Software Engineering Courses (SECourses)•3y ago

50% cartoon model merge

SSandstormleader 50% cartoon model merge

Stefanoid9343•10/8/23, 5:11 PM

do you have the base picture?

Stefanoid9343•10/8/23, 5:11 PM

what model do you use?

SandstormleaderOP•10/8/23, 5:16 PM

base picture of what?

Stefanoid9343•10/8/23, 5:16 PM

did you use images to train this model right?

SandstormleaderOP•10/8/23, 5:17 PM

Stefanoid9343•10/8/23, 5:18 PM

how many images did you use to train the model? I have a memory error and I could’t test the training

SandstormleaderOP•10/8/23, 5:20 PM

@Stefanoid9343 about 220

SSandstormleader @Stefanoid9343 about 220

Stefanoid9343•10/8/23, 5:20 PM

Wow

SandstormleaderOP•10/8/23, 5:20 PM

it's my wife so i had plenty haha

Stefanoid9343•10/8/23, 5:21 PM

on google colab I use to use like 17 and it works

SSandstormleader it's my wife so i had plenty haha

Stefanoid9343•10/8/23, 5:21 PM

I train my wife image too but like 30

Stefanoid9343•10/8/23, 5:21 PM

and it works

SandstormleaderOP•10/8/23, 5:21 PM

i trained with about 20 for the model i made of my self but the poses and angles didn't seem nearly as diverse

Stefanoid9343•10/8/23, 5:22 PM

I try to follow the new tutorial but I run out or memory error

SSandstormleader i trained with about 20 for the model i made of my self but the poses and angles...

Stefanoid9343•10/8/23, 5:22 PM

yeah needs to be different

SandstormleaderOP•10/8/23, 5:25 PM

ya i just don't have many professional photos of myself hanging around

Stefanoid9343•10/8/23, 5:25 PM

at this point you don’t need that those models are taking care off

SandstormleaderOP•10/8/23, 5:26 PM

ya but selfies are all mirror image not real image.

Stefanoid9343•10/8/23, 5:26 PM

that’s right

SandstormleaderOP•10/8/23, 5:28 PM

my working theory atm is that that higher quality images and resolution get better results

SandstormleaderOP•10/8/23, 5:28 PM

i tried a few celebs to experiment and i didn't like those nearly as much

Stefanoid9343•10/8/23, 5:28 PM

do you have you own card? or you are using runpod or kaagle

SandstormleaderOP•10/8/23, 5:29 PM

everything local

SandstormleaderOP•10/8/23, 5:29 PM

i figured buying a new pc would be cheaper in the long run

SandstormleaderOP•10/8/23, 5:30 PM

i need to try and experiment on product photography again but i didn't have much luck with products that had text on them in the past

SSandstormleader @Stefanoid9343 about 220

Elisoli•10/8/23, 5:34 PM

Looks really good, how many repeats did you do for 220 images?

Elisoli•10/8/23, 5:35 PM

or was it 1 repeat and the rest just epochs?

SandstormleaderOP•10/8/23, 5:43 PM

i just did 150steps (epochs) not sure what repeat is

Elisoli•10/8/23, 5:44 PM

so no reguleraztion images at all, 220 images * 150 33000 ste[s. did you not get overtraining issues? or did you choose an earlier checkpoint?

Neuromasters•10/8/23, 5:45 PM

Hello, long time ago I made models and already forgot how to customize them in colab,
I used to make initial descriptions in WD14 loras and edit them further. How can I do the same in colab?

If you can, can you share the config for better training accuracy
I want to put the training 768 x768 pixels, base model take anylora from lykon

EElisoli so no reguleraztion images at all, 220 images * 150 33000 ste[s. did you not get...

SandstormleaderOP•10/8/23, 5:47 PM

no i didn't like the early checkpoints, used the 33k one

SandstormleaderOP•10/8/23, 5:47 PM

once it's finished i usually merge it with another model i like at about 30% to help clean it up a bit

Elisoli•10/8/23, 5:52 PM

very nice workflow

looks really really good

SandstormleaderOP•10/8/23, 6:27 PM

@Dr. Furkan Gözükara anyone know if gradio link is working again yet?

Narva•10/8/23, 6:56 PM

sdxl gives very good results training specific objects, very versatile

SandstormleaderOP•10/8/23, 7:02 PM

nice

SandstormleaderOP•10/8/23, 7:04 PM

SandstormleaderOP•10/8/23, 7:04 PM

trying to make myself some wallpapers

SSandstormleader Click to see attachment

Stefanoid9343•10/8/23, 7:30 PM

My training Starts! I jus change the server to 1 x RTX 4090 8 vCPU 31 GB RAM!

Stefanoid9343•10/8/23, 7:30 PM

3hrs to go!

SSandstormleader Click to see attachment

Stefanoid9343•10/8/23, 7:32 PM

This is really nice!

rafraf•10/8/23, 8:17 PM

@Dr. Furkan Gözükara have you tested xformers/gradient checkpointing as well as Memory Efficient Attention? I am running without Gradient CH. and xformers but still have memory efficient attention on. Not sure if it decreases quality

rafraf•10/8/23, 8:38 PM

Also, wit A6000, on runpod, running 14 images dreambooth LoRA, at 4 batch size, I run into OOM, without gradient checkpointing and xformers. Using class images. Any thoughts if there are any memory optimizations better than these two? What do you recommend @Dr. Furkan Gözükara ??

SSandstormleader 220 professional photos, cropped but not down sized, trained at 768, xformers, 8...

Furkan Gözükara SECourses•10/8/23, 8:50 PM

220 pro photos is the key

NNeuromasters Hello, long time ago I made models and already forgot how to customize them in c...

Furkan Gözükara SECourses•10/8/23, 8:52 PM

i dont knoıw if colab supporting captions never tried

Furkan Gözükara SECourses•10/8/23, 8:52 PM

but colab is working

Furkan Gözükara SECourses•10/8/23, 8:52 PM

i did training few days ago

Furkan Gözükara SECourses•10/8/23, 8:52 PM

https://youtu.be/mnCY8uM7E50

YouTubeSECourses

Transform Your Selfie into a Stunning AI Avatar with Stable Diffusi...

Our Discord : https://discord.gg/HbqgGaZVmr. How to do free Stable Diffusion DreamBooth training on Google Colab for free. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on

https://www.patreon.com/SECourses

Playlist of Stable Diffusion Tutorials, Automatic1111 and Goo...

SSandstormleader @Dr. Furkan Gözükara anyone know if gradio link is working again yet?

Furkan Gözükara SECourses•10/8/23, 8:52 PM

i havent tested but it wasnt working like yesterday

NNarva sdxl gives very good results training specific objects, very versatile

Furkan Gözükara SECourses•10/8/23, 8:52 PM

it is super strong

50% cartoon model merge

Similar Threads

Similar Threads

Similar Threads