Software Engineering Courses (SECourses)•3y ago

is it me you're looking for

ZZono50 is it me you're looking for

Aldo Jones•6/8/23, 2:03 PM

Everybody could answer to my question

AAldo Jones Everybody could answer to my question 🙂

ashley•6/8/23, 2:05 PM

If this is on Windows, its already been answered in #programming

Aashley If this is on Windows, its already been answered in #programming

Aldo Jones•6/8/23, 2:07 PM

Cool... but I'm not that much skilled with this stuff. Should I install insightface usind Conda?

AAldo Jones Hello. I'm installing DeepFace software following this tutorial: Mind-Blowing De...

Furkan Gözükara SECourses•6/8/23, 2:08 PM

c++ tools error

Furkan Gözükara SECourses•6/8/23, 2:08 PM

you need to install

FFurkan Gözükara SECourses you need to install

Aldo Jones•6/8/23, 2:09 PM

Install what?

Aldo Jones•6/8/23, 2:09 PM

Aldo Jones•6/8/23, 2:09 PM

I did this. Do you think is ok now?

Aldo Jones•6/8/23, 2:09 PM

Or should I do this from ROOP ambience?

ashley•6/8/23, 2:10 PM

programming

ashley•6/8/23, 2:10 PM

"you can install visual studio community edition as well with c++ libs"

AAldo Jones I did this. Do you think is ok now?

ashley•6/8/23, 2:11 PM

Not ok, it didn't install

Aldo Jones•6/8/23, 2:11 PM

Ok, now I try

Aldo Jones•6/8/23, 2:12 PM

So in C++ libs should also be insightface?

ashley•6/8/23, 2:13 PM

Too many questions about #roop in #stable-diffusion-flux-such

, should move to #ai or get its own channel.

Aashley Too many questions about #roop in #stable-diffusion-flux-such 🙈 , should move t...

Aldo Jones•6/8/23, 2:15 PM

I agree

Aldo Jones•6/8/23, 2:15 PM

But people always asks stuff in more active channels

ashley•6/8/23, 2:16 PM

The other channels would be active too if people didn't ask absolutely everything in this one

Aldo Jones•6/8/23, 2:16 PM

Furkan Gözükara SECourses•6/8/23, 2:34 PM

c++ tools video added for roop : https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/1-Click-DeepFake-Tutorial.md

GitHub

Stable-Diffusion/1-Click-DeepFake-Tutorial.md at main · FurkanGozuk...

Best Stable Diffusion and AI Tutorials, Guides, News, Tips and Tricks - Stable-Diffusion/1-Click-DeepFake-Tutorial.md at main · FurkanGozukara/Stable-Diffusion

Zono50OP•6/8/23, 2:46 PM

little bit confused on captions, heard people say that captions are what you DON'T want the AI to learn, but then when i use captions to generate class images, it always generates images using my captions

Zono50OP•6/8/23, 2:51 PM

been having too much fun with roop, gonna turn myself into goldberg today

brenkhof•6/8/23, 3:23 PM

What is the difference between the pre installed Lora in Stable Diffusion and Kohya LoRA?

Bbrenkhof What is the difference between the pre installed Lora in Stable Diffusion and Ko...

ashley•6/8/23, 3:27 PM

First 4 trained by Kohya_ss, last one trained by Dreabooth extension for A1111.

brenkhof•6/8/23, 3:29 PM

Thanks for sharing

brenkhof•6/8/23, 3:29 PM

I don't see a huge difference

ashley•6/8/23, 3:30 PM

What? you must be blind, the last one is MUCH better. The middle 3 are really bad in comparison.

ashley•6/8/23, 3:30 PM

1st one is similar to the last one, but the last one is still better.

brenkhof•6/8/23, 3:32 PM

So that means you don't recommend Kohya in general?

Bbrenkhof So that means you don't recommend Kohya in general?

ashley•6/8/23, 3:34 PM

No, use whatever works best for you. A1111 extension's main branch is broken so I got better results with Kohya_ss for full dreambooth training. Just use whatever you get the best results with and feel the most comfortable with.

ashley•6/8/23, 3:35 PM

I was just using that image as an example of someone comparing the 2, and he chooses Dreambooth over Kohya_ss for that very reason, but if you prefer Kohya_ss then keep using it

brenkhof•6/8/23, 3:36 PM

I see. Yeah I'm trying to use Lora just for clothing and apply it to a style dreambooth model. But havn't tried lora yet and wanted to ask first

brenkhof•6/8/23, 3:36 PM

Thanks for sharing @ashleyk

Bbrenkhof Thanks for sharing @ashleyk

ashley•6/8/23, 3:37 PM

I suggest chatting to Anzhc on the d8ahazard Dreambooth Discord (A1111 Dreambooth extension Discord). He is an expert at training LoRA.

ashley•6/8/23, 3:37 PM

Those images I shared were made by him.

brenkhof•6/8/23, 3:39 PM

Great help my friend. I joined and might ask him later. Thank you so much!

Zono50OP•6/8/23, 4:18 PM

With lora training, if i want the AI to learn a womans eyes, would i put her eye color as a caption?

ZZono50 With lora training, if i want the AI to learn a womans eyes, would i put her eye...

felixsanz•6/8/23, 5:01 PM

you only caption what is not part of the subject

Zono50OP•6/8/23, 5:11 PM

oh ok good, than i'm doing it right

Furkan Gözükara SECourses•6/8/23, 5:42 PM

I have a great plan. If works planning to release huge fine tuning dataset and do fine training tutorial

Furkan Gözükara SECourses•6/8/23, 5:42 PM

Currently at the gym

Zono50OP•6/8/23, 5:44 PM

like the ai thinks my wife had brown eyes, but she has green eyes, would I put brown eyes as a caption? or is that a failure of the dataset itself

Zono50OP•6/8/23, 5:44 PM

that's great @Dr. Furkan Gözükara

ZZono50 like the ai thinks my wife had brown eyes, but she has green eyes, would I put b...

ashley•6/8/23, 6:04 PM

Probably an issue with your data set. Are there lots of distance shots or are they close ups?

Zono50OP•6/8/23, 6:07 PM

they are mostly close up

Zono50OP•6/8/23, 6:07 PM

couple of them are distant shots

ashley•6/8/23, 6:11 PM

Should learn the eyes then

ashley•6/8/23, 6:38 PM

Dreambooth extension is removed from the Runpod image again

ZZono50 like the ai thinks my wife had brown eyes, but she has green eyes, would I put b...

felixsanz•6/8/23, 6:39 PM

just use brown eyes as negative prompt

FrankXXXtremo•6/8/23, 6:50 PM

Hello dear Stable Diffusion community!! And thank you for let me take part. Excuse me.. Do you know if any Kaggle notebook works to train or create a checkpoint file model with Dreambooth? I would like to create a checkpoint file model with a dataset of images of myself in Kaggle. I ve tried a couple of Dreambooth extensions in Stable Diffusion WebUIs in Kaggle, but they show errors.

is it me you're looking for

Similar Threads

Similar Threads

Similar Threads