Software Engineering Courses (SECourses)•3y ago

I have entered discord, because I was a teacher, and they are the best tutorials for the use and ins

I have entered discord, because I was a teacher, and they are the best tutorials for the use and installation of stable diffusion that I have seen of all the ones on youtube, not only for the explanations of how to do the things, but there is an addition where You explain, for example, in dreambooth the parameters for which they are used in detail, then you also apply different resolutions and compare the results.
It is let's say a very scientific methodology, congratulations

Zono50•5/4/23, 5:03 PM

is having 200 images in my dataset using 5 class reg images the same as using 10 images in my dataset using 5 class reg images

Zono50•5/4/23, 5:03 PM

like how do i get it to learn every intricate detail of someones face

Iivaxsirc -4090 I have entered discord, because I was a teacher, and they are the best tutorials...

Furkan Gözükara SECourses•5/4/23, 5:05 PM

thank you so much

Furkan Gözükara SECourses•5/4/23, 5:05 PM

yes I am also a teacher. Assistant professor in a private university in Turkey

Furkan Gözükara SECourses•5/4/23, 5:06 PM

new video almost ready. waiting youtube to synch subtitles hopefully

Zono50•5/4/23, 5:24 PM

I did a search on stable diffusion, saw an article on linkedin, guess whose it was

Zono50•5/4/23, 5:34 PM

is the only point of the num class instancers per image or regularization images is to preserve the class?

Zono50•5/4/23, 5:34 PM

so if it overfits, if i type woman, it will only show the woman in my dataset, and no other women?

Zono50•5/4/23, 5:38 PM

can you explain doc?

Zono50•5/4/23, 5:43 PM

@Furkan Gözükara SECourses

ZZono50 is the only point of the num class instancers per image or regularization images...

Furkan Gözükara SECourses•5/4/23, 5:53 PM

not only that

Furkan Gözükara SECourses•5/4/23, 5:53 PM

it also prevents overtraining

ZZono50 so if it overfits, if i type woman, it will only show the woman in my dataset, a...

Furkan Gözükara SECourses•5/4/23, 5:54 PM

if you train very long, even when you dont type anything you get your trained subject

Furkan Gözükara SECourses•5/4/23, 5:54 PM

because unet becomes cooked

Zono50•5/4/23, 5:57 PM

overtraining is when the images start looking warped?

ZZono50 overtraining is when the images start looking warped?

Furkan Gözükara SECourses•5/4/23, 5:58 PM

usually

TheRustingDeadAdmin•5/4/23, 6:00 PM

I have a question, when you use the "ohwx man" and the <lora> to generate an image, are you still increasing the weights using ctrl+up e.g. (ohwx man:1.5) or are you keeping it at 1.1?
2nd question; Do we need both [the lora and the ohwx man] everytime we want to call our model?

TTheRustingDeadAdmin I have a question, when you use the "ohwx man" and the <lora> to generate an ima...

Furkan Gözükara SECourses•5/4/23, 6:01 PM

i usually test this because sometimes 1.1 or 1.2 works better

Furkan Gözükara SECourses•5/4/23, 6:01 PM

so you need to test

Furkan Gözükara SECourses•5/4/23, 6:01 PM

2nd : yes

TheRustingDeadAdmin•5/4/23, 6:02 PM

thank you!
I am seeing great results with 1.3 but gets worse higher than that
can you increase the weights on the <lora>? what effect would that have?

TTheRustingDeadAdmin thank you! I am seeing great results with 1.3 but gets worse higher than that ca...

Furkan Gözükara SECourses•5/4/23, 6:08 PM

actually i didnt test increased weight of lora but you can test easily with sr prompt

FFurkan Gözükara SECourses actually i didnt test increased weight of lora but you can test easily with sr p...

TheRustingDeadAdmin•5/4/23, 6:09 PM

Alright, thanks again!

Zono50•5/4/23, 6:34 PM

what's the best way to impose someones face using controlnet?

Zono50•5/4/23, 7:05 PM

i like codeformer but it seems to change the face a little too much, should i lower the weight? or try gfpgan

FFurkan Gözükara SECourses new video almost ready. waiting youtube to synch subtitles hopefully

ivaxsirc -4090OP•5/4/23, 7:10 PM

Ahhh! and a good teacher. Because good teachers explain things as if they were learning the first time.
It is not easy to explain things as if you were explaining them to yourself, when you are already an expert in the field.
But for students it is the best way for them to learn.
And in your tutorials you can see this way of teaching

Iivaxsirc -4090 Ahhh! and a good teacher. Because good teachers explain things as if they were l...

Furkan Gözükara SECourses•5/4/23, 7:20 PM

Thanks agreed

Rae•5/4/23, 8:10 PM

Rae•5/4/23, 8:10 PM

any alternatives to kohya for LoRa training?

Rae•5/4/23, 8:11 PM

or anyone able to help me with my 4090 just getting 3.5it/s?

RRae or anyone able to help me with my 4090 just getting 3.5it/s?

Furkan Gözükara SECourses•5/4/23, 8:25 PM

kohya is slow

Furkan Gözükara SECourses•5/4/23, 8:25 PM

dreambooth extension of web ui much better

Furkan Gözükara SECourses•5/4/23, 8:25 PM

also better quality

Rae•5/4/23, 8:25 PM

Furkan Gözükara SECourses•5/4/23, 8:26 PM

have you seen my videos?

Rae•5/4/23, 8:26 PM

i think i just watched one

Rae•5/4/23, 8:26 PM

on kohya

Rae•5/4/23, 8:27 PM

but can i do LoRa in dreambooth extension of automatic1111?

Rae•5/4/23, 8:30 PM

also is 16it/s in line with what to expect from a single 4090 in automatic1111?

AlexGilSeg•5/4/23, 8:51 PM

@Furkan Gözükara SECourses hello. I just watched your last video. Amazing stuff. One question. You said you trained The model based of The settings in that video but then you said you generated The images using realistic vision. I dont understand how that works.. Or did you train The images on realistic, because in The video you use The standard model..

RRae but can i do LoRa in dreambooth extension of automatic1111?

Furkan Gözükara SECourses•5/4/23, 8:52 PM

yes you can do

RRae also is 16it/s in line with what to expect from a single 4090 in automatic1111?

Furkan Gözükara SECourses•5/4/23, 8:52 PM

16 it is low

Furkan Gözükara SECourses•5/4/23, 8:52 PM

have you watch this video and compared same settings? https://youtu.be/lgP1LNnaUaQ

YouTubeSECourses

RTX 3090 Ti vs RTX 3060 Ultimate Showdown for Stable Diffusion, ML,...

Discord: https://bit.ly/SECoursesDiscord. RTX 3090 review along with RTX 3060. Stable Diffusion, OpenAI's Whisper, DaVinci Resolve and FFmpeg benchmarked. 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

Summary And Conclusions PDF

...

RTX 3090 Ti vs RTX 3060 Ultimate Showdown for Stable Diffusion, ML,...

AAlexGilSeg @Furkan Gözükara SECourses hello. I just watched your last video. Amazing stuff....

Furkan Gözükara SECourses•5/4/23, 8:53 PM

ye i used realistic vision with same configuration. so classification images and used model was different only

Furkan Gözükara SECourses•5/4/23, 8:55 PM

anyone reading can you please upvote i appreciate it : https://www.reddit.com/r/StableDiffusion/comments/137unbi/novel_how_to_find_best_stable_diffusion_any_ai/

Furkan Gözükara SECourses•5/4/23, 9:25 PM

what video i should i make next

Rae•5/4/23, 9:28 PM

steering, biased

Rae•5/4/23, 9:29 PM

can you passivly train with lora captions? like using the captions of the images which dont all represent the same object but the same theme make that theme bleed into the composition even the lora was focused on a very different object.

RRae can you passivly train with lora captions? like using the captions of the images...

Furkan Gözükara SECourses•5/4/23, 9:30 PM

sure you can train with captions. i prefer captions only if fine tuning model

Rae•5/4/23, 9:31 PM

na iam not about training with captions but passivly train the lora and composition to be biased

I have entered discord, because I was a teacher, and they are the best tutorials for the use and ins

Similar Threads