Software Engineering Courses (SECourses)•15mo ago

Hey some nice images @Dr. Furkan Gözükara ..do you mind sharing the prompt for this one and the acti

Hey some nice images @Dr. Furkan Gözükara ..do you mind sharing the prompt for this one and the action hero ?

You can DM if you like or in patreon chat ?

Ddxqb what inference parameters do you use? samplers and such

Furkan Gözükara SECourses•10/6/24, 6:19 PM

iPNDM

Furkan Gözükara SECourses•10/6/24, 6:19 PM

40 steps

Furkan Gözükara SECourses•10/6/24, 6:19 PM

flux cfg 4

PPixelMonkey Hey some nice images @Dr. Furkan Gözükara ..do you mind sharing the prompt for t...

Furkan Gözükara SECourses•10/6/24, 6:19 PM

i will share all prompts on patreon

Furkan Gözükara SECourses•10/6/24, 6:19 PM

finalizing atm

xlads•10/6/24, 6:41 PM

Elsewhere•10/6/24, 6:43 PM

Question: Anyone know of a way, or a node; which allows me to send an image down different branches, instead of all?

Elsewhere•10/6/24, 6:44 PM

in comfyui

Xxlads Click to see attachment

Furkan Gözükara SECourses•10/6/24, 7:01 PM

nice

Furkan Gözükara SECourses•10/6/24, 7:05 PM

really cool

1905-ohwx20man20as20a20character20in20a20shaft-style-FT_256_images_BS7_5e_01_xFormers-000040-1750876568.png

Furkan Gözükara SECourses•10/6/24, 7:06 PM

1905-ohwx20man20as20a20character20in20a20shaft-style-FT_256_images_BS7_5e_01_xFormers-000040-1750876570.png

Furkan Gözükara SECourses•10/6/24, 7:08 PM

1907-ohwx20man20as20a20character20in20a20mappa-style-FT_256_images_BS7_5e_01_xFormers-000040-1789627509.png

Maxivi•10/6/24, 7:10 PM

@Dr. Furkan Gözükara have you tried reg images with dreambooth?

MMaxivi @Dr. Furkan Gözükara have you tried reg images with dreambooth?

Furkan Gözükara SECourses•10/6/24, 7:22 PM

yes i am testing training results now

Furkan Gözükara SECourses•10/6/24, 7:22 PM

if you mean fine tuning

Furkan Gözükara SECourses•10/6/24, 7:22 PM

i think i tried didnt work

Furkan Gözükara SECourses•10/6/24, 7:22 PM

i am testing onetrainer fix

Furkan Gözükara SECourses•10/6/24, 7:22 PM

thing

Maxivi•10/6/24, 7:23 PM

yeah for dreambooth

Maxivi•10/6/24, 7:23 PM

ontrainer one i tried like 7 different trainings, couldnt get it to work

Furkan Gözükara SECourses•10/6/24, 7:24 PM

ye me neither

Furkan Gözükara SECourses•10/6/24, 7:24 PM

not working

Furkan Gözükara SECourses•10/6/24, 7:24 PM

FFurkan Gözükara SECourses Click to see attachment

Maxivi•10/6/24, 7:25 PM

Furnald Gozump

Furkan Gözükara SECourses•10/6/24, 7:25 PM

could cause i added text encoders?

Furkan Gözükara SECourses•10/6/24, 7:25 PM

Furkan Gözükara SECourses•10/6/24, 7:25 PM

i didnt train though

Maxivi•10/6/24, 7:26 PM

he said he wasnt training them

Furkan Gözükara SECourses•10/6/24, 7:26 PM

i also dont know what is meaning of include

Maxivi•10/6/24, 7:26 PM

on onetrainer you need to include them either way

Furkan Gözükara SECourses•10/6/24, 7:26 PM

it doesnt make sense

Maxivi•10/6/24, 7:26 PM

when you have the include option its like in kohya when you load in clip l and t5

Furkan Gözükara SECourses•10/6/24, 7:27 PM

photo of a man

Maxivi•10/6/24, 7:27 PM

yep had the same issue with mine

Maxivi•10/6/24, 7:27 PM

i wish @dxqbYD could provide a detailed workflow and training so we can see it properly

dxqb•10/6/24, 7:37 PM

including TE Is fine just no training

dxqb•10/6/24, 7:39 PM

let me know how I can help, but I currently don't know what you could do differently. if you look at my samples on lower left and lower right, the images are nearly identical. dreambooth-training

dxqb•10/6/24, 7:42 PM

you could provide config and images and I'll have a look? but I'm on a business trip from Mo-Wed

Ddxqb let me know how I can help, but I currently don't know what you could do differe...

Maxivi•10/6/24, 7:49 PM

you could provide info like, learning rates, optimizers, noise offset etc

dxqb•10/6/24, 7:49 PM

I could, I just have no reason to assume that is what makes the difference - because the difference is not just a little, but night and day

dxqb•10/6/24, 7:50 PM

dxqb•10/6/24, 7:50 PM

nevermind the "train TE1 embedding" switch, that's just something I'm currently trynig. my samples were without

dxqb•10/6/24, 7:50 PM

also, LR is not comparable to yours because I am using alpha 1

dxqb•10/6/24, 7:54 PM

I actually don't think this will be useful in the end, because it limits training of the concept you want to train (or 'instance' as kohya would call it) too much. but that's the opposite of the problem of what you're seeing

MMaxivi you could provide info like, learning rates, optimizers, noise offset etc

dxqb•10/6/24, 7:54 PM

that enough?

Maxivi•10/6/24, 7:55 PM

oh wow your learning rate is super high

dxqb•10/6/24, 7:55 PM

no it's not. it's because I'm using alpha 1.

dxqb•10/6/24, 7:55 PM

if you are using 16 you can practically divide the LR by 16

dxqb•10/6/24, 7:56 PM

what LR and alpha are you using?

Maxivi•10/6/24, 7:56 PM

128x128 and 3e-06

Hey some nice images @Dr. Furkan Gözükara ..do you mind sharing the prompt for this one and the acti

Similar Threads