: self explanatory post a real image and get an anime image don't post pictures where hands (or objects in general are disturbing) can be done but take longer, or if you want to do it but then don't complain about bad hands etc.
>don't post pictures where hands can be done but take longerWhat the fuck does this even mean you ESL retard
>>19600071>What the fuck does this even mean you ESL retardexactly what post says, retard. to make good ‘hands’ in addition to making several attempts, and the ‘prompt’ is not enough, you need controlnet and adtelier which make the photo take longer to generate, as this is meant to be a thread (there is not only your photo, although you can make more requests) avoid posting them or don't cry.
>>19600075>exactly what post says, retardThe post makes no fucking sense. Do you want people to not post pictures with hands because you're too stupid to figure out AI hands in 2025?
>>19600103>>19600103>you're too stupid to figure out AI hands in 2025?it takes me 5 minutes per gen with everything on. without 30 seconds sherlockit takes me 5 minutes per gen with everything active. without 30 seconds sherlock now if you're done investigating, post a picture or gtfo this thread. i simply don't have 48948954779849849849489898984489498 Vram that allows me to do everything in 10 seconds
Try this one but do a good job this time. Make the dress look accurate, use the right eye color tags, use the proper facial expression tags.
>>19600126use the proper facial expression tags.Are you really asking me to commit to a thumbnail photo? No, thank you. Normally I would just insult you, but since you're obsessed, that image only serves to make it clear what the thread is for since 90% of the posts here are by ‘real’ people (e.g. Nudify)
>>19600151>thumbnail photoWhat
Nigger if you can't even get the first picture in your thread done properly then don't even bother making these threads
>>19600126this>>19600126 and this >>19600054 (the real one) are the same image, I always create a generic image for the Catalog(so a thumbnail) I have already explained the reason here >>19600151
>>19600159last time reach 300+ maybe because you're complaining about all the shit instead of starting, for me it's all the same, I even have fun in a different way
>>19600173Didn't OP abandon the thread initially and a whole bunch of people mock OP for doing a shit job when he actually did show up?
>>19600220the thread was initially abandoned because there were no requests (like now that it has been open for 5hours and there are only rant posts) and other users ‘carried it on’ in my place but they made occasional images, even if they made better ones than mine, there were some users who asked for multiple images and I satisfied them up to post 300 the criticism (from requesters) came at the beginning because ‘omg you left the thread’and because i set the rule of one girl only for this reason>>19600122 because i want keep things simple, despite the fact that I can even generate 3 together
bump
>>19600054Hey that's my girl
>>19600509not a good image to make 1:1 I explained the reasons abovepost a infornt pic solve the problem
>>19600536
>>19600545
>>19600543thank you so much!
>>19600567np feel free to ask more
>>19600567thank you!
>>19600581
>>19600596Thank you so much master
>>19600594
>>19600584
>>19600557Thanks
>>19600509
>>19600722Wait I think this looks better
>>19600624>>19600584made a better version
>>19600689enjoy your futanari
>>19600787
>>19600824
>>19600862
>>19600890
>>19600915Lol didn't knew Dolly Parton was Sakura Haruno in anime, thanks.
>>19600992
>>19600126
>>19600998Thanks wiz
>>19601010
I know, the guns.But the photo is funny. Could you try?
>>19601001
>>19601069you're damn good at these. mind sharing your tools and workflow?
>>19601072I don't use ComfyUI so there's no "workflow" but I can explain the rest:- Clean up, rotate and/or crop the image beforehand if needed (don't wanna generate a bunch of empty sky, for instance). The img2img process will be fairly high strength, so don't worry about the cleanup being pixel perfect. The image you quoted I just made the background white, for >>19601029 I changed the background color a little so it wouldn't get in the way.- The checkpoint I use the most is autismmixSDXL but any good anime model should do really.- Pay attention to the image tags you're using in the prompt and be willing to tweak them if the facial expression seems off compared to the original image. Becoming familiar with these tags can take a while.- Artist LORAs help a ton in not making these look generic. The one I've used in this thread is based on Cirenk's work.- The rest is just an img2img process. 60-70 steps, CFG scale at 6 or 7, strength varies according to picture (anywhere between 0.5 and 0.8 depending), and Canny ControlNet (prioritizing ControlNet, but Ending Control Step set at 0.2 to 0.4 depending on picture).- If some details come out wrong try fixing them through inpainting, or by editing the source image by removing whatever's causing trouble (again, this part doesn't have to be perfect either).
>>19601079
>>19601079Are you the same anon who did the last thread?
>>19601083If you mean almost all the images in >>19591698 then yes
>>19601086You also doing the realification and Chun li threads?
>>19601091Yeah I did the better looking images in the realification thread and did those Chun Li images but I didn't notice the Chun Li thread LOL. I'll add something to that thread later tonight.
>>19601082
>>19600890"Jolene" ain't got shit on this
>>19601099That turned out pretty damn good>>19600994
>>19601021
>>19601047Thanks wizard
>>19601079Thanks wiz
>>19601108Hot diggidy damn
>>19601027
>>19601029Thanks wizard
