>>462 That Lora really does her tits perfect

There used to be a really good artist who used to do stuff like bikinis uniforms maids etc and he made the girls really thick. He was on DA and then it got deleted and that was it. Shame. I think his name was AIHUB

>>462 >>501 It really does. I not a huge tit guy, but good Fran gens are something to see, thx.

Wet and wild

Thanks MKUltra!

One more quickie/study, done now.

>>524 >>526 So cool Tell MKUltra that I'm waiting for him here Tell him to add me on disc: ryannabel

>>527 How about you go back to your containment thread instead and nobody gets hurt?

>>527 get out

how is it the 'no rules' thread has less posts then the celeb gen thread? anyway, does anyone know how bumps work here? Seems like I can't post in the other thread anymore due to some sort of bump limit.

>>619 what I was going to post over there is: tfw I'm the op for over 20-30 of the images reposted here. \\\(only 3 in this thread)/// feels good man. now my work is out in the ether, floating around different chans for all time It's nice that this place doesn't strip metadata so it's easier to see peoples prompts- and we don't have to upload to catbox anymore

>>620 >It's nice that this place doesn't strip metadata Wut? I just tried it with a few pics of mine I accidentally uploaded without stripping metadata but that certainly doesn't seem to be the case. Am I doing it wrong? :P

>>685 huh, I'm not able to do it for anyone else's posts, but I can download mine and PNGinfo can read it. A lot itt are either not PNG or made with comfy (guessing by the filename format), so I was only able to test a few. want to post a png made with forge so I can test?

>>719 I don't have/use Forge but that could explain it. It would depend on exactly what metadata/chunks it filters. Comfy saves into custom text chunks for png and there's a node that uses the "Manufacturer" metadata for jpg to put workflows in, but both appear to be filtered.

can someone help me out with making a lora? i followed the guide(s) for Biglust on the OP, i have 120 images, mostly good quality with a couple smaller (640x480) but most between 1024-3096 (in width or height). i did use bucketing, but i dont know if this works well in sdxl/pony based models like it does with flux. i have trained a few flux loras that pretty good, but the pony/sdxl ones always come out not-at-all like the person. training went without problems, but the lora itself was pretty bad, as in not like the person at all. i use the biglust_v6 model in kohya_ss, i didnt change anything other than the folders/names and total epochs (per the guide, 3500/num of images which came to around 30 epochs). does anyone have ideas, suggestions, or share a kohya json settings file that works for you please?

>>748 I've heard that it's better to train with fewer, better pictures (ie taking out your smaller ones) but that shouldn't make the lora look nothing like your target, it usually just reduces quality. Not an expert by any means, but I hope some kind anon can come along to help

>>750 i actually did end up using only 60 images (half the set) as a test and it was a very similar result - no match for the person. i think something in the biglust link instructions is not right, or otehrwise i have some completely messed up thing on my machine (but then again the flux loras are fine?)

>>754 only thing I can think of is you didn't update the "pretrained_model_name_or_path" to the path of your BigLust v1.6 model. not sure why there would be such a big difference between flux and biglust

>>787 I know it's no rules but I genuinely think Ryans are better in the containment thread. I mean, this is actually a nice image (even it's a repost, I assume). Then again the whole point of Ryan originally was derailment, so what's left if it's not derailing anymore. :P

I love taytay!

>>841 >the whole point of Ryan originally was derailment Ryan is a celeb Ryan is not derailment Ryan is legit post get mad, troll no gen no opinion

>>911 Well your retirement didn't last long, did it? Welcome back, ryanfag. I mean you quoted the wrong post but hey, nobody's perfect.

Any chance of getting Kat McNamara or Yvonne Strahovski?

>>911 i missed this motherfucker like you wouldn't believe lol

>>911 >Sees people having fun posting nude celeb women >Posts shitty male gens of Ryan >Some anon keks >Proceeds to repeat 9001 times like child >4chan is kill >Comes into new board You have always known exactly what your doing and it's just not the same anymore is it? Move on.

18yo Kirsten Dunst 18yo Isabela Merced 18yo Marjorie Taylor Greene 18yo Queen Elizabeth II of United Kingdom 18yo Hillary Clinton 18yo Diane Franklin 18yo Candela Anton de Vez 18yo Aurora Aksnes 18yo Queen Juana I of Castile (based on Juan de Flandes' painting, ca. 1500 AD) 18yo Brett Cooper 18yo Lauren Southern 18yo Greta Thunberg 18yo Ivanka Trump 18yo Melania Trump 18yo Princess Leonor de Borbon y Ortiz 18yo Xochitl Gomez 18yo Laura Prepon 18yo Michelle Trachtenberg 18yo Sarah Chalke 18yo Ornella Muti 18yo Ana Kasparian 18yo Ivy Wolk 18yo Jamie-Lynn Sigler 18yo Grimes 18yo Ava Lalezarzadeh 18yo Nancy Pelosi 18yo Empress Agrippina the Elder (based on the marble bust from Pergamon, ca. 1st century AD) 18yo Maria Valverde 18yo Lola Naymark 18yo Sofia Coppola 18yo Elva Guerra 18yo Irene Escolar 18yo Sophia Lillis 18yo Barbara Streisand 18yo Denims 18yo Priscilla Alvarez 18yo Gillian Anderson 18yo Farha Khalidi 18yo Frogan 18yo Queen Frederica of Greece 18yo Princess Victoria Louise of Prussia 18yo Kate Winslet 18yo Jennette McCurdy 18yo Gaby Hoffmann 18yo Mckenna Grace 18yo Lena Dunham 18yo Colleen Haskell 18yo Paula Usero

[Expand Post]

18yo Kaitlyn Dever 18yo Amber Midthunder 18yo Anamaria Vartolomei