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


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: ADT.jpg (2.72 MB, 3924x2065)
2.72 MB JPG
Previous: >>109726164

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

>Image Models
IL: https://rentry.org/illustrious_checkpoints
Anima: https://huggingface.co/circlestone-labs/Anima
Krea: https://huggingface.co/krea/Krea-2-Turbo
NovelAI (paid): https://novelai.net

>Video Models
H3: https://huggingface.co/Comfy-Org/MiniMax-H3
Turbo lora: included above

>How to generate anime:
image : https://rentry.org/comfyui_guide_1girl
video: https://rentry.org/wan22ldgguide (No H3 guide yet)

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

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

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

>/adt/ logo:
https://files.catbox.moe/pakfgt.png

>Beware inorganic activity
https://rentry.org/animanon
>>
*going to sleep,*
>>
File: output_20s_4mb_max.mp4 (3.64 MB, 1216x672)
3.64 MB
3.64 MB MP4
Expect a meltdown from a "anon"
>>
>>109740196
Lol
>>
Living the dream rn
>>
>>109740196
>>109740204
why do you embarrass yourself constantly? I am not entirely sure what you get from doing any of this
>>
>>109740286
>trying to peg others for what he's done
pathetic
>>
>>109740196
That's so funny, what anima is that?
>>
>>109740293
what did I do?
>>
>>
File: Saint Crispin.jpg (1.19 MB, 1920x1088)
1.19 MB JPG
Man, I love how picrel came out. GPT Image 2 ftw! Tried with Nano Banana (at twice the cost) and EzRemover first but both fucked up in different ways, the former even returning the output as-is, the fucker!

Source for the uninitiated:
https://www.youtube.com/watch?v=JRPdFFTjaFI

>>109740035
>Baker
You called?
https://www.youtube.com/watch?v=U-gsLFenB14
>>
>>109740652
*the input as is, i.e. not doing anything to it
>>
>>109740652
>>109740667
Read the subject of the thread retard
>>
>>109740820
Harsh but fair. Posted in the wrong tab, my bad. I'll leave you (fellow) weebs to it then o7

Becky sexo btw.
>>
https://civitai.red/models/2385278
>>
>>
>>109740838
>chibi EOM and Slaaneshi(?) xenos
Lore QRD?
>>
>>109740851
Unwilling chaos mutant used as a punching bag by all four factions, 40k would be the one universe where turbo regen is a curse instead of a buff
>>
>>109740870
>all four
Meaning? In my mind there is a whole bunch more than just four factions in W40k. Just the main ones are the Imperium of Man, Eldar, Chaos, Orkz, Nids, Necrons, ...

Also, that doesn't really explain why the EOM is ogling her like that while she's drooling, eyes (well, 2 out of 3 anyway) rolled up.
>>
>>109740883
All four Chaos factions, turned her world and people into contested so none could fully claim it. She's meant to be Erda reborn, trying to fix her fuck ups
>>
>>
>>109740894
They really need to shoehorn skaven into 40k eventually
>>
https://litter.catbox.moe/0ifo422lkov2acbk.png
>>
>>109740945
>>109730356
>>
>>109740950
>>>/h/hdg
>>
>>109740960
Based
>>
>>109740962
its completely dead..
>>
Why is Anima anti drugs? I just wanna see my oshi snort a big line of coke
>>
>>109741030
no e261. furries are the only ones that can afford coke in this economy
>>
well, I tried training a krea style lora. It was shit.
$25 here I go
>>
>>109741035
see >>109740834
>>
Don't worry localfriends, you'll get a nai5-tier model in 2028
>>
>>109741081
my artist mix always turn out like shit with these anima finetunes, but I'll try out this one
>>
>>109740945
>>109738787
>>
>>109740154
Need some prompt formatting help. I am using a template in comfy that can take up to 3 image inputs, run them through a +/- Quen text prompter, and render an image.

How do you format basic prompts for something like this? I just want it to do very basic stuff, like have the guy from image 1 standing behind the guy from image 2 with his sword raised up, ect. and maybe some basic facial expressions. It seems like the prompter doesn't understand what model I want to do which action. Its always getting mixed up and then starts adding really random shit that I never even asked for, or specifically said not to in the negative prompt.
>>
>>109741011
Know what isn't dead? /b/'s /degen/. Fucking WILD general, I tell ya hwat.
>>
>>109741162
>two guys straight non-stop autism spamming seed variants of singular prompts
C O O L
>>
>>109740154
>>109740702
Someone compared online video/audio generation to gambling and they're absolutely right. You start spending a few bucks to buy credits and then try 3,4 sometimes even 5 to 7 times to get a good output and quickly burn through the credits abd then you think you'll get a better output if you try again and but more credits abd before you know it, you've spend one tenth of our measly monthly payment to feed a cooming addiction. God I hate being a poorfag khhv.
>>
>>109741336
You shouldn't enter into slopping with the mindset of getting something exact and precise out of it. Expecting a happy surprise every time you hit the button is a good approach
But this is 4chan and we're all autistic as fuck here so if it doesn't match our mind's eye our psyches shatter and we chimp out on our keyboards
>>
>>109740716
Sadly, yes.
And sadly, that is the case for the majority of people in this hobby in general, and on this website.
>>
>>109741336
In this hobby, they call it "gacha rolling," and sadly, a lot of lonely men need to fill the void in their lives with something. Local or not, you will end up involved in some way that requires you to spend real money. NovelAI addicts, GPU addicts, GPUlets who rent GPUs to train poorly baked models, or subscribe to Claude Code to enhance your ComfyUI workflow. Lonely men are just as consumerist as an upper class American girl.
>>
>>109741134
What you're describing is an edit model issue. Edit models are still a new addition to this hobby and remain quite unstable. Your best bet is to check the cookbook for the specific model you're using, copy the sections that cover prompting guidelines, and feed that to your trusted AI, asking it to rewrite your prompt idea using the lexicon that model was trained on. If you're working with edit LoRAs, you'll also need to check what lexicon the trainer used and which editing tasks the LoRA was actually trained for.
>>
>>109741334
I see you've never actually been in it. Probably best you don't start to either, actually.
>>
Why is Anima anti scat?
>>
>>109740716
if you look at other places of the internet it is very easy to come to the conclusion that normalfags should not have access to AI because they will make the most mundane shit imaginable
>>
>>109741574
Because its creators hate black people. There, I said it.

Anti-scat is anti-brown, run these racists out of town!
>>
>>109741575
>NOOOOOOOOOOOO, imagine boomers creating slop with the slop machine, the horror!
Who cares, anon? The minor annoyance of their stupid outputs is massively outweighed by their value to the relevant companies as consoooomers. Without them to market to/replace this tech wouldn't be pushed as hard as it is.
>>
>>109741039
>>109741575
Sad truth. Those people will tell you model X is great for anime and styles, "just train a lora bro," "just curate a dataset bro," "skill issue bro." And since you're new, you believe them. But then a new model drops and suddenly everyone starts complaining about the exact same things you were told to shut up about. That's when you realize this hobby runs on half truths
>>
>>109740838
Cute
>>
>>109740894
Am I missing something, anon, or does that still not address
>Also, that doesn't really explain why the EOM is ogling her like that while she's drooling, eyes (well, 2 out of 3 anyway) rolled up.
>>
>>109740196
lol
was that yoland defending comfy? kek
>>
>>109740028
yes: houtengeki
>>
>>109740838
Curious question though, do you use NAI because Anima just doesn't have Warhammer character knoweldge? From what I've seen, Anima's pretty weak with Mortal Kombat or Street Fighter characters too.
>>
File: big head.jpg (354 KB, 896x1152)
354 KB JPG
>>
>>109741336
AI is a hobby, anon and hobbies exist for socializing, laughing, and making friends. If you're the kind of anon who needs shippable results to feel okay, this isn't your scene yet because right now the field runs on speculation and vibes.

A good rule of thumb is, whenever you see a shiny new release number like Fable 5, GPT6, or NAI5, slide the decimal point over, Fable 0.5. GPT 0.6. NAI 0.5 and the hype settles down to something a lot closer to the truth.
>>
>>109741863
Was for him >>109741628
>>
>>109741856
Big head <:3
Small pussy >:3
>>
>page 3
Where is /adt/ engagement squad? :ɛ
>>
I fucking hate when Pixiv takes down my stuff for being not censored enough.
>>
>>109742055
But you use NAI tho, thought?
>>
>>109742056
He meant pixiv, not pixai, silly nonnie.
>>
>>109742055
This is adorable, mind sharing your Pixiv??
>>
>>109741744
ty
>>
I want to bake an Anina LoRA I made, but in Illustrious this time. Problem is I lost my original SDXL LoRA settings. If I recall correctly it was 300 images, batch size 16, 2800 steps total. Does anyone still have good Illustrious LoRA settings for that kind of dataset size?
>>
>>109741162
Oh no, /r9k/'s diffusion general also doesn't exist anymore :(
>>
File: gemini_3.1_image_00004_.png (1.17 MB, 1024x1024)
1.17 MB PNG
From the previous thread, with regards to krea2 turbo vs base, it's like a repeat of z-image in a way, but at least the underbaked main model is large enough to handle a big lora without collapsing or forgetting. I trained my danbooru lora at lr: 5.0e-05 at 15000 steps - which doesn't sound like much, but it takes quite a long time to get through each step when there's 40K images.

I see openrouter has openai-5.4-image-2, so here's an example from it. I'd do something more lewd but I'm pretty sure there's refusals even on the API side, right?
>>
>>109742055
Isn't Pixiv oversaturated with AI slop already? Genuinely wonder what's harder nowadays, make it as a youtuber or makin it as an AI artist.
>>
>>109742559
>openai-5.4-image-2
Is this a new separate model from gpt image 2, or the same thing renamed?
>m pretty sure there's refusals even on the API side
Yeah, confirmed. I tried curating my anime dataset using gpt image through API and still got refusals. Fewer than when using the website directly, but they're still there.
>>
>>109742612
You can make a brand new twitter account and get a post to break >10k likes in less than 2 weeks with no pre-planned external support(and no paid botting) if you know what you're doing. And no, I don't mean "could," I mean CAN. And once you start gaining momentum it's pretty easy to keep it rolling as long as you keep pace and make a post every 2-3 days. Though this all assumes you're posting things of quality that people actually want to see.
>>
>>109742559
Why is Teto obese?
>>
File: Krea2_turbo_00338_.png (1.38 MB, 1448x1448)
1.38 MB PNG
>>109742696
Someone made a image like this in anime, I was just trying to copy the style.

Can you make anima do a Migu like this? I'm trying to build up a collection of animations which I can covert to ILDA output, being able to do line drawings like this is a very nice headstart. They get simplified at the png to svg step - basically turned to outlines with the eyes and mouth fully-drawn.
>>
>>109742559
Not trying to be a dick, but if you're the anon with 48GB VRAM, let me ask you something. Why are you wasting all that power and compute on this? Tdrusell himself said a real anime finetune needs like 20k-100k in funding.
With your setup you could actually be training:
-An Anima Edit finetune using the noob 2 edit dataset (gated on his HuggingFace, but he shares the link)
-Anima ControlNet
-Anima IPAdapter
-Anima style transfer
I kept quiet till now, but every time I see your posts I wonder why so much GPU power out there just gets wasted on dumb stuff like what you're doing right now.
Nobody asked for a Krea 2 NSFW finetune when Anima already does that. Nobody asked for a Z Image finetune either. Yeah I know it's your GPU, your money, your time, but you know this isn't a real improvement and it's not helping anyone. We already have plenty of good paths open with these new models. No point opening new ones when we all know literally any model has "potential" for something.c
If you're willing to do any of the kinds of things I pointed out, I wouldn't mind reaching out and asking LAX for access to his Noob 2 edit dataset, or the Xinsir ControlNet dataset, or whatever is necessary to get a specific dataset for some of these things I mentioned. Put your name or /ldg/ name to it, I don't care, but dude, do something useful for once.
>>
Hello, new to this thread. I was wondering if anyone was working on a Krea 2 NSFW finetune? It's clearly leagues above other local base models like Z-Image and Cosmos 2. I've been browsing CivitAI and the stuff people make with Krea 2 is already amazing. Really all it needs is a bit of danbooru finetuning to push local forward.
>>
File: Krea2_turbo_00351_.png (1.42 MB, 1448x1448)
1.42 MB PNG
>>109742752
Nah that's not being a dick, those are valid points. The main thing which captured my interest in krea2 is it does things you either can't do in other models, or which take a complex workflow. The secondary thing I wanted to see if a big lora could somewhat break the safetycucking - it does. Same reason why I have a bunch of ID4 loras, even though everyone hates ID4 - it does a few things you can't do in other models.
I totally agree, you can't do a finetune at home (unless you're a Saudi Prince). I wanted to see how far you could go on a single gpu.
Is it dumb? I guess it is if you're happy with Anima. I haven't played with it so I don't talk about it, other than I appreciate how hard it is to do something at that scale. Sucks that cloud keeps pulling away from local, the bones they toss us are increasingly cucked, and expectations for image gen push the hardware requirements further and further from what even a medium-size corp can fund.
Finally, I think there's a few anons in /lmg who have 6000 pros. I missed the boat on that, I waitfagged and now they're Rolex-tier.
>>
File: anime girl.jpg (451 KB, 832x1216)
451 KB JPG
>>
>>109742055
I thought they recently became less draconian and allowed cameltoe/cleft as long as it was low detail
>>109741715
Giving her a minute to 'catch her breath' (since her IQ flatlined)
>>109741806
ye it's v5
>>
>>109743448
But why's she in his presence anyway? Thought she's from some four-way-fucked Chaos world.
>>
File: kneeling-migu1.mp4 (1.79 MB, 640x640)
1.79 MB
1.79 MB MP4
Damn minimax-h3 does such a good job with generating coherent spatial stuff if only I could stop it from shading things. It wasn't even a year ago I was using cosmos 7b and genning stuff where limbs would fall off and bodies would melt.
>>
File: Keikaku doori.jpg (472 KB, 1216x832)
472 KB JPG
>>109743495
I'm actually working on a comic about it, first part is canon events, second part is about Donut Steele lol
>>
>>109743669
Well, you better make sure to upload it somewhere once it's finished, anon, so you can just link to it when asked about it rather than dumping it page by page each time.
>>
>>>109742802
Shut up Ani! This is not fun! Nobody will finetune Krea. Anima is our last model before the APIcalypse
>>
File: Krea2_turbo_00382_.png (1.64 MB, 1448x1448)
1.64 MB PNG
>>109743857
It's uncuck vs unfuck. Do you unfuck cosmos or uncuck krea2?
>>
>>109743857
Ani is right about everything newfag
>>
>>109743684
Absolutely, still touching it up
>>
File: file.png (880 KB, 1015x627)
880 KB PNG
>>109743969
except about how to behave irl in a way that doesn't get you banned from physical spaces apparently
>>
>>109743996
>double akimbo bolters wielded by a quasi-immortal
Enemies of the Imperium of Man beware! She's comin' for ya!
>>
>>109743646
thats a big butt
>>
>>109743684
>>109743669
This I want to read it too she's adorable
>>
>>109741716
Yes
>>
>>109744043
what the fuck are you talking about now schizo?
>>
File: file.png (107 KB, 466x530)
107 KB PNG
>>109744337
glad you asked so more newfrens can know what an irl psycho you are
>>
>>109744354
didn't you post a vid where comfy explains ani isn't actually banned?
>>
>>109744361
notice how you don't deny you said you would "get rid of yoland" and that you are "not allowed near yoland"?
>>
File: Capture.jpg (182 KB, 2486x1801)
182 KB JPG
>>109744361
it's here >>109681847
catjack is just a delusional retard nobody should listen to. now get back on topic
>>
>>109744408
>you can go near him, I just don't think it's a good idea
Yeah, you're banned, you sperg. Learn to take a hint.
>>
>>109744413
>you can go near him
>Yeah, you're banned
it's one or the other illiterate freak. go make more seethe videos nobody wants to see
>>
>>109744337
>>109744361
>>109744361
>>109744408
that's a lot of effort and energy you're putting into cocksucking some avatarfag that is totally not you btw
>>
>schizotrooning out
I just want to know how to create a lora for this shit.
>>
File: Return_00501_.jpg (1.42 MB, 1776x2368)
1.42 MB JPG
>>109744408
How's that career going ani?
Why do you keep calling for me when I have nothing to do with this current spat?
Now I'm going to post images and you're going to cuck in the corner because even your schizo rentry falls to the dirt and ages poorly with the recent revelations of your desires to service endowed "women" with 12 inch dicks
>>
>>109744427
this isn't about defending ani, this is about warning people you are an unhinged faggot that has been doing this shit 24/7 for years. GET HELP NOW
>>
>>109744449
do keep going, your desperate self-whiteknighting just tanks your reputation further.
>>
Catjack is the worst schizo in all the diffusion generals...
>>
>>109744444
>How's that career going ani?
Hasn't had a job since getting booted from the Comfy Org founding team. Now he begs money from them like they forgot he said he was going to "get rid of" their CEO. lol
and samefags on 4chan full time without pay
>>
>>109744444
Hello, rentry.org/ranfaggot , again derailing our anime thread? You are seriously making more people to wake up who you really are
>>
yes! post your demented rentry! i still can't believe you don't realize that you just advertise how mentally ill you are when you do lol
>>
>>109744470
Catjak is a disgrace
>>
File: Krea2_turbo_02571_.jpg (1.63 MB, 1776x2368)
1.63 MB JPG
>>109744475
I know and now the guy that doxxed himself and continues to post the most embarrassing shit non stop because he has wetbrain. Now he's going to do the projection routine.
>>109744470
Who has a rentry with verifiable proof of schizo behavior?
>>109744511
This doesn't help you it just ages poorly now that your /d/ post are logged.
He keeps doing kamikaze play without hitting the target
>>
we support ani here. no matter how many of you come in to derail our general with your anti ani bullshit, ani is very appreciated here. we accept his past, and we accept anistudio as a work in progress ui
>>
>>109744590
ah yes that post isn't suspicious at all LMAO
very organic
>>
>>109744585
What's with your extreme cuckolding obsession? That's just really weird, dude. Stop blaming Ani for your own problems.
>>
>>109744590
This guy just stokes the fire to make him crash out even more.
I don't know a single place where he's not treated like the special needs time bomb that he is
>>
File: Krea2_turbo_01841_.jpg (1.79 MB, 2368x1776)
1.79 MB JPG
>>109744601
You see cuckholding out of that image?
That says a ton about you, the point is despite not being racist and a member of the LGBT he defaults to newfag 4chan insults that hold no water and contradict his leaf nature.
I think he's a troubled man that refuses to confront his problems and doubles down.
Also where are your gens ani defender?
>>
>>109744590
Retard
>>
>>109744660
Agreed, it's incredible how he can post that with a straight face and not realize it just makes him look retarded.
>>
>>109744667
I think anons gas him up to make him crash out more
>>
>>109744585
>>109744618
is this the place to troon out talking about drama nobody cares about? return to your discord circlejerk immediately and seethe there. leave us in peace
>>
>>109744590
It's really easy to spot you, Julien
>>
>>109744745
Who's us?
>>
File: kotomi-rotation1.mp4 (1.78 MB, 640x640)
1.78 MB
1.78 MB MP4
A smoothly-looping animation of a cute anime girl drawn in a vector plot style, slowly rotating 360 degrees. The image remains drawn in a pure un-shaded green line style, and the background remains pure black. An hard techno beat plays in the background.
>>
>>109744804
Sorry, I assumed you had friends but it makes sense nobody likes you
>>
>>109744820
Read the room
>>
>>109744833
Multiple anons telling you to fuck off and some really petty self replies to encourage more schizo posting from yourself. What do I win?
>>
>>109744833
Says "מנא מנא תקל ופרסין". Huh. Anyone make heads or tails of that?
>>
File: concept34445.mp4 (2.56 MB, 1376x768)
2.56 MB
2.56 MB MP4
He's having a full on melty and zero gens posted. Just the fact I breathe is enough to put him in a rage. You have been doing this and failing for almost 2 years now, you don't even show your face anywhere because people are tired of you and your mental illness.
Either post a gen or go back to talking about how much you want to drink cum in thr /d/ thread
>>
>>109744833
Anon, this is 4chan, the closest thing to that here is "reeeee the room".
>>
File: Krea2_turbo_00389_.png (1.93 MB, 1448x1448)
1.93 MB PNG
krea2 has definitely seen some blender wireframe stuff. This is pretty damn good despite is being quadrangles and not triangles. Is there a 3D AI model yet which understands how to put clothing onto characters and properly bone hair and other dangly stuff?
>>
>>109743684
this
i at this point have a collection of her saved, but i only go on /g/ and /e/ so possibly it's incomplete
>>
>>109744888
so it is some sort of fetish for you
>>
I miss the anon that used to post
>ani is crying
He never stopped crying
>>
>>109744991
where is ani is the bigger question
>>
>>109745063
All over this thread, cocksucking and whiteknighting yourself.
>>
>>109745076
I didn't know every anonymous poster is just ani. Thanks for informing me
>>
>>109745076
He still can't accept the fact the rentry is the most damning piece of evidence about his actions. If ani is so great why do the so called defenders never post a gen from his UI?
If he's so great why are anons afraid to attach a gen while defending him
If he's such a valid goy why can we find him repeating lies he post here verbatim in other threads where he post his avatar.
>>
File: 1759017770703571.png (152 KB, 582x431)
152 KB PNG
>>109742690
The fuck are you talking about, my posts literally all have single digit views
Not likes, views
>>
>>109743996
Please do more of her in Sister of Battle power armor in non-chibi form.
>>
>>109744427
>>109744449
being an unhinged faggot to indicate to others what an unhinged faggot is
hmmm curious, many such cases.
>>
File: Krea2_turbo_00429_.png (1.63 MB, 1448x1448)
1.63 MB PNG
Time to call it a day!
explicit version:
https://files.catbox.moe/gt1vw2.png
>>
>>109745136
Simply pointing to OP invalidates his claims and thus his narcissism loop short circuits and he crashes the fuck out.
>>
Destroy /adt/ all you want faggots, but I will resurrect it from the ashes as many times as necessary. This is the only general I browse on this shitty website, and I’m not going anywhere.
>>
>>109745171
Just need the schizo dev to get occupied with something else.
>>
my favorite thing to do on pixai is to do a hi res upscale and it adds extra fingers
>>
File: teto coffee.jpg (564 KB, 1152x1344)
564 KB JPG
>>
>>109745182
My favorite thing is to go to /adt/ and tell the schizo I enjoy his gens and ask for more so he keeps staying there.
Whoops wrong thread
>>
thanks for bakering
>>
>>109742274
I post cute and lewds in different accounts though
>>
Going back to illustrious because of bondage loras :(
>>
File: HRUJzAkWUAY-UM8.png (918 KB, 743x731)
918 KB PNG
I need to learn how to make good enough porn of my niche fetishes so I can pay for college like 500 bucks a month
>>
>>109745935
Peruvianon?
>>
>>109745935
Based PeruGOD
Drinking mate Loras when?
>>
>>109743309
Woa
>>
>>109743646
Minimax is sota edit model
>>
>>109746151
Can Anima or NAI do a "cutting Coca Cola plastic bottle to make a fernet cup" concept LoRA?
Can Anima or NAI make anime girls drinking mate?
Can Anima or NAI make an barbecue cooked in a cut in half oil drum?

Why can't local APIs and local models handle simple third world concepts like this?

Where is my third world lora concept?
>>
Going back to illustrious because of out of the box scat :(
>>
>>109746167
anime girls are great, don't you agree?
>>
>>109746222
Isn't Pablo Escobar into anime gooning? Why not launder money by making a local anime model?
>>
>>109746348
If you can rent gpu with crypto is a good idea
>>
>>109746357
openrouter accepts crypto but there's nothing in the way of uncucked imagegen.
>>
>>109745130
have you considered it's a skill issue? an issue of skill?
it took like a week for twitter's anti-bot checks to get passed and for it to actually start serving to people and to get views at all. Once you actually start getting content served to people all it takes is one of them to give it a like(not even an RT) for it to start getting served to more people, of which will be in the same category of user as the people that hit the like button. As long as you focus on the right type of characters/the right type of content you'll pick up traction. I know this because I've done this very recently. Just make a single post each day until something goes hard, let it run for two days then make sure the next post is actually compelling and the prior success will compound and ride a 2-3 day schedule after.
oh and if you couldn't tell, the "right type of content" is of popular characters with light ero touches or weird "safe" fetishization. I don't make the rules.
>>
>>109746222
I'm pretty sure you can achieve "coca bottle cut in half as fernet cup" with good enough prompting, krea should be capable enough
>>
>>109744812
cute
>>
File: Purple potato (Hopeful).jpg (275 KB, 832x1216)
275 KB JPG
>>109745132
>>
NAI won
>>
>>109747405
Never seen this race in Dawn of War 1 despite having played it. You're saying it's a slave race under Chaos?
>>
>>109747491
Why? Can Krea 2 not do this?
>>
>>109745195
Prompt/Catbox?
>>
>>109747508
Slaanesh weighted transformation OC race (sea life, suckers, tentacles). Slaaneshis used them as onaholes
Khornites used them as punching bags
Nurgle cursed them with regen on top of their now "natural" regen. Then tested plagues and diseases on them
Birdboi gave them the third eye (which only screws with their depth perception, making them easier prey), glowing blue blood (making them easier to track while injured), and sat back while basking in the constant conflict
>>
>>109747537
Nah style figting 1girl and pray for the style and look gets assimilated
>>
>>109747586
Interesting, is there any W40k novel that talks about them?
>>
>>109747537
No, training a model from scratch is prohibitively expensive and new base models contain some flaw that makes them a hard no for anyone considering financial investment in fine tuning but Anima bought some breathing room.
>>
>>109747537
>forced to boomerprompt
>safetycucked
just train a lora
>>
>>109747635
main rulebooks
https://warhammer40k.fandom.com/wiki/Mutants
kinda paraphrases it
just being near a rift can screw over entire militias
all four chaos factions dogpiling one planet would straight fuck most of the population
easy explanation would be the blue bird of suicidal thoughts pointing out a planet near a rift to the other three daemon gods and everyone fighting to get through the door at the same time lol
>>
can i take a png without metadata and add the prompt to it easily with comfy? what do i need? the image was already generated, it just doesn't have the metadata
>>
>>109747873
Yes, ComfyUI can add metadata to a pre-existing PNG if you load the image into a workflow and use custom nodes or save it out again with metadata-handling extensions.
Load the image: Use a standard Load Image node to bring your pre-existing PNG onto the ComfyUI canvas.
Use custom metadata nodes: Install utility packs like ComfyUI-PNG-Metadata via ComfyUI Manager to inject custom key-value pairs or AUTOMATIC1111-compatible parameter strings
Save the file: Connect your image and metadata nodes into a Save Image node so ComfyUI writes the new text chunks directly into the exported PNG file
I asked googai and that's what it coughed back at me
Can also just use Photostop File-File Info and plug it in there
>>
>>109747744
Okay, so you say "just train a lora." Then I run into the problem:
>2023 era taggers, and even the newer ones like TiMM or PixAI taggers hallucinate constantly and just aren't reliable for tagging at scale
Or, well, now there's the "miracle" of natural language captioning, so I figured that'd help, except:
>local VLMs are trash tier and safety cucked. The few that claim to be obliterated are poorly trained and still have gaps where censorship leaks through
Oh well, corpo...
>corpo LLMs are little better, sure, but they're expensive and extremely censored
So no, it's not as easy as "just train a lora."
>>
anyone got retarded prompts like
>showering, riding motorcycle
>flying, swimming
>bowling ball, blood
?
>>
>>109748375
Majority of them, split the gen in two scenes
>>
>>109748414
*copy and paste*
>>
>>109743929
You sound like someone who has never inpainted in his life.
>>
>>109748582
>>109748375
Test with >>109746222
>>
>>109748375
One of these is not like the other, dududuu~
>>
>>109747586
Poor girl! :(
>>
Happy Monday /adt/ :3
>>
>>109750100
>Happy
I hate Ma-ma-mondays.
>>
>>109750149
Luckily, our leader Ani doesn't need to work
>>
File: 220504195044.jpg (19 KB, 404x356)
19 KB JPG
https://litter.catbox.moe/bh9hhdjcmwix8mde.png
https://litter.catbox.moe/18bxbzhxorkqbejg.png
>>
>>109750469
Been on medical for about three weeks now and it happened to line up almost exactly with the NAI5 release, couldn't have timed it better desu.
>>109750483
Nice, are you him? >>109745562 very similar style
>>
>>109750499
All the good stuff are mine...
>>
>>109750149
I hope you stub your toe nonners
>>
>>109750517
Cute~
>>
>>109750469
Got promoted, did he?

>>109750483
Not bad. /degen/ might like it.

>>109750525
>nonners
Nigga, what?
>>
>>109747744
I have eyes, anon. I can see what a LoRA gives you and what it costs you. You lose plasticity, you lose versatility, it's not the same as a base model that already has the knowledge trained in from the start.
>>
>>109750700
Not sure how vram friendly this would be, but is there such a thing as regional lora for anima or sdxl?
>>
>>109750713
I'm guessing maybe it doesn't work? In ComfyUI, the LoRA is after the checkpoint and before the KSampler. The prompt is separate from that, it's more connected to the text encoder.
>>
File: makimayamada.png (1.83 MB, 1106x1422)
1.83 MB PNG
Makima or Tayama?
>>
/ani hours/
>>
>>109750469
Why would a dipshit avatarfag be our leader? Why would a general have a leader?
>>
>>109750483
Thanks for sharing <3 I don't go there but I know /u/ has its own diffusion thread too so you can share there also, I don't remember if it is sfw or nsfw tho.
>>
>>109750699
>Got promoted, did he?
promoted from pre-org comfy contributor to unemployed e-beggar due to his behavior, yeah
>>
>>109750809
it's just thread culture, anon
/ldg/ - catjak rentry.org/ranfaggot
/sdg/ - debo rentry.org/debo
/adt/ - ani rentry.org/animanon
>>
>>109750713
There is
>>109750764
I created this using anima + Regional lora
https://civitai.red/images/142069248
>>
>>109750831
Holy cringe.
>>
>>109750831
Could we at least choose our own schizo instead of it being forced on us? :(
>>
>>109750855
No, this shizo thing is not new, primordial anons decided this a long time ago, It's like the masons or the illuminati.
>>
>>109750764
>>109750843
Cool, thanks
>>
>no cris the spic
>>
>>109750855
Tamzay has schizo potential for /adt/, he was there from the start and single handedly bumped the thread through the great /adt/ depression.
>>
>>109750764
'kima all day.
>>
>>109750920
You know the drill
>>
>>109750909
Nah, it doesn't have any memorable anecdote worth putting in a rentry, Ani beats him by a mile.
>>
>>109750930
I actually don't, honkler. Say more.
>>
>>109750713
Forge Neo has a regional prompter that Haoming02 himself maintains, but I'm not sure if it supports regional loras.
>>
>>109750920
sexy
>>
>>109750997
Ain't she!? Shame for y'all that Astolfanon got bullied out and now won't share (well, not really, just spending his time more fruitfully, teehee).
>>
>>109748370
I used minimax-m3 via openrouter to caption my danbooru 1024 eq 40K dataset. I had my local qwen 3.8 27b LLM agent talk to it and figure out what the prompt should be to get it to fully, explicitly caption sex scenes. So now I have a script which can caption a 1024x1024 image in about 10 seconds, or call it 3 if you consider I'm batching four requests at a time to openrouter.
As for accuracy, I used the danbooru 2026 tag set so I could tell minimax exactly who was in each image and what they should look like, so minimax did a very accurate job of describing what it saw each character doing.

>>109748665
Nigger please, I've been doing this shit since stylegan. Of course I've inpainted and did regional prompting and smart detailers etc... if you still enjoy that, then good for you, but I prefer to use models which can use complex NLP, or hell, even json, if ID4 wasn't cucked garbage.
>>
File: 00002-1874262534.png (2.14 MB, 1344x1920)
2.14 MB PNG
>>
>>109740154
Im downloading and installing Krita plugin for making gens with Illustrious you fags better be right that this is the way. I also dont see support in it for Anima only illu/noobai.
Anyway any advic
>>
>>109751176
Wouldn't grok be better for the coom material?
>>
>>109751276
Krita isn't necessary. It was built by a dev, not a genner, so a lot of the settings feel awkward. Things like checkpoint, sampler, cfg, steps, and quality tags all get buried inside something called "style," and if you want to switch samplers or schedulers you have to go edit a json file. There's a 2ch anon who made a custom fork that supposedly fixes most of this, but he doesn't share it. Really, Krita wasn't built for someone coming from genning, it was built for a novice or casual drawer who wants to get into diffusion.
Now, for that kind of inpaint focused workflow, since you're already a genner I'd recommend Invoke first. It supports Illustrious, NoobAI, Anima, and other models. Next would be SwarmUI, whose canvas is less advanced than Invoke's but runs on a ComfyUI backend, so memory management is safer and more reliable. And last but not least, the good old reliable Forge Neo.
>>
>game cg now is called game asset
Fuck, time to edit all my dataset :(
>>
>>109751437
In terms of multimodal vision? I can't stand grok's tryhard flippant redditor tone, it's the last place I'd look for LLM stuff.
>>
>>
>>109751557
Forgot to add, the annoying thing about Invoke is that its seed, LoRA strength, and tag weight values don't carry over to Comfy or Forge. It's its own closed ecosystem in that sense. But then again, Krita's seed can't be replicated in Comfy or Forge either.
>>
>>109751575
I really do hate danbooru fucking with the tags all the time. Not to mention the shit they pulled with the oppai_loli tag until Ilulu showed up
>>
Who needs 32ch VAE?
https://civitai.red/models/2767064/anima-detailer
>>
>>109750831
Funny how the /ldg/ one comes off as written by someone who's butthurt.
>>
>>109752024
>germanic architecture
>nippon clothes
cuppa nani
>>
If anima had perfect text there would be no need for us to push krea2. Is it really that hard for the creator to upgrade the text encoder?
>>
>>109751575
There are Danbooru tag manager GUIs that do batch regex replace on all your txt. files, anon.
>>
>>109748674
but I was going for retardation, not non sequitur split screens
>>
>>109751872
>the shit they pulled with the oppai_loli tag
explain
>>
File: 1787837532121818.png (69 KB, 795x254)
69 KB PNG
>>109752383
Doubt that actually happened, at least not from Circlestone Labs themselves, Anima was abandoned due to lack of funding.
>>
>>109752782
Not using a better text encoder is exactly why it's no longer a good candidate for support when new models like krea 2 exist. Why invest in something that crippled itself from the start?
Why advertise text if it's so fucking shit?
>>
>>109752647
Went on a warpath against oppai_loli and shortstack type bodytypes and manually removed the tags from thousands of images
Literally one year later, Ilulu detonated the interet and Danbooru consneeded
>>
>>109752581
>so slick she can't even hold onto the handlebars
>>
>109752782
Don't you get tired of pulling the same bullshit every time?
>>
>>109752858
He's severely mentally ill and is trying to defend his name in the other general too. I don't get the point of it all when all anon has to do is read his rentry and it's all null and void.
>>
>>109752818
Also if you are training Anima you need text captions in there or watch it catastrophically forget TM everything like a goldfish
>>
>>109752900
You need better catchphrases, you're such a miserable pile of shit.
So fucking bitter even when anons say anima is done for due to newer models you still need to spread misinfo and shit your pants.
>>
>>109752782
Seek help, touch grass. still using Anima lol
>>
>>109752922
What makes it funny is he seethes both here and /ldg/ lies constantly and is still confused on why his rentry is updated
>>
>>109752933
I get why he visits /ldg/ given his emotional drama with Comfy, but I don't get why he comes here. Does he also visit /sdg/ or /de3/?
>>
>>109751688
Did the meta tag prefix actually do anything, or is it just snake oil? Does the style feel more "official" to you with it?
>>
>>109753031
self-promotion.
>>
>>109753031
/sdg/ is too tarnished for him so /ldg/ gets bombarded with both the schizo that /sdg/ was given to.
He got chased out of multiple threads when he was an attack dog for comfy and pretty much settles in /d/ now which lead to him embarrassing himself even more in the rentry.
They tend to outright ignore him there which imo might be worse for his mental health.
>>
>>109752782
>until I have the funds for
your screenshot literally says he wants to keep training models but needs to wait for some money to come in first. which is understandable, given he spend over 50 grand out of pocket making anima
>>
>>109753410
Here's what's gonna happen when he gets enough money from royalties to be able to train more models:
>man it sure is a nice feeling having all this money
>>
>>109753473
if he can be net profitable on anima, why would he stop there, and risk get surpassed by future models? you would obviously roll that profit over into training an anima2, or a realism model, or whatever
>>
>>109753519
Not obviously. Think pdxlv6 vs pdxlv7.
You can just sink money and FAIL miserably.
>>
>>109753530
the anon I'm responding to is suggesting tdrussell will just stop completely and never even attempt to train another model
>>
>>109753530
anyone with a half functioning brain could've told you pdv7 was going to be an absolute failure before it even started, though.
Dumb as fuck architecture choice, chosen specifically to prevent people from running it locally unless they had a xx90 class card, because he wanted to make money off of SaaS.
Dataset was made worse in every way he could possibly have made it worse due to him completely misunderstanding what people actually want from diffusion models.
And also the guy is just plainly a full on retard who doesn't know jack shit about how to train, with pdv6 being a cosmic bit flip tier rarity in being "functionally" overfit.
The worst part though is that I'm pretty sure he's still wildly ahead in money earned vs money spent due to how much he got out of pdv6.
>>
>>109753592
He also destroyed Chroma in the process which baffles my mind why people even trust or care about the faggot that made that model when he let that retard of all people influence his decisions
>>
>>109753578
It's very probable. In fact I'm betting on him getting cold feet now that comfy will not be funding him further.
When he gets the money he'll just say forget it and keep the money.
But I'm all for getting pleasantly surprised and being wrong.
>>
Suport Ani
>>
Fart on trani
>>
File: Krea2_turbo_03889_.jpg (1.52 MB, 2368x1776)
1.52 MB JPG
>>109753772
Supporting him is helping him leave this place so that he can get the help he needs.
Check out this fusion of styles
>>
>>109747571
>1girl, solo, (cowboy shot), (@andro juniarto, @maruko tooto:0.5), kasane teto \(sv\), white shirt, black leather jacket, jeans, walking, (plump:0.75), curvy, looking ahead, holding disposable coffee cup, sidewalk, outdoors
>>
>>109751557
Really? You are not trolling me??
At first Krita seems to be good though, at least the default gens seem ok, but I kinda get what you're saying about the settings being buried.

Im gonna try invoke later then. Im just used to ComfyUI, but I wanted a good inpainting workflow Comfy tools suck.
I had always assumed Comfy was the best of the best for power users, was I wrong? IsForge Neo better?
>>
>>109753984
Comfy is not a good tool if your intention is to just gen and inpaint. If you're willing to wait forge will be better 90% of the time. Comfy still has shit tier UX and odd bugs like lora block weight breaking when using fp8 models. Many people raise the alarm bells and they ignore stupid shit hoping people will fix it for them even though the extension ecosystem is often an attack vector save for a few trusted sources. Not even trying to troll I use comfy but I typically use it as a backend only in many cases.
>>
>>109754019
Also the documentation is fucking garbage and scattered across blog post that assumes the user knows what the fuck they are talking about without doing a proper preface, so many things that webui based frontends have just work with minimal effort. The high res fix WF to this day is still unacceptable from a user perspective and subgraphs do not fix the issue when it makes basic things like previews require you to step into them if you actually want a neat A to Z workflow.
>>
>>109753793
Avant grade catjat-kun
>>
Wow NAI 5 gens are so boring, so souless
>>
>>109753984
Nta but I feel like Krita AI is made for someone who's like 70% drawing and 30% messing with AI. The drawing tools are front and center (obviously, since it's drawing software), but stuff like context padding, blur, and mask grow are tucked away inside settings menus. That's something that doesn't really exist for a pure AI genner. It's kind of like burying pencil thickness and type inside a settings tab instead of having them right there for a drawer. You get what I mean anon?
>>
>>109740154
Hi /adt/, coming over from discord, has anyone downloaded MageTrail yet?
https://civitai.red/models/2918004/magetrail-mageflow-4b-dannborue621-finetune

It's a concept model based on Microsoft's Mage Flow. Some highlights from the author's comments:
About how easy and cheap is to train:
>"Only 100 dollars spent, it's a minor miracle that it's learning this well. The model has shown great promise in quickly learning and adapting booru concepts and tags to its knowledge base"
About how is faster and better than Cosmos being double it size:
>Being 15 to 20% faster than NVIDIA Cosmos2/Anima on inference despite being 2 billion parameters larger.
About how good the VAE is:
>Using MageVAE, which performs better than QwenVAE across the board, only behind the strongest open source VAE currently available, Flux2VAE (which it was distilled from). It also has the ability to slot in Flux2VAE for inference
TE:
>Having a decent Qwen 3 VL 4B text encoder
Native res:
256 to 2048 resolution native support
--------------------
And most importantly:
"Being fairly quick to learn and adapt to new knowledge without any knowledge forgetting"
--------------------
If I were you, I'd download it, give it a try, and maybe attempt training some LoRAs with it.
>>
>>109754339
very cool, SD 1.5 era artifacts are back
>>
>>109754339
Thanks, good to know Anima isn't the only option!
>>
File: 17450340.png (12 KB, 669x97)
12 KB PNG
>>109754359
It feels more natural and intuitive than Anima, it is easier to use. You should try it too, It is popular on reddit now.
>>
>>109754439
you are absolutely right!
it is perhaps popular on X (formerly Twitter) too?
>>
>>109754442
- Shift: 5.0
- Steps: 30
- CFG: 5
- Sampler: euler_ancestral or er_sde
- Scheduler: simple or beta
Try it, anon! https://files.catbox.moe/pzpa8a.png
>>
>>109754499
Holy shit! better text than Anima!
>>
>>109754499
>ModelSamplingAuraFlow
To the trash it goes
>>
>>109754339
>The model has shown great promise in quickly learning and adapting booru concepts and tags to its knowledge base
Ok I don't want to be throwing too much shade because any kind of experimentation with training a new model is a good thing. But this statement is giga cope. I tried the model and it has learned basically nothing from the booru datasets. It's learned a generic anime style bias and that's about it. IDK how the fuck anyone is concluding "wow it learns so quickly, much better than anima, amazing model architecture". It is like 0.1% fully trained, if even that, you can't conclude anything from this test.

For all we know anima looked way better than this at the same amount of training.
>>
>>109754563
>For all we know anima looked way better than this at the same amount of training.
proof?
>>
>>109754585
His ass. When preview 1 released, it cost $50k thanks to the Comfy grant. This finetune of MageFlow only cost 100 bucks.
>>
>another local attempt at nothing, as usual
where my NAI bros at? come on, post those 32ch SOTA gens
>>
>>109754585
I don't have proof that's why I said "for all we know". We don't know anything. We don't know if MageFlow is a better base model than Cosmos. What I DO know, is that a bunch of people on the noobcord have made some big cultish echo chamber about how bad anima is, how terrible the architecture is, and if we could just train an ACTUALLY good model like MageFlow it would solve every problem and completely surpass anima. And this experiment is supposed proof of that.

The model looks like shit. Yeah maybe it's just starting to learn some stuff but you have no way at all to do an apples-to-apples comparison with how well anima learned when it started training. I don't know how you can conclude anything from this experiment. It's been trained with only $100 of compute ffs.
>>
>>109754626
This desu. Kind of embarrassing seeing two cavemen still debating spear vs arrow.
>>
>>109754638
funny thing is that by only comparing this new model to anima theyre basically proving anima is the best anime model out there. doesnt make sense to make comparisons unless youre trying to dethrone the number 1 spot
>>
File: state of the ass.jpg (269 KB, 1056x1536)
269 KB JPG
>>
>>109754339
so with $200 it would be twice as good and then at 50k it would surpass anima
>>
>>109754638
Good objective point. The poor bastard fell for the meme and is now starting to spend actual money on halfbaked projects. It's kind of funny watching Bluvoll lead everyone down the same road to failure, getting them to sink 8k just like he did with MUGEN.
>>
>>109754599
>cost 100 bucks.
Four NAI subs would go further for that price.
>>
>>109754672
He will spend $700 for his next version, he's falling for the meme hard. Guess he will learn the local ways the hard way.
>>
File: 1758365451225249.png (1.1 MB, 768x1216)
1.1 MB PNG
>>109754712
Are you saying that improving this seven fold would not mog the hog?
>>
>>109754339
What discord
>>
>>109754740
nta but 'oobcord probably
>>
File: 1776651525312643.png (1.82 MB, 768x1280)
1.82 MB PNG
>>109754339
Actually based on how his anima loras look, I think his training settings are fucked in general. Like you could convince me this too was mageflow.
>>
File: ComfyUI_05157_.png (887 KB, 1024x1024)
887 KB PNG
>>
>>109754816
shut up schizo
>>
>>109754839
hi ani
>>
>>109754789
Total samples seen: ~ 326656
Training resolutions : 10242
## Version 0.1 (initial 20 epoch run extended 10 epoch run)
- Learning rate: 7e-6
- LR scheduler: Warmup -> Constant -> REX to 0e-7
- Precision: Full BF16
- Optimizer: AdamW8bit with Kahan summation (to offset BF16 precision roundoff)
- Weight decay: 0.02
- Timestep sampling: Logit-Normal, shift 6, sigmoid scale 1.0
## Additional training features
- Tag dropout: 10%
- Caption dropout: 5%
- Mixed captions at 25/25/25/25 ratio (tags only, NL only, tags-nl, nl-tags)
- Tag shuffle
- Caption shuffle
- Artist trigger attribution system
what went wrong here? what settings would you change to fix it
>>
File: file.png (2.27 MB, 1216x832)
2.27 MB PNG
>>
>>
>>109754863
>>109754880
Thanks for curing my eyes anons, staring at these local caveman paintings was starting to hurt
>>
>>
>>109746846
post account?
>>
>>109754858
Shift 6 seems high for a finetune but I am not confident since I have never trained on mageflow. It's just weird how even the Anima LoRAs have the same melted hair and eye issue.
Besides that...
> Mixed captions at 25/25/25/25 ratio (tags only, NL only, tags-nl, nl-tags)
> Tag shuffle
> Caption shuffle
For a $100 test run I think it would have been fine to have a strict tag only prompt style since the target is just going to stray at that scale.
>>
>>
>>109754915
Holy quality, thanks NAI for existing.
>>
how to make money as an anime ai artist
>>
>>109754664
Oh, I've seen that character get fucked in the ass, creampied and plugged up in a doujin before (for totally legit "mana resupply" reason of course). Why TF did she call the (male) MC "mommy"?
>>
>>109754816
Can you make one that says "Don't care, still a VRAMlet?"
>>
>300
>suspicious bumps
Is julien here?
>>
>>
>Schizo Sleuth reporting for duty
>>
File: file.png (1.94 MB, 832x1216)
1.94 MB PNG
>>
File: redriding.png (2.46 MB, 832x1216)
2.46 MB PNG
>>
I summon dark magician girl gens.
>>
File: redriding2.png (2.04 MB, 848x1239)
2.04 MB PNG
>>
>>109755019
who is julien?
>>
File: ComfyUI_05176_.png (994 KB, 1024x1024)
994 KB PNG
>>
New thread:
>>109755079
>>109755079
>>109755079
>>
>>109755075
The boogeyman.
>>
>>109747405
Nice. Thank you.
>>109747586
>Nurgle give them turbo regen
You got it very right there.



[Advertise on 4chan]

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

[Enable Mobile View / Use Mobile Site]

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