(((Celeb + E-Celeb AI))) Parody Nudes

(((Celeb + E-Celeb AI))) Parody Nudes

Anne Hathaway or Dakota Johnson?

need Bryce Dallas Howard Lora please

Is the government admitting movie and television violence is harmful enough to be banned as well? Those actors aren't actually dying on screen but it looks exactly like they are. I mean I see blood and body parts flying of in movies. How is fake murder illegal but fake tits are?

mart sharts genuinely fear a knock from the police every time they post a gen

Dog Bless KWABmerica

What website should I upload AI pr0n video tutorials to?

don't answer to this man he wants to know where you live so he can bring his computer police squad to it

I don't follow your logic.

bulk download i a premium feature

gay

can't you use something like JDownloader 2 to bulk download?

not from gofile, but you can still download one file at a time it's just annoying

man what a hassle if there could be hundreds. This is why people use web crawlers to rip everything from sites

Are the far files just archives of all the individual ones?

thanks bro

Can someone gen Hilary Duff?
Pretty Please!

yea it's just some anon's library

or just not use a trash scam hoster like gofile and use pixeldrain instead

What. The. Fuck.

loooool

Well if I am understanding this correctly it seems you can bulk download them. I just put the link for the page in JDownloader2 and they are all there from that first Gofile link. That is a lot of files and of course I won't be downloading them, my poor hard drives won't like it, plus I don't do images anyway, I just wait for them to appear and "borrow" them for videos.

file.png - 3450x1738, 346.13K

sir, this is the celeb ai thread

plenty of that on Anon Babble

sir, go fuck yourself no one cares

Need this unpantsed

1.png - 614x849, 836.08K

i hate the people who name their loras j3nfr_4nstn etc. and use similar triggerwords.

that's just the normal natalie portman lora with some 'young' proooompts

I suppose it would not be that bad if once you use the Lora it could automatically use that name as the trigger word without having to type it in. So 'im_a_fag_666.safetensor' would be 'im_a_fag_666' automatically

the training is more accurate if the person being trained is in the base sdxl model and their real name is used to trigger that base trained model

Actually you can use a node pack in your workflow that pulls the index 0 of metadata from the lora file, which is the trigger word. I have done it in my workflow and it works.

well, you need to do that to avoid concept bleed. Always hilarious to see how all the hermione or emma watson lora's used her real names as trigger words lol.

Any Tessa violet?

in Forge, if you go to the Lora tab and click on the wrench/hammer icon on the lora frame, it will let you set the trigger words and save them for that lora so when you click on it, it automatically adds the trigger words to your prompt along side the lora tag. other UI's might be the same, but i haven't used them

wrong, since that trash likeness that is baked into the base model then gets merged with your lora likeness

no, it doesn't influence the training that much. it just gives the training a starting point to gen the samples from, instead of trying to train from scratch. this was advice passed down from ppl who worked on SD

That is one of the good things about comfy is that it has so many nodes and i suppose ppl can write custom nodes as well. I will stick to Pinokio tho to do my Wan2.1 videos tho despite the limitations. although ironically you can install Comfy on Pinokio anyway so I suppose Wan2.1 could also be used on that as well. I might look to try SDXL and Flux for images now I have a 5060ti, it will at least be faster than my 2070.

Hey, didn't expect my Nicki gens here.

Might take some requests, if you're interested.

_Image 551.jpg - 831x1234, 1.03M

Well sometimes the celeb name is linked to other shit for example 'Eva Green' can come up with green shit instead of Eva Green so in some cases it helps to use a new word that's not trained into the model at all

Roxy Striar?

yeah, that's highly possible. another problem is if you're trying to train someone at a very different age than they were trained on the base sdxl model

i mean ai cum, tittydrops and the like.
you dont need a likeness lora for i2v.
but you mean image only?

Have you tried telling AI to generate a high speed video, and then you could slow it down on playback or re-encode it to increase effective length? Even 3fps can be very effective for repeated animations like fucking.

I'm looking for a Deanna Troi Lora since I found an AI chat card for her that is surprisingly good. The lora is an old 1.5 model one but that's fine.
Fucking your therapist is the most therapeutic therapy.

When you see ohxw woman, you know it is going to be shit.

Can we please make Roxy Striar take some Dick

Thanks for sharing! There were quite many duplicates but that's okay.

Proves you know nothing lmao. Plenty of excellent loras that just use the names

How tf do you guys get consistent gens on SD 1.5?

Every model I used turned into nightmare fuel, looked mid, or made bodywise errors.

When Civitai had Loras, I might've tried searching.
Sorry, little one.

Never thought I'd see high speed dubbing make a comeback

Yeah, Hilary Duff riding cock, looking over her shoulder, sucking her thumb.

Damn no Roxy, I'll go cry lol

I'm not saying 1.5 is good, I'm just saying if that's all there is it can work. AI chat has a limit in image sizes for it's expressions anyway.
Back when 1.5 was the thing to use, cyberrealistic was actually fantastic and knew a lot of celebrities on its own. Don't know how they fucked up their XL model so hard, totally useless.
Giving models more steps can help and you need a lot of negatives. All pointless now instead of just using XL with dmd at 5-7 steps and resizing with another 7 and still being faster.

I spent half a day making an animated png (which I previously didn't even know existed), by mass generating inpainted vaginas and shifting the entire image up and down slightly.

yeah - i've made a few hundred loras at this point and have never seen much difference between using weird "triggers" and the actual name. Also everyone thinks clip is trained on tags but it's actually tokens... concept bleed isn't that big of an issue. It sort of was for daisy ridley, because you'd get daisies on her nipples or something. But that's pretty rare. It just makes more sense to use their name for ease of use even at the expense of potential concept bleed.

in fact, the main thing that matters is having good quality images. the captions and training settings are like 20% of the work. Captions aren't even technically needed. for most anons the amount of work to create some "perfect" lora isn't worth it when you can just get some images and run someone else's training settings.

I-I kno... I-I k-know I-I shouldn't h-have her...

6 types of logins

Still chicken

Eh, I know a source.

file.png - 502x353, 429.55K

Can anyone share this without having to login?

how the fuck are you this bad at navigating the internet in 2025?

ease of use is more important than completely eliminating concept bleed from your main tag? Just give your lora your main tag as a name lmao.
Also, training settings are far more important than the actual images. Another very common misconception that gets spread by all the dumb fucks in this space

Do we have any of Streamer Biscuit? She's constantly posting teasing pics then getting mad about being sexualized. Soooo let's see her do it all

You have no idea what you're talking about and that's been proven clearly. Kill yourself

You use a unique word for the trigger so that the model itself does not fuck up what you want to generate like all those retarded loras made by Indians using images of the celebrity 20-30 years after they were famous that absolutely nobody wants to look at because those images are higher resolution. That's what is in the base model using their name. Then there are other cases like Emma Stone and Sadie Sink where you don't want everything generating fucking stones and sinks.

With the proper settings you can literally train a pretty high quality lora with blurry and heavily artifacted thumbnails. Ofc quality is higher with higher quality pics but thats not the point. Always hilarious to experience first hand how clueless everyone actually is

instead of bitching about what's best, just download the sdxl base model and set it as your checkpoint and try genning the person with their real name. if the gen looks like the person you want to train, then use their name to help with the training. if not, then don't.

You know nothing. Kill yourself

I tried comfy once because I wanted to do batch facial replacement to generate expressions.
It was a huge clusterfuck where everyone can make nodes that use the same names and conflict with each other.
I just made a wildcard for expressions and used dynamic prompt to generate all combinations in forge instead.

piracy is illegal world-wide, yet millions if not billions of people still do it and ive never known anybody in my life receive punishment for downloading a movie or game.

even this site has a torrent board that shares paid and copyrighted content and its doing just fine

No gen no opinion

it's fine he's just retarded/trolling

my face is ready, have a seat

Imagine taking any sort of advice from someone who produced something that look like this lmao

well that would look better if interpolated, it's 16fps and any fast motion looks poor

you don't go to jail for downloading any movie, but people have gotten in legal trouble for uploading movies in America.
same here.

i pirated the Until Dawn movie and i still feel ripped off.

A man named Jack Yates has been sentenced to 6 months in prison for pirating ‘The Love Guru’.
lol

I think the prison clause has to do with illegal illegal gens, you know the ones, who let the dogs who who who who

i tried watching it 3 hours ago. that movie sucks ass. couldn't last more than 30min

the 10 seconds of Maia Mitchell is the only redeeming feature so far

different case

Kristopher Dallmann and four others operated Jetflicks, an illegal streaming service that uploaded and distributed pirated movies and TV shows
Outcome: All five were convicted of conspiracy to commit criminal copyright infringement in a Las Vegas federal court.

it's because he pirated a movie with gay title like he's a faggot

don't know who that was, doesn't matter to me. i have standards, horror movies is never about chicks for me.

it starts slow but the second half was pretty good. kind of like final destination crossed with cabin in the woods or something. I liked the second half.

i hate both of those movies

maybe you're a fag

cyberrealistic was actually fantastic and knew a lot of celebrities on its own. Don't know how they fucked up their XL model so hard

Probably to avoid the hastle of AI fakes, I guess.

riiiight, so the big guys at the top, and what about the thousands that downloaded them, are they also convicted in a federal court?

hope the Backrooms movie will be better than this drivel

no those movies are for pansies. i rather watch a chunkblower with a cool story. the writing, effects, acting all were a waste

i never said its never happened, just that its never happened to anybody around me and ive never heard of anybody (from people around me) that have been fucked for piracy.
obviously people get fucked for it and you can find that online, but as i said, its extremely uncommon relative to the amount of pirates there are world-wide and that its barely enforced on the internet

but that's what I told you.
downloading= no real issue
UPLOADING= jail time .

if you download Lilo and stitch right now, whatever
but if you UPLOAD it online, jail.
same here?

that's dumb, and why doesn't they just go after the ai companies making the programs?

Streamer Biscuit

One, multiple people have that in their name.

Two, we have a Lora shortage, so don't count on it.

Three, good luck due to limited reach.

And four, post pic or GTFO

no
what I was saying is that downloading songs or movies it's whatever
but do you know anyone that Uploads movies or songs online?

because uploading stuff online can fuck you up. recording something illegally and uploading it is heavy.
but most people don't do that.
h

i get what youre saying, but its still piracy even if you just download something though, right?

it's supposed to be distributing and sharing of material.

The US baked in a clause to forbid regulation of AI into their "Big Beautiful Budget" scam bill.

WHY THE FUCK ARE YOU RETARDS USING GOFILE

large uploads without having to create an account

Is there still a way to see the celebrities loras in civitai or did they finally remove them all?

a spoonful of semen makes the medicine go down

civit will no longer allow loras or gens of any real people and they've killed all celeb loras. cant even do movie characters like indiana jones

Where can I get good taylor swift loras?

Thank god I downloaded pretty much every interesting LoRA on that site couple days back. There were just so many flux loras I couldn't go through them all so I think I missed some. Also I didn't get the WAN celeb loras but there wasn't that many of them.

check these links

ehhhh I guess. I agree car manufactory or gun companies shouldn't be sued for what people so w/e. It's all about money, those that have it all can lose it overnight if we ever get a un-neutered system

So there is a bill outlawing the celeb law?

I have 70 GB of XL/pony celeb loras but there isn't a reasonable place to put them.

no, the provision in the bill just prevents states from regulating AI

has anyone made some ice spice stuff?

you can put them on my hard drive

I think the free tier of mega will do 50GB

Any of billie eilish? She is perfect for monster fucking

Bitch I have 120 GB

so, does this bill cover only nsfw use? what if its got nothing nsfw about it or is fan art?

if it's nude or sex I think. but a bikini pic I don't know. I guess not

only nsfw is banned

freedom everywhere but the us

so just make all of them like ken and barbie dolls, no nips or danglys EZ

freedom everywhere but the us

and europe

unfortunately theres no taylor swift loras there

transferir.png - 1024x1024, 1.31M

This is Kamala's new job now guys... I don't think she likes it.

blowjob or handjob for you? make your selection

sad bj.mp4 - 592x640, 1.12M

wait it's not a steamdeck, it's kamala harris

ever seen a roach cumshot?
well, now you have.

cinema

kek

the dung beetles have been, time from the semen beetle to finish the job off

would be nice to see nikki naked and bound engaging in some consensual breathplay like fighting for air with a plastic bag ziptied over her head

No Olivia Rodrigo smh

i downloaded the offline easy diffusion but am lost as to how to train this thing. any links to guides for dummies?

Any of these have Kiera Knightly?

The first one might, it has a lot of celebs. I was looking for Emma Roberts so i didn't notice if KK was one of them or not.

Any Amy Lee

this is some advanced laziness
I get being too lazy to learn how to bake your own LoRA but being too lazy to click a link and cmd+f a name is next level
I can't see any Kiera's there but there are 2 links with several LoRA compressed into RAR files so maybe there you lazy fuck

I was asking about the compressed ones. Don't want to bother downloading if there isnt what I want.

i7 4790k, RTX 3060 12gb,
should i even bother trying?

for images?
sure easy with dmd2 and a decent workflow it will take less that 20 secs to gen an image
480p videos would take over half an hour

I had a 2070, I genned videos (on the wan2.1 model) at 480p 15 steps and they took around 30 mins for 5 sec clips (81 frames). You should do better than that especially as you have 12gb and not 8gb

this thread is boring now. where's the snake guy?

What's the problem with it?