with sdxl captions really reduces likeliness but perhaps only eyeglasses can be added 😄
with sdxl captions really reduces likeliness but perhaps only eyeglasses can be added 
Dreduces likeness if you use descriptive tokens, but adding accessories like glasses or clothing is not a problem
Gbruhhh
Gi do step2 batch and it breaks
Ji asked chat GTP how to, it worked very well for me, no need for new windows install 
Gi updated to 3.10.11 python too
Fdont use that
Fuse my starter and see
Fand any error at step 2?
M@Dr. Furkan Gözükara what is this
Its justa mazing!!
Its justa mazing!!
Sdr Furkan when do you think could be a kohya flux tutorial update_? i saw and i did SDXL but seem to diferretn form flux process
Dto avoid the problem I use captions with comma "ohwx man, prescription glasses" on that way it separate the concepts
Gno error on step 2, says START GUI, when I use your starter it pops up for 1 sec and dissappears
Fi am almost there to start recording it
Fwell i cant know sadly. something broken for some reason
Gwhats the command to PAUSE on a batch file?
F
Fopen cmd
Fand type bat file name
Fyou will see error
F
G
FWell apparently the training has started but I've messed around with so much stuff that I don't know why lol
F
Flooks good
Fbut too many steps how many images youy have
Fye it fail due to something broken
Fno ideas
FI spoke too fast lol it doesn't seem to launch, I have 62 images, how many should I put?
Fstill to many steps
F62 * 200 = 12400 steps
FShould I lower the steps or the number of images?
Fyou should fix setup
Fwhy you have that many steps you know?
FI have no idea
FMaybe because I clicked on prepare training data?
FThis is what I have
Kohya_ss_GUI_v24.2.0.pdf1.86MB
Fyou have 40 repeats in config
Fshow config folder names
Nseems like discussion here is usually around only 10-20 training images, and that by 1000-2000 steps the lora produces a great likeness (assuming you're going for a character lora). since you have 3x-6x as many training images, you'll have 3x-6x as many steps... steps is images * repeats * epochs. the doc's configs always have repeats at 1, so it's just images * epochs. might be worth additional testing to loras made by 10 images at 200 epochs (2000 steps) vs 20 images at 100 epochs (also 2000 steps). if we increase images, do we get a "good" lora faster? or if we increase images, do we need more steps to better "merge" all of them? i am just a hobbyist and now i'm rambling lol - would be interested in the Dr's feedback.
Fi hate mountain house electricity went off :/
Fi will test more images better or not with flux but still optimizing hyper params
Freg images definitely not helping
Fi am going to sleep probably electricty wont come until morning :/ 2 am here
Fhaha electricty came faster
GWhere you at bro? South America?
FTurkey
Fin mountain house
GI thought you were Hispanic
Fnope
F