if you want people to think you are using inferior tech.
Buttchins are very hard to gen on sdxl. On flux you can turn them off if you want people to think you are using inferior tech.
Anonymous 01/12/25(Sun)18:17:57 No.103871074
>>103870965
"a dog"
"a dog"
Anonymous 01/12/25(Sun)18:19:50 No.103871098
the blackmagic model is interesting. it can either give you really good abstract looking ones like this,
Anonymous 01/12/25(Sun)18:20:18 No.103871105
>>103870837
i am scared but also horny
i am scared but also horny
Anonymous 01/12/25(Sun)18:20:51 No.103871112
or slop things like this
Anonymous 01/12/25(Sun)18:22:44 No.103871133
>>103871074
I can't believe how important I am, and this image is. I bet hank hill would like my autograph.
I can't believe how important I am, and this image is. I bet hank hill would like my autograph.
Anonymous 01/12/25(Sun)18:23:07 No.103871135
>>103871012
>I hate it when I get baited like a muppet.
dont take it too hard, they repeat the same thing mostly so youll figure out how to spot it soon enough
>I hate it when I get baited like a muppet.
dont take it too hard, they repeat the same thing mostly so youll figure out how to spot it soon enough
Anonymous 01/12/25(Sun)18:23:35 No.103871140
Anonymous 01/12/25(Sun)18:23:58 No.103871147
>>103871098
thats a nice once. gonna check it out
the 2nd one just looks like that fenner lora (kind of)
thats a nice once. gonna check it out
the 2nd one just looks like that fenner lora (kind of)
Anonymous 01/12/25(Sun)18:23:59 No.103871148
>>103871133
A very important panda.
>>103871098
I like the color thing it's doing. What's the blackmagic model?
A very important panda.
>>103871098
I like the color thing it's doing. What's the blackmagic model?
Anonymous 01/12/25(Sun)18:25:27 No.103871160
>>103871148
search civitai retard
search civitai retard
Anonymous 01/12/25(Sun)18:25:27 No.103871161
Flux doesn't know what a sistrum is.
https://en.wikipedia.org/wiki/Sistrum
https://en.wikipedia.org/wiki/Sistr
Anonymous 01/12/25(Sun)18:26:26 No.103871164
>>103871148
this one. i passed "neon" and "blacklight" for the color parts. pic rel is with out those 2 prompts
https://civitai.com/models/979440/illustrious-black-magic-illustrious-noob-more
this one. i passed "neon" and "blacklight" for the color parts. pic rel is with out those 2 prompts
https://civitai.com/models/979440/i
Anonymous 01/12/25(Sun)18:26:56 No.103871171
>>103871160
I don't think civitai allows retard loras, but I'd download one and gen your momma.
I don't think civitai allows retard loras, but I'd download one and gen your momma.
Anonymous 01/12/25(Sun)18:26:56 No.103871172
>"stop spamming slop" poster comes out at the beginning of every thread
>"another anon" starts spamming slop
>first anon nowhere to be found
Curious
>"another anon" starts spamming slop
>first anon nowhere to be found
Curious
Anonymous 01/12/25(Sun)18:27:28 No.103871177
Anonymous 01/12/25(Sun)18:28:29 No.103871187
>>103871172
filtered
filtered
Anonymous 01/12/25(Sun)18:28:41 No.103871190
Anonymous 01/12/25(Sun)18:29:12 No.103871194
Anonymous 01/12/25(Sun)18:30:54 No.103871208
>>103871177
but shes no longer maylionkey :(
but shes no longer maylionkey :(
Anonymous 01/12/25(Sun)18:31:16 No.103871214
Anonymous 01/12/25(Sun)18:33:43 No.103871241
Anonymous 01/12/25(Sun)18:35:08 No.103871260
Anonymous 01/12/25(Sun)18:35:26 No.103871263
Surely one day they will coincide and we'll see him reply to an actual slop spammer
Surely
Surely
Anonymous 01/12/25(Sun)18:36:12 No.103871269
>>103871263
that can't happen, I filtered them lol
that can't happen, I filtered them lol
Anonymous 01/12/25(Sun)18:37:23 No.103871280
>>103870600
>surely
>surely
>>103871263
>Surely
>Surely
Surely one day he'll say "Surely" more than 2 times, surely
>surely
>surely
>>103871263
>Surely
>Surely
Surely one day he'll say "Surely" more than 2 times, surely
Anonymous 01/12/25(Sun)18:37:59 No.103871288
Certainly
Anonymous 01/12/25(Sun)18:38:24 No.103871296
Anonymous 01/12/25(Sun)18:38:32 No.103871299
>>103871263
Can you identify the slop spammer posts and stop spamming slop posts by linking them so I can get a good idea of who you think each personality is?
Can you identify the slop spammer posts and stop spamming slop posts by linking them so I can get a good idea of who you think each personality is?
Anonymous 01/12/25(Sun)18:38:43 No.103871300
Anonymous 01/12/25(Sun)18:39:20 No.103871306
>>103870467
>anyone notice how girls looks more attractive in Hunyuanvideowrapper vs the normal Comfyui way?
Doesn't comfy sometimes fuck up implementation? I remember anon bringing this up before.
>anyone notice how girls looks more attractive in Hunyuanvideowrapper vs the normal Comfyui way?
Doesn't comfy sometimes fuck up implementation? I remember anon bringing this up before.
Anonymous 01/12/25(Sun)18:39:52 No.103871313
Anonymous 01/12/25(Sun)18:41:11 No.103871328
>>103871306
I'm not noticing a drop in quality when I switched to comfy core, I even prefer using it because it has Q8, and recently he added "res_multistep" sampler and that shit makes hunyuan gens less blurry, so I'm eating pretty good right now
I'm not noticing a drop in quality when I switched to comfy core, I even prefer using it because it has Q8, and recently he added "res_multistep" sampler and that shit makes hunyuan gens less blurry, so I'm eating pretty good right now
Anonymous 01/12/25(Sun)18:43:08 No.103871350
Anonymous 01/12/25(Sun)18:43:43 No.103871359
>>103871350
no it's a sampler not a scheduler
no it's a sampler not a scheduler
Anonymous 01/12/25(Sun)18:44:18 No.103871363
>>103871359
kek, excuse my poor reading comprehension
kek, excuse my poor reading comprehension
Anonymous 01/12/25(Sun)18:44:25 No.103871364
Anonymous 01/12/25(Sun)18:44:54 No.103871373
>>103871363
it's all right kek
it's all right kek
Anonymous 01/12/25(Sun)18:46:16 No.103871393
Anonymous 01/12/25(Sun)18:46:20 No.103871394
>>103871373
I appreciate the info, on the topic of scheduler, which are you using with the "res_multistep" sampler?
I appreciate the info, on the topic of scheduler, which are you using with the "res_multistep" sampler?
Anonymous 01/12/25(Sun)18:47:30 No.103871403
>>103871394
I went for "simple", before I was using "normal" but I think it's broken, comfy has to fix that one
I went for "simple", before I was using "normal" but I think it's broken, comfy has to fix that one
Anonymous 01/12/25(Sun)18:47:34 No.103871405
If you were to guess, what percentage of the population either doesn't care about AI or uses it themselves?
Wondering when the "AI witchhunts" will cease.
Wondering when the "AI witchhunts" will cease.
Anonymous 01/12/25(Sun)18:48:01 No.103871411
Anonymous 01/12/25(Sun)18:48:52 No.103871421
Anonymous 01/12/25(Sun)18:48:57 No.103871422
>>103871405
nowdays? I'd say 20 to 30%, there's still a shit ton of people who hate when something is made by AI, like the new coca cola ad got a lot of dramas
https://www.youtube.com/watch?v=4RSTupbfGog
nowdays? I'd say 20 to 30%, there's still a shit ton of people who hate when something is made by AI, like the new coca cola ad got a lot of dramas
https://www.youtube.com/watch?v=4RS
Anonymous 01/12/25(Sun)18:49:46 No.103871433
>>103871405
The ones who piss and shit themselves seething with fury whenever they see AI art are definitely just a vocal minority. Normies don't give a fuck.
The ones who piss and shit themselves seething with fury whenever they see AI art are definitely just a vocal minority. Normies don't give a fuck.
Anonymous 01/12/25(Sun)18:51:03 No.103871456
Anonymous 01/12/25(Sun)18:55:04 No.103871493
>>103871433
Most artists who are good at their craft are not threatened by AI. If you go through the profile of "Artists" who hate AI you'll notice a growing trend of their art being slop. The further you go back in their history you'll notice another trend of it never getting any better.
Most artists who are good at their craft are not threatened by AI. If you go through the profile of "Artists" who hate AI you'll notice a growing trend of their art being slop. The further you go back in their history you'll notice another trend of it never getting any better.
Anonymous 01/12/25(Sun)18:55:06 No.103871495
>>103871405
>Wondering when the "AI witchhunts" will cease.
I doubt it'll calm down, AI is taking a shit ton of jobs, especially chatgpt, that shit is replacing many many software engineers, so the resentment will still be there
>Wondering when the "AI witchhunts" will cease.
I doubt it'll calm down, AI is taking a shit ton of jobs, especially chatgpt, that shit is replacing many many software engineers, so the resentment will still be there
Anonymous 01/12/25(Sun)18:56:17 No.103871509
>>103871164
not sold. nice tho that modelmaker gives detailed info on what he actually merged together, and I must say that faceless triangular shaped wing-thing is cool AF, ty for the idea muhahah
>>103871405
large percentage of population thinks exactly what tv/news tells them to, as for the rest, I dunno? just noise anyways. seen some witchhunt-tier stuff lately, yeah, worrying.
not sold. nice tho that modelmaker gives detailed info on what he actually merged together, and I must say that faceless triangular shaped wing-thing is cool AF, ty for the idea muhahah
>>103871405
large percentage of population thinks exactly what tv/news tells them to, as for the rest, I dunno? just noise anyways. seen some witchhunt-tier stuff lately, yeah, worrying.
Anonymous 01/12/25(Sun)18:56:29 No.103871511
>>103871493
>Most artists who are good at their craft are not threatened by AI.
at some point AI will be able to make great drawings and not just slop, we got a glimpse of that with MJ, it's the first model that made me believe a real human could draw some of its outputs
>Most artists who are good at their craft are not threatened by AI.
at some point AI will be able to make great drawings and not just slop, we got a glimpse of that with MJ, it's the first model that made me believe a real human could draw some of its outputs
Anonymous 01/12/25(Sun)18:57:34 No.103871518
Anonymous 01/12/25(Sun)18:58:22 No.103871529
>>103871511
>at some point AI will be able to make great drawings and not just slop
We're already there, but it requires more than just typing 1girl big sex into the latest slop model and posting the result, which is what 99% of people do.
>at some point AI will be able to make great drawings and not just slop
We're already there, but it requires more than just typing 1girl big sex into the latest slop model and posting the result, which is what 99% of people do.
Anonymous 01/12/25(Sun)18:59:28 No.103871539
>>103871529
yeah my b, that's what I implied, an AI model that could create masterpiece with a simple prompt
yeah my b, that's what I implied, an AI model that could create masterpiece with a simple prompt
Anonymous 01/12/25(Sun)19:00:39 No.103871554
>>103871511
The average MJ gen will be better than the average SD gen but the greatest gen you will ever see will be from a local user because we have the tools and MJ does not.
The average MJ gen will be better than the average SD gen but the greatest gen you will ever see will be from a local user because we have the tools and MJ does not.
Anonymous 01/12/25(Sun)19:01:02 No.103871560
A new revolution is happening gentlemen
https://github.com/SonyResearch/micro_diffusion
>This repository provides a minimalistic implementation of our approach to training large-scale diffusion models from scratch on an extremely low budget. In particular, using only 37M publicly available real and synthetic images, we train a 1.16 billion parameter sparse transformer with a total cost of only $1,890 and achieve an FID of 12.7 in zero-shot generation on the COCO dataset.
https://github.com/SonyResearch/mic
>This repository provides a minimalistic implementation of our approach to training large-scale diffusion models from scratch on an extremely low budget. In particular, using only 37M publicly available real and synthetic images, we train a 1.16 billion parameter sparse transformer with a total cost of only $1,890 and achieve an FID of 12.7 in zero-shot generation on the COCO dataset.
Anonymous 01/12/25(Sun)19:01:38 No.103871564
The third and last of my experiment
Anonymous 01/12/25(Sun)19:01:45 No.103871565
Anonymous 01/12/25(Sun)19:02:06 No.103871567
>>103871560
>using only 37M publicly available real and synthetic images
I'd pay them $1891 to not do that.
>using only 37M publicly available real and synthetic images
I'd pay them $1891 to not do that.
Anonymous 01/12/25(Sun)19:02:28 No.103871573
>>103871567
kek
kek
Anonymous 01/12/25(Sun)19:02:50 No.103871578
Anonymous 01/12/25(Sun)19:03:18 No.103871585
Anonymous 01/12/25(Sun)19:05:24 No.103871599
>>103871422
DESU 126k doesnt seem like a lot in the grand scheme of things
DESU 126k doesnt seem like a lot in the grand scheme of things
Anonymous 01/12/25(Sun)19:06:24 No.103871602
Anonymous 01/12/25(Sun)19:07:00 No.103871605
Anonymous 01/12/25(Sun)19:07:57 No.103871611
>>103871554
idk about that, you don't have tools to "unslop" an image and make it look like a human draw that, the best combo would be making a first picture with MJ and then inpaint it with SD tools
idk about that, you don't have tools to "unslop" an image and make it look like a human draw that, the best combo would be making a first picture with MJ and then inpaint it with SD tools
Anonymous 01/12/25(Sun)19:09:02 No.103871617
Anonymous 01/12/25(Sun)19:09:35 No.103871622
>>103871617
please do tell
please do tell
Anonymous 01/12/25(Sun)19:10:29 No.103871631
>>103871605
nice!
nice!
Anonymous 01/12/25(Sun)19:11:48 No.103871640
>>103871611
almost every tool created is in some way meant to help unslop an image if it isnt for a speed increase desu
>>103871622
does this look hand drawn?
almost every tool created is in some way meant to help unslop an image if it isnt for a speed increase desu
>>103871622
does this look hand drawn?
Anonymous 01/12/25(Sun)19:11:53 No.103871641
>>103871328
I'm the original anon, yeah the quality is fine in comfy core, it's just that actual attractiveness of girls in general are not as good, I also prefer using comfy core now, but that's something that's bothering me still lol
Try to do a comparison without loras, and you might see what I mean
I'm the original anon, yeah the quality is fine in comfy core, it's just that actual attractiveness of girls in general are not as good, I also prefer using comfy core now, but that's something that's bothering me still lol
Try to do a comparison without loras, and you might see what I mean
Anonymous 01/12/25(Sun)19:12:35 No.103871645
>>103871260
Moar steps desu
Moar steps desu
Anonymous 01/12/25(Sun)19:13:19 No.103871658
Don't really like how dark vpred is compared to epsilon but hey, sometimes it works for the outputs
>>103871602
Thanks!
>>103871602
Thanks!
Anonymous 01/12/25(Sun)19:14:25 No.103871669
>>103871641
try the new "res_multistep" sampler and see if that fixes anything
try the new "res_multistep" sampler and see if that fixes anything
Anonymous 01/12/25(Sun)19:14:32 No.103871675
>>103871631
is that the summerjack girl?
is that the summerjack girl?
Anonymous 01/12/25(Sun)19:15:34 No.103871684
>>103871669
nta but I want to try it. How does it compare to lcm?
nta but I want to try it. How does it compare to lcm?
Anonymous 01/12/25(Sun)19:15:55 No.103871690
Feels like depending on the scheduler the effect of the LoRA becomes better or worse.
Anonymous 01/12/25(Sun)19:16:01 No.103871694
>>103871669
yeah I tried that, it makes the images sharper and clearer, but the actual girls stay like the same person.
It's not a big issue because loras etc fixes this issue, it's just more of a weird mystery thing.
yeah I tried that, it makes the images sharper and clearer, but the actual girls stay like the same person.
It's not a big issue because loras etc fixes this issue, it's just more of a weird mystery thing.
Anonymous 01/12/25(Sun)19:16:08 No.103871696
>>103870592
Blessed
Blessed
Anonymous 01/12/25(Sun)19:17:01 No.103871707
>>103871690
That looks clean, which sampler was that
That looks clean, which sampler was that
Anonymous 01/12/25(Sun)19:17:21 No.103871710
>>103871684
Might be something positive to come out of cosmos.
Can't even do a PoV of a robot tasing amazon workers for being lazy right. Isn't that what the whole model was designed for?
Might be something positive to come out of cosmos.
Can't even do a PoV of a robot tasing amazon workers for being lazy right. Isn't that what the whole model was designed for?
Anonymous 01/12/25(Sun)19:17:35 No.103871713
>>103871694
maybe it's an issue with the scheduler, I have no idea what sheduler kijai is using
maybe it's an issue with the scheduler, I have no idea what sheduler kijai is using
Anonymous 01/12/25(Sun)19:17:37 No.103871714
Anonymous 01/12/25(Sun)19:18:02 No.103871720
>>103871690
Can you make a realistic tentacle monster scene
Can you make a realistic tentacle monster scene
Anonymous 01/12/25(Sun)19:18:22 No.103871723
>>103871707
res_multistep. Seems pretty good
res_multistep. Seems pretty good
Anonymous 01/12/25(Sun)19:19:07 No.103871731
>>103871723
yeah better than euler for sure, although I wonder if these samplers affect animation too, what if euler has better animation while res_multistep has better image quality, hmmm
yeah better than euler for sure, although I wonder if these samplers affect animation too, what if euler has better animation while res_multistep has better image quality, hmmm
Anonymous 01/12/25(Sun)19:19:30 No.103871734
>>103871710
That's hilarious
That's hilarious
Anonymous 01/12/25(Sun)19:19:30 No.103871735
>>103871723
>res_multistep. Seems pretty good
yeah that's a really good sampler, I was bothered with the blurry gens from hunyuan and that makes it more "HD" even though I'm still going for 512x512 res kek
>res_multistep. Seems pretty good
yeah that's a really good sampler, I was bothered with the blurry gens from hunyuan and that makes it more "HD" even though I'm still going for 512x512 res kek
Anonymous 01/12/25(Sun)19:20:31 No.103871746
>>103871731
Does lcm work with huan ding dong or whatever the slanty eyed trash merchants call it?
Does lcm work with huan ding dong or whatever the slanty eyed trash merchants call it?
Anonymous 01/12/25(Sun)19:20:37 No.103871747
>>103871658
>Don't really like how dark vpred is compared to epsilon
It's to the point where I cannot use some samplers because they overcook the output. That was something eps did better.
>Don't really like how dark vpred is compared to epsilon
It's to the point where I cannot use some samplers because they overcook the output. That was something eps did better.
Anonymous 01/12/25(Sun)19:21:56 No.103871760
>>103871713
yeah I think that's it
FlowMatchDiscreteScheduler
Supposedly closer to what Hunyuan is supposed to use or something IIRC
yeah I think that's it
FlowMatchDiscreteScheduler
Supposedly closer to what Hunyuan is supposed to use or something IIRC
Anonymous 01/12/25(Sun)19:22:56 No.103871772
>>103871746
it's there as an option but haven't tried it
it's there as an option but haven't tried it
Anonymous 01/12/25(Sun)19:23:54 No.103871781
for how much you guys pride yourselves on discussion here it doesn't show because all I see is fluxslop from the same three posters and the asian bukkake richdad spammer
Anonymous 01/12/25(Sun)19:24:14 No.103871785
>>103871747
Is there no way to remedy it? Not even cfg rescale fixes how utterly broken the contrast is.
Is there no way to remedy it? Not even cfg rescale fixes how utterly broken the contrast is.
Anonymous 01/12/25(Sun)19:24:51 No.103871792
Anonymous 01/12/25(Sun)19:25:18 No.103871800
Anonymous 01/12/25(Sun)19:25:44 No.103871806
>>103871785
euler_cfg_pp with a cfg value of 0.5 (on comfy, which i think translates to 1.5 on forge?) is the only way ive found to mitigate it as much as possible
rescale does nothing for me either
euler_cfg_pp with a cfg value of 0.5 (on comfy, which i think translates to 1.5 on forge?) is the only way ive found to mitigate it as much as possible
rescale does nothing for me either
Anonymous 01/12/25(Sun)19:27:47 No.103871827
>>103871714
I only generate important work.
I only generate important work.
Anonymous 01/12/25(Sun)19:27:51 No.103871830
>>103871675
>summerjack gir
Many moons ago there was a summerjack girl poster I wonder where he is now
>summerjack gir
Many moons ago there was a summerjack girl poster I wonder where he is now
Anonymous 01/12/25(Sun)19:28:53 No.103871843
>>103871760
>FlowMatchDiscreteScheduler
this is what claude told me about its scheduler
https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/blob/836b26aff3944a06f592d7446b2f68159ab48bb8/hyvideo/diffusion/schedulers/scheduling_flow_match_discrete.py#L48
>FlowMatchDiscreteScheduler
this is what claude told me about its scheduler
https://github.com/kijai/ComfyUI-Hu
Anonymous 01/12/25(Sun)19:29:21 No.103871850
>>103871785
vpred is also tricky prompt wise, might need lower weights on some stuff (or higher!)
>>103871806
that is the sampler I ended up using as well, but higher cfg, still learning tho
vpred is also tricky prompt wise, might need lower weights on some stuff (or higher!)
>>103871806
that is the sampler I ended up using as well, but higher cfg, still learning tho
Anonymous 01/12/25(Sun)19:32:13 No.103871881
>>103871827
>>103871800
>>103871792
>>103871714
>>103871710
>>103871690
>>103871684
>>103871605
this is all just /sdg/ spam except with a more expensive computer so I guess that makes it better right, just merge at this point
>>103871800
>>103871792
>>103871714
>>103871710
>>103871690
>>103871684
>>103871605
this is all just /sdg/ spam except with a more expensive computer so I guess that makes it better right, just merge at this point
Anonymous 01/12/25(Sun)19:32:52 No.103871891
>hes still seething
Anonymous 01/12/25(Sun)19:33:03 No.103871895
>>103871830
Yeah, a lot of people vanished after a while.
Yeah, a lot of people vanished after a while.
Anonymous 01/12/25(Sun)19:33:25 No.103871898
>>103871881
/ldg/ is the country club and /sdg/ is the ymca. It only benefits /sdg/ to merge back, not visa versa.
/ldg/ is the country club and /sdg/ is the ymca. It only benefits /sdg/ to merge back, not visa versa.
Anonymous 01/12/25(Sun)19:33:44 No.103871904
Anonymous 01/12/25(Sun)19:34:34 No.103871915
Anonymous 01/12/25(Sun)19:35:28 No.103871927
Anonymous 01/12/25(Sun)19:35:55 No.103871933
WOW! SUCH BEAUTY! MUCH PRETTY!
Anonymous 01/12/25(Sun)19:37:27 No.103871949
https://civitai.com/models/1053376/cara-delevigne?modelVersionId=1275773
That smooth skin is scaring me
That smooth skin is scaring me
Anonymous 01/12/25(Sun)19:39:14 No.103871969
Anonymous 01/12/25(Sun)19:40:24 No.103871978
>>103871927
neat
neat
Anonymous 01/12/25(Sun)19:40:54 No.103871985
>>103871898
/ldg/ is the psych ward and /sdg/ is the fun cozy place
/ldg/ is the psych ward and /sdg/ is the fun cozy place
Anonymous 01/12/25(Sun)19:41:02 No.103871987
>>103871760
>>103871843
I think it's something like that at the end, that's really similar to karras or exponential
>>103871843
I think it's something like that at the end, that's really similar to karras or exponential
Anonymous 01/12/25(Sun)19:45:38 No.103872030
Anonymous 01/12/25(Sun)19:46:49 No.103872042
Average /ldg/ discussion
>So anyway I think we can get cleaner results with this scheduler seen in this paper here https://arxiv.org/pdf/2308.02157
>SAYS NOT NO GEN a1111pooopooxl.102445.jpg
>Hey we don't really want to flood the conversation with meaningless images
>NO GEN! Lol how mad are you? a1111pooopooxl.102446.jpg
>Well, we already have a thread if you want to post your work but we're more about discussion of the tech here with a few examples.
>SOUNDS LIKE NOGEN COPE a1111pooopooxl.102447.jpg
>So anyway I think we can get cleaner results with this scheduler seen in this paper here https://arxiv.org/pdf/2308.02157
>SAYS NOT NO GEN a1111pooopooxl.102445.jpg
>Hey we don't really want to flood the conversation with meaningless images
>NO GEN! Lol how mad are you? a1111pooopooxl.102446.jpg
>Well, we already have a thread if you want to post your work but we're more about discussion of the tech here with a few examples.
>SOUNDS LIKE NOGEN COPE a1111pooopooxl.102447.jpg
Anonymous 01/12/25(Sun)19:48:29 No.103872062
>>103872042
yeah, that's pretty much this unfortunately
yeah, that's pretty much this unfortunately
Anonymous 01/12/25(Sun)19:48:41 No.103872065
WELL I THINK ANON IS A CUTIE WHAT ABOUT THAT HUH????
Anonymous 01/12/25(Sun)19:49:40 No.103872072
>>103872042
this but it's just me samefagging each character
this but it's just me samefagging each character
Anonymous 01/12/25(Sun)19:51:10 No.103872082
Anonymous 01/12/25(Sun)19:51:35 No.103872086
Crazy how faggots that had no hand in this general being created keep screeching what it ought to be
Anonymous 01/12/25(Sun)19:51:56 No.103872088
Anonymous 01/12/25(Sun)19:52:01 No.103872091
>>103872030
i havent strayed far away from that with vpred. sometimes ill go back to euler_smea_dy but even that one tends to burn.
>>103872086
this
i havent strayed far away from that with vpred. sometimes ill go back to euler_smea_dy but even that one tends to burn.
>>103872086
this
Anonymous 01/12/25(Sun)19:52:51 No.103872100
This is now a Mayli thread
Anonymous 01/12/25(Sun)19:54:00 No.103872114
Anonymous 01/12/25(Sun)19:54:16 No.103872115
>>103872086
We're NOT faggots and you SHOULD post MORE LOW QUALITY GENS NOW!
We're NOT faggots and you SHOULD post MORE LOW QUALITY GENS NOW!
Anonymous 01/12/25(Sun)19:54:44 No.103872119
>took them that long to reply
Pathetic
Pathetic
Anonymous 01/12/25(Sun)19:57:35 No.103872153
Anonymous 01/12/25(Sun)19:59:28 No.103872168
>>103872030
looks good no? upscales come out great btw (imo)
>>103872100
you are the guy who posted her when flux came out, right?
looks good no? upscales come out great btw (imo)
>>103872100
you are the guy who posted her when flux came out, right?
Anonymous 01/12/25(Sun)19:59:51 No.103872172
>>103872100
you're half the thread already, standing lora 1girl so original
you're half the thread already, standing lora 1girl so original
Anonymous 01/12/25(Sun)20:03:39 No.103872204
>>103872168
>looks good no
Yeah, i like it. Inpaint requires going back to other samplers though, at least on swarm/comfy
>looks good no
Yeah, i like it. Inpaint requires going back to other samplers though, at least on swarm/comfy
Anonymous 01/12/25(Sun)20:05:03 No.103872212
>>103872100
Do a weather map, they have funny names when it does them.
Do a weather map, they have funny names when it does them.
Anonymous 01/12/25(Sun)20:05:10 No.103872214
>>103872168
No, Im one of the several anons who play with her lora
No, Im one of the several anons who play with her lora
Anonymous 01/12/25(Sun)20:05:17 No.103872215
I'm just here because I hate purple witch and the quokka. They're fake ass people and nobody calls them out on it.
Anonymous 01/12/25(Sun)20:07:35 No.103872229
>>103872204
Very nice
Very nice
Anonymous 01/12/25(Sun)20:08:01 No.103872231
>>103872215
you can filter their filename ass to never see them ever again though
you can filter their filename ass to never see them ever again though
Anonymous 01/12/25(Sun)20:08:01 No.103872232
>>103872204
not tried inpainting yet, I'll get to that eventually, watercolor yay
not tried inpainting yet, I'll get to that eventually, watercolor yay
Anonymous 01/12/25(Sun)20:09:00 No.103872242
>>103872231
My hatred for them goes well beyond filtering.
My hatred for them goes well beyond filtering.
Anonymous 01/12/25(Sun)20:10:22 No.103872254
God I love mayli
Anonymous 01/12/25(Sun)20:11:46 No.103872263
>>103870672
>>103870689
What are you using to generate these? I've been trying to get Flux to do pixel art and haven't had much luck so far
>>103870689
What are you using to generate these? I've been trying to get Flux to do pixel art and haven't had much luck so far
Anonymous 01/12/25(Sun)20:13:01 No.103872274
>>103872263
iirc that guy was using Flux. Can't remember how, could be wrong.
iirc that guy was using Flux. Can't remember how, could be wrong.
Anonymous 01/12/25(Sun)20:13:13 No.103872277
>>103872263
It's pixart sigma
It's pixart sigma
Anonymous 01/12/25(Sun)20:13:13 No.103872278
>>103872263
he channels the energy of every Chang currently in existence across infinite parallel universes
he channels the energy of every Chang currently in existence across infinite parallel universes
Anonymous 01/12/25(Sun)20:13:15 No.103872280
o7
Anonymous 01/12/25(Sun)20:14:02 No.103872290
>>103872277
checked. aha, sorry
checked. aha, sorry
Anonymous 01/12/25(Sun)20:14:50 No.103872296
>>103872277
Basted
Basted
Anonymous 01/12/25(Sun)20:17:29 No.103872316
>>103871641
Can you post a comparison so I can judge your taste in women?
Can you post a comparison so I can judge your taste in women?
Anonymous 01/12/25(Sun)20:22:39 No.103872368
>>103872274
the most important LIFE ALTERING 1girl ever.
the most important LIFE ALTERING 1girl ever.
Anonymous 01/12/25(Sun)20:24:35 No.103872390
Anonymous 01/12/25(Sun)20:26:26 No.103872402
>>103871760
>>103871987
I have some news for you anon, kijai's FlowMatchDiscreteScheduler and Comfy core "Simple" scheduler are absolutely identical when you put a shift > 0
>>103871987
I have some news for you anon, kijai's FlowMatchDiscreteScheduler and Comfy core "Simple" scheduler are absolutely identical when you put a shift > 0
Anonymous 01/12/25(Sun)20:26:46 No.103872404
>>103872390
Can that only do asians?
Can that only do asians?
Anonymous 01/12/25(Sun)20:27:15 No.103872408
>>103872404
In theory yes.
In theory yes.
Anonymous 01/12/25(Sun)20:27:40 No.103872412
Why Mayli?
Anonymous 01/12/25(Sun)20:28:02 No.103872418
Why not Mayli?
Anonymous 01/12/25(Sun)20:31:23 No.103872458
>>103871800
Is this flux? I haven't used it but I would like to make pictures like this. Can you walk me through it?
Is this flux? I haven't used it but I would like to make pictures like this. Can you walk me through it?
Anonymous 01/12/25(Sun)20:34:41 No.103872494
Anonymous 01/12/25(Sun)20:34:52 No.103872496
another memory leak, comfy? Every pull I take, the more slow it gets
Anonymous 01/12/25(Sun)20:38:02 No.103872529
>>103872496
I have the latest pull and I'm not getting memory leak, that's weird
I have the latest pull and I'm not getting memory leak, that's weird
Anonymous 01/12/25(Sun)20:39:16 No.103872538
I love this one. I just need her to have smoother skin.
Anonymous 01/12/25(Sun)20:40:36 No.103872552
>>103872458
>Is this flux?
yes
>I haven't used it but I would like to make pictures like this.
check if your computer has a good gpu with lots of vram
if "no" then upgrade
if yes
>download comfyUI windows portable
>install the comfyUI manager
>download flux and put flux into the right folders
>download lora and put lora into lora folder
>load workflow or make one yourself
>hit queue
>wait
>done
>Is this flux?
yes
>I haven't used it but I would like to make pictures like this.
check if your computer has a good gpu with lots of vram
if "no" then upgrade
if yes
>download comfyUI windows portable
>install the comfyUI manager
>download flux and put flux into the right folders
>download lora and put lora into lora folder
>load workflow or make one yourself
>hit queue
>wait
>done
Anonymous 01/12/25(Sun)20:41:30 No.103872561
>>103872538
As usual, I am saving the world with my gens.
As usual, I am saving the world with my gens.
Anonymous 01/12/25(Sun)20:42:10 No.103872567
A good deal of the flux hate in these threads is vramlet cope.
Anonymous 01/12/25(Sun)20:42:58 No.103872577
>>103872567
A good deal of anons are vramlets.
A good deal of anons are vramlets.
Anonymous 01/12/25(Sun)20:43:02 No.103872578
>>103872529
its happening with a flux workflow that I have, I have to use the clean vram command to make it work, otherwise the vram doesnt clean up automatically
its happening with a flux workflow that I have, I have to use the clean vram command to make it work, otherwise the vram doesnt clean up automatically
Anonymous 01/12/25(Sun)20:43:11 No.103872579
>>103872567
Many of the tools you enjoy were made by vramlets. never forget that.
Many of the tools you enjoy were made by vramlets. never forget that.
Anonymous 01/12/25(Sun)20:44:30 No.103872587
>>103872577
just buy a 4090 or a 5090 next month.
just buy a 4090 or a 5090 next month.
Anonymous 01/12/25(Sun)20:44:44 No.103872590
>>103872577
idk, I just think that if you critique flux you should at least prove you can run it.
idk, I just think that if you critique flux you should at least prove you can run it.
Anonymous 01/12/25(Sun)20:44:53 No.103872593
>>103872552
can you share your workflow?
can you share your workflow?
Anonymous 01/12/25(Sun)20:45:54 No.103872600
>>103872590
Anyone with over 10gb of vram can run it no?
Anyone with over 10gb of vram can run it no?
Anonymous 01/12/25(Sun)20:46:03 No.103872601
>>103872593
NTA, but just use the examples found here
https://github.com/comfyanonymous/ComfyUI_examples/tree/master/flux
NTA, but just use the examples found here
https://github.com/comfyanonymous/C
Anonymous 01/12/25(Sun)20:46:54 No.103872609
>>103872590
i did when it came out but havent in quite awhile. i should assuming teacache makes a noticable difference. im still burned by anon saying quants are faster than fp8 on my shitcard.
>>103872600
i have less and can run it but its not fun
i did when it came out but havent in quite awhile. i should assuming teacache makes a noticable difference. im still burned by anon saying quants are faster than fp8 on my shitcard.
>>103872600
i have less and can run it but its not fun
Anonymous 01/12/25(Sun)20:47:05 No.103872612
Anonymous 01/12/25(Sun)20:47:50 No.103872619
the only reason I keep using stable diffusion is due to an extremely niche fetish of having 70 year old muscular women with a giant ass masturbating with squat fuck on a dildo with her 40 year old daughter watching and masturbating to it
Anonymous 01/12/25(Sun)20:48:08 No.103872625
>>103872609
I just found it too stuborn and the morons that released it did a piss poor job with documentation. I really think that alone can make or break model adoption. You should know what the fuck your model can do before setting it out to the wild and not praying some autistic redditor figures it out for you.
I just found it too stuborn and the morons that released it did a piss poor job with documentation. I really think that alone can make or break model adoption. You should know what the fuck your model can do before setting it out to the wild and not praying some autistic redditor figures it out for you.
Anonymous 01/12/25(Sun)20:48:09 No.103872626
>>103872612
>mayli, your space suit fitting is completed. you can return to your pod now
>thank you, mayli
>mayli, your space suit fitting is completed. you can return to your pod now
>thank you, mayli
Anonymous 01/12/25(Sun)20:49:14 No.103872640
>>103872619
Pic not related right?
Pic not related right?
Anonymous 01/12/25(Sun)20:49:25 No.103872641
>>103872619
nice self insert
nice self insert
Anonymous 01/12/25(Sun)20:50:51 No.103872654
Anonymous 01/12/25(Sun)20:51:01 No.103872655
Anonymous 01/12/25(Sun)20:52:20 No.103872666
>>103872655
I hope you get shot
I hope you get shot
Anonymous 01/12/25(Sun)20:52:56 No.103872671
>>103872666
kekd, samed, and checked
kekd, samed, and checked
Anonymous 01/12/25(Sun)20:55:57 No.103872691
Do you think we should give up the terminology war with "AI art" and start calling them "AI gens"?
Would doing this:
- Help normalize AI because it's less aggro to artists, it carves out a space for us.
OR
- Give them an inch and they'll take a mile, they'll see it as a win to further attack AI?
Would doing this:
- Help normalize AI because it's less aggro to artists, it carves out a space for us.
OR
- Give them an inch and they'll take a mile, they'll see it as a win to further attack AI?
Anonymous 01/12/25(Sun)20:56:30 No.103872697
Anonymous 01/12/25(Sun)20:56:33 No.103872698
This isn't a softcore fapping thread.
Anonymous 01/12/25(Sun)20:57:08 No.103872703
Anonymous 01/12/25(Sun)20:57:17 No.103872704
Anonymous 01/12/25(Sun)20:57:46 No.103872708
>>103872666
oops, this is better.
oops, this is better.
Anonymous 01/12/25(Sun)20:57:51 No.103872709
>>103872625
>You should know what the fuck your model can do before setting it out to the wild and not praying some autistic redditor figures it out for you.
I think they did know what their model could do but neglected to tell everyone. I can't think of another reason to distill other than prevent it from becoming "more dangerous" than what it already "is".
>>103872691
They're all just images.
>You should know what the fuck your model can do before setting it out to the wild and not praying some autistic redditor figures it out for you.
I think they did know what their model could do but neglected to tell everyone. I can't think of another reason to distill other than prevent it from becoming "more dangerous" than what it already "is".
>>103872691
They're all just images.
Anonymous 01/12/25(Sun)20:58:18 No.103872711
Anonymous 01/12/25(Sun)20:58:50 No.103872716
>>103872709
Then they get no play compared to other models. Same retarded behavior that made the previous company they worked for die.
Then they get no play compared to other models. Same retarded behavior that made the previous company they worked for die.
Anonymous 01/12/25(Sun)20:59:28 No.103872725
>>103872703
Calling it slop is what anti-ai people do. If you're pro-ai you can't call it slop and hope to normalize ai art, people will just see you as contradicting yourself.
It's basically like "white guilt" or some you've got going on here. You're ashamed of using AI?
Calling it slop is what anti-ai people do. If you're pro-ai you can't call it slop and hope to normalize ai art, people will just see you as contradicting yourself.
It's basically like "white guilt" or some you've got going on here. You're ashamed of using AI?
Anonymous 01/12/25(Sun)21:01:20 No.103872745
Anonymous 01/12/25(Sun)21:01:33 No.103872747
>Calling it shlock is what anti-movie people do. If you're pro-movie you can't call it shlock and hope to normalize film art, people will just see you as contradicting yourself.
>It's basically like "white guilt" or some you've got going on here. You're ashamed of enjoying movies?
>It's basically like "white guilt" or some you've got going on here. You're ashamed of enjoying movies?
Anonymous 01/12/25(Sun)21:03:02 No.103872754
>>103872747
I don't watch movies, I'm not gay.
I don't watch movies, I'm not gay.
Anonymous 01/12/25(Sun)21:03:08 No.103872755
>>103871949
Looks like the user used post-wall images for the dataset. Pretty gross desu
Looks like the user used post-wall images for the dataset. Pretty gross desu
Anonymous 01/12/25(Sun)21:04:54 No.103872767
>>103872754
that's not even close to the point
that's not even close to the point
Anonymous 01/12/25(Sun)21:05:30 No.103872771
>>103872402
For those who want to see the comfy's core equivalent to kijai's hunyuan node, here's the settings:;
FlowMatchDiscreteScheduler = Euler + Simple (Shift > 0)
SDE-DPMSolverMultistepScheduler = dpmpp_2m_sde + Simple (Shift > 0)
DPMSolverMultistepScheduler = dpmpp_2m + Simple (Shift > 0)
SASolverScheduler = Stochastic Adams Solver (not on ComfyUi?) + Simple (Shift > 0)
https://github.com/scxue/SA-Solver
UniPCMultistepScheduler = uni_pc_bh2 + Simple (Shift > 0)
For those who want to see the comfy's core equivalent to kijai's hunyuan node, here's the settings:;
FlowMatchDiscreteScheduler = Euler + Simple (Shift > 0)
SDE-DPMSolverMultistepScheduler = dpmpp_2m_sde + Simple (Shift > 0)
DPMSolverMultistepScheduler = dpmpp_2m + Simple (Shift > 0)
SASolverScheduler = Stochastic Adams Solver (not on ComfyUi?) + Simple (Shift > 0)
https://github.com/scxue/SA-Solver
UniPCMultistepScheduler = uni_pc_bh2 + Simple (Shift > 0)
Anonymous 01/12/25(Sun)21:07:06 No.103872779
>>103872725
I use AI but I still call it slop.
I use AI but I still call it slop.
Anonymous 01/12/25(Sun)21:07:13 No.103872780
>>103872725
>If you're pro-ai you can't call it slop and hope to normalize ai art
not at all, as long as it still looks like AI and not a real drawing, it'll be called slop, I'll stop calling them slop the day AI will be good enough you won't be able to differentiate with a real drawing
>If you're pro-ai you can't call it slop and hope to normalize ai art
not at all, as long as it still looks like AI and not a real drawing, it'll be called slop, I'll stop calling them slop the day AI will be good enough you won't be able to differentiate with a real drawing
Anonymous 01/12/25(Sun)21:07:43 No.103872785
>>103872771
thanks
thanks
Anonymous 01/12/25(Sun)21:08:35 No.103872791
>>103872747
Acting like schlock is a bad descriptor for a movie proves that you're unable to grasp the concept that sometimes accidentally awful films can carry some charm.
Acting like schlock is a bad descriptor for a movie proves that you're unable to grasp the concept that sometimes accidentally awful films can carry some charm.
Anonymous 01/12/25(Sun)21:09:17 No.103872797
>falling for whataboutism
disrespectful
disrespectful
Anonymous 01/12/25(Sun)21:09:39 No.103872801
Anonymous 01/12/25(Sun)21:09:43 No.103872802
Anonymous 01/12/25(Sun)21:09:43 No.103872803
>>103872771
He could have just labeled them right from the start, yeesh
He could have just labeled them right from the start, yeesh
Anonymous 01/12/25(Sun)21:10:22 No.103872805
>>103872803
kijai never does things like the others, that's his charm
kijai never does things like the others, that's his charm
Anonymous 01/12/25(Sun)21:10:49 No.103872806
>>103872802
I'm better than you.
I'm better than you.
Anonymous 01/12/25(Sun)21:11:42 No.103872814
>>103871405
Artists will always be retards. They were before AI and they will be until the end of time. A few people bought into it but most people don't care.
>>103871422
The ad looks kinda ass at times. That dog is stiff as a board. I think they could've done a better job with some scenes, though the glossy AI look suits their brand well with the polar bears and stuff.
One of the most catastrophic effects of AI is that people are accused of using AI all the time now when simply speaking with their own voice, drawing their own images, etc. I see this constantly on Youtube on small channels.
Artists will always be retards. They were before AI and they will be until the end of time. A few people bought into it but most people don't care.
>>103871422
The ad looks kinda ass at times. That dog is stiff as a board. I think they could've done a better job with some scenes, though the glossy AI look suits their brand well with the polar bears and stuff.
One of the most catastrophic effects of AI is that people are accused of using AI all the time now when simply speaking with their own voice, drawing their own images, etc. I see this constantly on Youtube on small channels.
Anonymous 01/12/25(Sun)21:11:48 No.103872815
>>103872767
The point is you are worthless, you do worthless things like watch movies. Pathetic.
The point is you are worthless, you do worthless things like watch movies. Pathetic.
Anonymous 01/12/25(Sun)21:11:58 No.103872818
Anonymous 01/12/25(Sun)21:13:00 No.103872828
>>103872814
>Artists will always be retards. They were before AI
this, when the printer was invented, those luddites wanted it gone, history repeats itself with AI
>Artists will always be retards. They were before AI
this, when the printer was invented, those luddites wanted it gone, history repeats itself with AI
Anonymous 01/12/25(Sun)21:14:09 No.103872837
Anonymous 01/12/25(Sun)21:15:20 No.103872843
>>103872815
prove I watch movies
prove I watch movies
Anonymous 01/12/25(Sun)21:15:31 No.103872845
>>103872828
Really? The printer, even?
Really? The printer, even?
Anonymous 01/12/25(Sun)21:16:45 No.103872857
>>103872771
>not on ComfyUi?
it's still an open PR
https://github.com/comfyanonymous/ComfyUI/pull/4454
you can try with this custom node thuogh
https://github.com/mira-6/comfyui-sasolver
>not on ComfyUi?
it's still an open PR
https://github.com/comfyanonymous/C
you can try with this custom node thuogh
https://github.com/mira-6/comfyui-s
Anonymous 01/12/25(Sun)21:20:57 No.103872887
>Token indices sequence length is longer than the specified maximum sequence length for this model (251 > 77). Running this sequence through the model will result in indexing errors
I thought Flux had more tokens?
I thought Flux had more tokens?
Anonymous 01/12/25(Sun)21:21:03 No.103872890
>>103872815
>The point is you are worthless, you do worthless things like watch movies. Pathetic.
You post on an image board in a thread specifically full of pathetic people. What does that make you?
>The point is you are worthless, you do worthless things like watch movies. Pathetic.
You post on an image board in a thread specifically full of pathetic people. What does that make you?
Anonymous 01/12/25(Sun)21:21:33 No.103872897
>>103872887
Flux still uses clip so once it reaches 77+, clip doesn't work anymore
Flux still uses clip so once it reaches 77+, clip doesn't work anymore
Anonymous 01/12/25(Sun)21:22:07 No.103872901
>>103872843
kekd hard
kekd hard
Anonymous 01/12/25(Sun)21:22:37 No.103872906
>>103872890
>in a thread specifically full of pathetic people
I'm not pathetic, my mom told me I was a good boi :(
>>103872843
oh he's good
>in a thread specifically full of pathetic people
I'm not pathetic, my mom told me I was a good boi :(
>>103872843
oh he's good
Anonymous 01/12/25(Sun)21:22:39 No.103872907
>>103872745
Very cool, thanks anon. I'll track down some of these LoRAs and see if it helps.
What's the purpose of the magic word "20250825v1d00ml0r4d" at the start of the prompt? Looks like a trigger word for a Doom LoRA, but none of the LoRAs I see being used sound like they're Doom-specific (aside from the db79... Doom LoRA, which uses a different trigger word and also isn't connected in the workflow)
Very cool, thanks anon. I'll track down some of these LoRAs and see if it helps.
What's the purpose of the magic word "20250825v1d00ml0r4d" at the start of the prompt? Looks like a trigger word for a Doom LoRA, but none of the LoRAs I see being used sound like they're Doom-specific (aside from the db79... Doom LoRA, which uses a different trigger word and also isn't connected in the workflow)
Anonymous 01/12/25(Sun)21:24:42 No.103872925
Anonymous 01/12/25(Sun)21:29:49 No.103872947
>>103872907
It's a lora of my own for Doom-esque graphics, ultimately it's not very good for pixel art but I enjoy using it for more scenic images. I'd recommend messing around with using several different loras.
It's a lora of my own for Doom-esque graphics, ultimately it's not very good for pixel art but I enjoy using it for more scenic images. I'd recommend messing around with using several different loras.
Anonymous 01/12/25(Sun)21:30:44 No.103872956
Anonymous 01/12/25(Sun)21:34:49 No.103872988
:skull: :skull: :skull:
Anonymous 01/12/25(Sun)21:35:19 No.103872993
Anonymous 01/12/25(Sun)21:36:37 No.103872999
:fire: :fire: :fire:
Anonymous 01/12/25(Sun)21:43:54 No.103873054
tried to proompt these dorks yugioh style eyes but couldnt manage it in one pass
Anonymous 01/12/25(Sun)21:43:55 No.103873055
>>103872887
clip-l has a 77 token limit, t5 has 512.
clip-l has a 77 token limit, t5 has 512.
Anonymous 01/12/25(Sun)21:48:07 No.103873088
Anonymous 01/12/25(Sun)21:56:36 No.103873161
Anonymous 01/12/25(Sun)21:59:35 No.103873185
Anonymous 01/12/25(Sun)22:00:45 No.103873196
>>103873185
Not a fan of the 1girl, but I like the background, what prompt?
Not a fan of the 1girl, but I like the background, what prompt?
Anonymous 01/12/25(Sun)22:02:34 No.103873208
>>103873055
ehhh? how do I force it to use t5 then?
ehhh? how do I force it to use t5 then?
Anonymous 01/12/25(Sun)22:04:07 No.103873232
>>103873208
t5 is always working, even when clip is shitting itself at 77+ tokens
t5 is always working, even when clip is shitting itself at 77+ tokens
Anonymous 01/12/25(Sun)22:04:44 No.103873242
why is it when I search sdg in the catalog it refers me to this shithole?
Anonymous 01/12/25(Sun)22:04:48 No.103873243
Anonymous 01/12/25(Sun)22:05:17 No.103873248
>>103873185
fan of the style, but i dont like the colosseum for ants
fan of the style, but i dont like the colosseum for ants
Anonymous 01/12/25(Sun)22:05:32 No.103873249
>>103873242
because searshing for a shithole refers you to all the shitholes?
because searshing for a shithole refers you to all the shitholes?
Anonymous 01/12/25(Sun)22:05:43 No.103873252
>>103873243
furks done it again that bastard man!
furks done it again that bastard man!
Anonymous 01/12/25(Sun)22:05:53 No.103873254
Anonymous 01/12/25(Sun)22:06:33 No.103873261
>>103873243
lmao that's amazing
lmao that's amazing
Anonymous 01/12/25(Sun)22:07:08 No.103873268
>>103873248
You can always inpaint huge black guys with huge dicks and blue hats that say Hillary 2016
You can always inpaint huge black guys with huge dicks and blue hats that say Hillary 2016
Anonymous 01/12/25(Sun)22:09:22 No.103873285
>>103873161
sdg best keep moving, this is a FLUX neighborhood.
sdg best keep moving, this is a FLUX neighborhood.
Anonymous 01/12/25(Sun)22:09:29 No.103873287
I need to stop genning and get back to training.
Anonymous 01/12/25(Sun)22:10:47 No.103873303
>>103873208
here, something like this. can just use a primitive node for the text input. the clip just goes through a LORA pipeline before it goes into that 'guidance pos' node, also no neg needed obviously if you run it at cfg1.
here, something like this. can just use a primitive node for the text input. the clip just goes through a LORA pipeline before it goes into that 'guidance pos' node, also no neg needed obviously if you run it at cfg1.
Anonymous 01/12/25(Sun)22:16:13 No.103873335
Anonymous 01/12/25(Sun)22:16:21 No.103873336
Anonymous 01/12/25(Sun)22:19:56 No.103873359
>>103873303
I've never seen that, how's it work?
I've never seen that, how's it work?
Anonymous 01/12/25(Sun)22:20:41 No.103873367
Anonymous 01/12/25(Sun)22:23:53 No.103873383
>>103873287
patiently awating your noobvpred family guy lora
patiently awating your noobvpred family guy lora
Anonymous 01/12/25(Sun)22:25:51 No.103873398
>>103873383
I don't think a vpred version would make much sense. I wanna try training a finetune of illustrious first.
I don't think a vpred version would make much sense. I wanna try training a finetune of illustrious first.
Anonymous 01/12/25(Sun)22:26:56 No.103873407
Anonymous 01/12/25(Sun)22:26:57 No.103873410
>>103873367
lora?
lora?
Anonymous 01/12/25(Sun)22:29:47 No.103873427
>>103873410
just noob vpred 1.0 and this shitmix lora
https://civitai.com/models/1068866/chadmix-finetune-extract-for-noobai-vpred-10
just noob vpred 1.0 and this shitmix lora
https://civitai.com/models/1068866/
Anonymous 01/12/25(Sun)22:30:15 No.103873433
dont pull bros, maskeditor is broken now, do these people test anything before pushing their commits? Its baffling
Anonymous 01/12/25(Sun)22:37:37 No.103873490
>follow guide to gen vids in 4-6 mins on 3060 with RAM offloading
>takes 20 mins per vid with exact same workflow and hardware
>takes 20 mins per vid with exact same workflow and hardware
Anonymous 01/12/25(Sun)22:39:43 No.103873505
>>103873490
furked
furked
Anonymous 01/12/25(Sun)22:40:35 No.103873511
>>103873505
kek
kek
Anonymous 01/12/25(Sun)22:40:53 No.103873518
>>103873427
thanks
thanks
Anonymous 01/12/25(Sun)22:41:06 No.103873519
>>103872204
catbox?
catbox?
Anonymous 01/12/25(Sun)22:42:20 No.103873531
>>103873519
I understand why (I don't think) he's yet to share. That style is so cool.
I understand why (I don't think) he's yet to share. That style is so cool.
Anonymous 01/12/25(Sun)22:42:46 No.103873534
>>103873490
This why I don't even bother with video. Takes way too long and all the online shit is leagues better.
This why I don't even bother with video. Takes way too long and all the online shit is leagues better.
Anonymous 01/12/25(Sun)22:43:12 No.103873536
Anonymous 01/12/25(Sun)22:43:42 No.103873538
>>103873534
Just gen 2d stuff for later i2v.
Just gen 2d stuff for later i2v.
Anonymous 01/12/25(Sun)22:44:33 No.103873546
>>103873536
box? impressive. very nice.
box? impressive. very nice.
Anonymous 01/12/25(Sun)22:49:54 No.103873600
Anonymous 01/12/25(Sun)23:01:34 No.103873718
Anonymous 01/12/25(Sun)23:38:11 No.103874012
>>103872612
Damn that looks great! catbox? Please and thank you
Damn that looks great! catbox? Please and thank you