Software Engineering Courses (SECourses)•3y ago

Hey all, trying out ControlNet from the repo, with the prompt `photo of an owl`. DDIM 30 steps. `con

Hey all, trying out ControlNet from the repo, with the prompt

photo of an owl

photo of an owl

photo of an owl

photo of an owl. DDIM 30 steps.

control_sd15_scribble.pth

control_sd15_scribble.pth

control_sd15_scribble.pth

control_sd15_scribble.pth Anybody know what might be going wrong here?

maniac6420OP•2/15/23, 4:30 PM

Using this.

Sonic•2/15/23, 4:45 PM

Hey there, sorry for bothering,
I already have Automatic1111 with 1.5 since a long time and it s working perfectly, But, I followed your tutorial to install ControlNet step by step, but still couldn't make it to work (I didn't have that ControlNet Bar/Triangle in the UI)
I would be very thankful if someone could guide me to know the issue
Thank you in advance !

Yee•2/15/23, 4:55 PM

Thanks for the tutorial on last video i got this result

Yee•2/15/23, 4:55 PM

YYee Click to see attachment

kingsley•2/15/23, 4:56 PM

Damn this is nice

GP•2/15/23, 4:56 PM

yeah looks great!

Sesarr•2/15/23, 4:56 PM

Wow, which SD model is this?

Yee•2/15/23, 4:57 PM

im using 1.5, the one that come with the automatic1111, i use prompts, helmet, bioshock

Sesarr•2/15/23, 4:58 PM

Thank you, incredible result

Yee•2/15/23, 4:59 PM

yeah really impressive, i noticed the sketch has to be clear

Yee•2/15/23, 4:59 PM

otherwhise u get weird results lol

Yee•2/15/23, 4:59 PM

maniac6420OP•2/15/23, 5:00 PM

I created a super easy to easy scribble tool for ControlNet: https://diffusionmagic.web.app/#/

magic_canvas

A new Flutter project.

maniac6420OP•2/15/23, 5:01 PM

(but I'm not sure if it's working as expected yet, see the owl

)

Armando•2/15/23, 5:29 PM

Hello, Hello, Quick question does anyone know is there a Formula for Embedding Learning Rate and Batch Size for best results

Armando•2/15/23, 5:30 PM

I am trying to train Textual Inversion on my RTXA5000 card for Stable diffusion 2.1 and not getting quite good results.

FFurkan Gözükara SECourses yes i am trying to improve my video editing skills

Colored Tonalities•2/15/23, 5:45 PM

you are improving

Yee•2/15/23, 7:06 PM

how are u guys getting good results? every model i download looks, kinda ok ;/ i think it's my prompts, idk.

HHaashim GM @Dr. Furkan Gözükara … I noticed a Runpod user was able to use the kohya gui ...

Furkan Gözükara SECourses•2/15/23, 8:16 PM

runpod io runs on a linux os

GGP Hey! So I've downloaded and replaced but still getting these issues

Furkan Gözükara SECourses•2/15/23, 8:17 PM

hello. could you install directly in C drive? still looks to me as path permission error

SSonic Hey there, sorry for bothering, I already have Automatic1111 with 1.5 since a lo...

Furkan Gözükara SECourses•2/15/23, 8:17 PM

sure i can help. but need more info

YYee Click to see attachment

Furkan Gözükara SECourses•2/15/23, 8:18 PM

very good congrats

Mmaniac6420 I created a super easy to easy scribble tool for ControlNet: https://diffusionma...

Furkan Gözükara SECourses•2/15/23, 8:18 PM

nice

CColored Tonalities you are improving

Furkan Gözükara SECourses•2/15/23, 8:18 PM

thank you. any more suggestions?

AArmando Hello, Hello, Quick question does anyone know is there a Formula for Embedding ...

Furkan Gözükara SECourses•2/15/23, 8:19 PM

I would first test 1 batch size 1 gradient and default learning rate

Furkan Gözükara SECourses•2/15/23, 8:19 PM

you tried that?

Furkan Gözükara SECourses•2/15/23, 8:19 PM

the batch size and gradient actually changes how model learns . it is mini batches vs full batches. which one better is debated

FFurkan Gözükara SECourses you tried that?

Armando•2/15/23, 8:20 PM

Yeah, Tried with 2 Vectors but seems very inflexible

Armando•2/15/23, 8:21 PM

v5 for 20 input images working slyly better but not quite there yet

Furkan Gözükara SECourses•2/15/23, 8:23 PM

ye it requires experimenting

Armando•2/15/23, 8:24 PM

I'll try more combinations and see what it comes up with.

FFurkan Gözükara SECourses hello. could you install directly in C drive? still looks to me as path permissi...

GP•2/15/23, 9:37 PM

Hi! I think somehow the models didn't download properly, they were only just over 1gb when I did a git pull, but 5gb when I downloaded each individually

GGP Hi! I think somehow the models didn't download properly, they were only just ove...

Furkan Gözükara SECourses•2/15/23, 9:46 PM

ah so fixed now?

Furkan Gözükara SECourses•2/15/23, 9:46 PM

when you do git pull

Furkan Gözükara SECourses•2/15/23, 9:46 PM

you have to wait a lot

GP•2/15/23, 9:46 PM

I followed your suggestion of watching the ethernet, but must have biffed it

GP•2/15/23, 9:47 PM

Thank's for you video btw, super helpful

GGP Thank's for you video btw, super helpful

Furkan Gözükara SECourses•2/15/23, 9:47 PM

glad to help

GP•2/15/23, 9:47 PM

and thanks for taking the time to respond!

GGP Hi! I think somehow the models didn't download properly, they were only just ove...

Hurlecke•2/15/23, 9:54 PM

I think you need to install git LFS to download large files

Hurlecke•2/15/23, 9:54 PM

With git

Furkan Gözükara SECourses•2/15/23, 9:55 PM

it still downloads

Furkan Gözükara SECourses•2/15/23, 9:55 PM

but doesnt show progress

Furkan Gözükara SECourses•2/15/23, 9:55 PM

by the way i installed gitbash

Furkan Gözükara SECourses•2/15/23, 9:56 PM

maybe that is why i am able to

baban890•2/15/23, 10:07 PM

how to use controlnet in colab ?

Bbaban890 how to use controlnet in colab ?

Furkan Gözükara SECourses•2/15/23, 10:16 PM

if there is a colab supporting automatic1111 could be possible but i havent tested yet

Furkan Gözükara SECourses•2/15/23, 10:17 PM

i plan to show in runpodio

Furkan Gözükara SECourses•2/15/23, 10:37 PM

guys give me more ideas about runpod

Furkan Gözükara SECourses•2/15/23, 10:37 PM

Greetings everyone. In this video I am going to Show how to use Automatic1111 Web UI for stable diffusion tasks on runpod io like you are using it on your computer. I will cover many topics such as how to upload and download files quickly, how to delete directories, how to install and run extensions, how to quickly download and run custom models, how to do dreambooth training on sd 1.5 and sd 2.1 versions, how to do fine tuning via dreambooth, how to do textual inversion training. I will also explain how their pricing system works, how you can use bidding, how you can transfer your files from 1 pod to another pod. I will also demonstrate how you can use new famous controlnet on runpod io.

Hey all, trying out ControlNet from the repo, with the prompt `photo of an owl`. DDIM 30 steps. `con

Similar Threads