Software Engineering Courses (SECourses)•2y ago

https://www.reddit.com/r/StableDiffusion/comments/1bujpn3/onetrainer_fine_tuning_vs_kohya_ss_dreambo

https://www.reddit.com/r/StableDiffusion/comments/1bujpn3/onetrainer_fine_tuning_vs_kohya_ss_dreambooth/

From the StableDiffusion community on Reddit: OneTrainer Fine Tunin...

Explore this post and more from the StableDiffusion community

FFurkan Gözükara SECourses yep it is in production

ihsass•4/5/24, 8:06 PM

Thank you sir will waiting fir the release

FFurkan Gözükara SECourses use same resolution as your training images

NAMU•4/5/24, 8:07 PM

I thought because OneTrainer allows different resolutions, it would be the same for regularization images. I'm not to that part of the tutorial yet lol.

NNAMU I thought because OneTrainer allows different resolutions, it would be the same ...

Furkan Gözükara SECoursesOP•4/5/24, 8:07 PM

well there are 2 sceneraios

Furkan Gözükara SECoursesOP•4/5/24, 8:07 PM

either they get full resized

Furkan Gözükara SECoursesOP•4/5/24, 8:07 PM

or resized into buckets

Furkan Gözükara SECoursesOP•4/5/24, 8:07 PM

depending on your setup

Furkan Gözükara SECoursesOP•4/5/24, 8:08 PM

if you manually set all equal size

FFurkan Gözükara SECourses https://www.reddit.com/r/StableDiffusion/comments/1bujpn3/onetrainer_fine_tuning...

ihsass•4/5/24, 8:08 PM

I did upvote

Furkan Gözükara SECoursesOP•4/5/24, 8:08 PM

you get more control

Furkan Gözükara SECoursesOP•4/5/24, 8:08 PM

i prefer 1024x1024

Iihsass I did upvote

Furkan Gözükara SECoursesOP•4/5/24, 8:08 PM

thank you

NAMU•4/5/24, 8:13 PM

Just for clarification, your regularization images are mostly of people of color, and I'm training an east Asian and white person mostly. The ethnicity of the people in the regularization images don't have an effect on the training, right? Regularization images are for help with poses?

Edit: just looked at the latest folder of regularization images and it's a really good mix of different ethnicities, so I guess you can ignore this question, lol.

NNAMU Just for clarification, your regularization images are mostly of people of color...

Furkan Gözükara SECoursesOP•4/5/24, 8:30 PM

well as much as deverse better

Furkan Gözükara SECoursesOP•4/5/24, 8:30 PM

yes i tried to collect from every nation

NAMU•4/5/24, 8:31 PM

You did a great job.

NAMU•4/5/24, 8:31 PM

I honestly don't know where you get the energy and drive to do all these things. It must have taken forever just to gather all those images!

NNAMU I honestly don't know where you get the energy and drive to do all these things....

Furkan Gözükara SECoursesOP•4/5/24, 8:36 PM

i am full time spending on these right now

so patreon support is my almost whole revenue

FFurkan Gözükara SECourses sadly people didnt even upvote

kareem9000•4/5/24, 8:40 PM

i gotchuu

kareem9000•4/5/24, 8:40 PM

2 upvotes fromf both my accs

Kkareem9000 2 upvotes fromf both my accs 🤣

NAMU•4/5/24, 9:03 PM

one from me, as well lol

NAMU•4/5/24, 9:04 PM

@Dr. Furkan Gözükara Maybe every time you make a new post on reddit, you send out a server-wide note and ask people to upvote it?

NNAMU @Dr. Furkan Gözükara Maybe every time you make a new post on reddit, you send ou...

Furkan Gözükara SECoursesOP•4/5/24, 9:10 PM

nah this is fine. nevermind reddit :d

NAMU•4/5/24, 9:13 PM

With an RTX 4070 12gb, training 1024x at float16 using your tier1sdxl 10gb config, do you know a rough estimate of how long it would take to train using 25 images? Just a ballpark figure. I'm wondering if I should I set it and leave the house to run errands for a few hours, or if I should set it and leave it to run overnight.

NNAMU With an RTX 4070 12gb, training 1024x at float16 using your tier1sdxl 10gb confi...

Furkan Gözükara SECoursesOP•4/5/24, 9:20 PM

you can calculate

Furkan Gözükara SECoursesOP•4/5/24, 9:20 PM

what is your second / it

NAMU•4/5/24, 9:21 PM

Around 3.6 to 3.9

NNAMU Around 3.6 to 3.9

Furkan Gözükara SECoursesOP•4/5/24, 9:46 PM

ok lets say 4 at worst

Furkan Gözükara SECoursesOP•4/5/24, 9:46 PM

so 25 images with lets say 140 epoch

Furkan Gözükara SECoursesOP•4/5/24, 9:46 PM

25 * 140 * 2 = 7000 steps

Furkan Gözükara SECoursesOP•4/5/24, 9:47 PM

so it would be 7000 * 4 / 60 / 60 = 7.8 hours

NAMU•4/5/24, 10:35 PM

Overnight it is, then! Thanks for doing the math for me. I can do the math, I just don't enjoy the process, lol.

NAMU•4/5/24, 10:36 PM

Although looking at cmd, it looks like I'm averaging around 2s/it - so maybe if I'm lucky, around 4 hours

NAMU•4/6/24, 12:19 AM

So I stopped the training at 3100 and tested the saved model, and if I use all the common tags from the training in the same prompt, I get a pretty damned good result. But if I remove all or even just some of those common tags, or if I try to use any kind of creative prompt (like "sitting in a golden throne wearing golden armor"), the face likeness goes away.

Furkan Gözükara SECoursesOP•4/6/24, 1:07 AM

recording onetrainer tutorial already over 2 hours

NNAMU So I stopped the training at 3100 and tested the saved model, and if I use all t...

Furkan Gözükara SECoursesOP•4/6/24, 1:08 AM

did you train with captions?

Furkan Gözükara SECoursesOP•4/6/24, 1:08 AM

you can see caption effect : https://www.patreon.com/posts/compared-effect-101206188

Patreon

Compared Effect Of Image Captioning For SDXL Fine-tuning / DreamBoo...

Get more from SECourses: Tutorials, Guides, Resources, Training, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözükara on Patreon

MMorpheus SX343 I guess the base model only knows DDD extra large boobs 😄 But I dont want these...

Neo•4/6/24, 1:15 AM

Yes, that means the model you're using for inference of that word "boob" is still trained "stronger" than your LoRA concept. If you're training say the concept "girlfriend woman" with captions on your dataset you could leave out any reference to breasts. However it may make your LoRA less flexible. You could also experiment with doing the caption dropout option (available in Kohya) or for an experiment train your LoRA with no captions and see if the accuracy increases...

crapthings•4/6/24, 1:31 AM

hello i have a question
i have 2000 reg images

which should i use？
30 images 1 reg
or
30 images 30 reg

FFurkan Gözükara SECourses you can see caption effect : https://www.patreon.com/posts/compared-effect-10120...

NAMU•4/6/24, 1:46 AM

Ahhh, that makes sense! I created the training data before watching the tutorial, so I forgot that the caption text was included in the folder with the training images already. Damn it, back to the drawing board, lol.

NAMU•4/6/24, 1:48 AM

@Dr. Furkan Gözükara Someone recomended removing backgrounds for training data images, do you agree with them? I can do that pretty easily.

NAMU•4/6/24, 1:49 AM

Or I can use the tool to add the masks

NAMU•4/6/24, 1:53 AM

like this

NNAMU @Dr. Furkan Gözükara Someone recomended removing backgrounds for training data ...

EyeSpyBekoAI•4/6/24, 2:11 AM

I can chime in on that a bit. Personally, that gives me disastrous results, but give it a shot if you'd like

NNAMU like this

Furkan Gözükara SECoursesOP•4/6/24, 2:15 AM

yes

Furkan Gözükara SECoursesOP•4/6/24, 3:57 AM

still recording and tutorial is now over 3 hours. this tutorial actually a standalone full course haha

Furkan Gözükara SECoursesOP•4/6/24, 4:12 AM

other than introduction here the tutorial durations :d onetrainer

Furkan Gözükara SECoursesOP•4/6/24, 4:13 AM

i need to record introduction and edit it now

FFurkan Gözükara SECourses still recording and tutorial is now over 3 hours. this tutorial actually a stand...

EyeSpyBekoAI•4/6/24, 4:56 AM

Allah razı olsun hocam!!!

Furkan Gözükara SECoursesOP•4/6/24, 4:58 AM

good day / night

EyeSpyBekoAI•4/6/24, 5:00 AM

Seriously though you’re probably sick of me saying this over and over but I and I’m sure this community is very thankful for everything you do. I’ve learned a ton here.

https://www.reddit.com/r/StableDiffusion/comments/1bujpn3/onetrainer_fine_tuning_vs_kohya_ss_dreambo

Similar Threads

Similar Threads

Similar Threads