Software Engineering Courses (SECourses)•3y ago

I saw this workflow today

MmonsieurBoutte here's a great short series on all things git 101 https://www.youtube.com/watch?...

HaashimOP•7/22/23, 8:00 PM

THank you

HHaashim I saw this workflow today

Furkan Gözükara SECourses•7/22/23, 8:02 PM

i mean the controlnet model files wouldnt be compatiable but i didnt test

FFurkan Gözükara SECourses i mean the controlnet model files wouldnt be compatiable but i didnt test

HaashimOP•7/22/23, 8:04 PM

I see, and I think if they trained a 0.9 SDXL version they would have to retrain for 1.0

Shack•7/22/23, 8:07 PM

thanks @Dr. Furkan Gözükara and @ashleyk for response, from what age do you think i should start using ohwx girl or boy? 18yrs and below?

SShack thanks @Dr. Furkan Gözükara and @ashleyk for response, from what age do you thi...

ashley•7/22/23, 8:09 PM

I wouldn't train minors, Civit AI doesn't even accept models that contain minor content.

HHaashim I see, and I think if they trained a 0.9 SDXL version they would have to retrain...

Furkan Gözükara SECourses•7/22/23, 8:10 PM

probably

SShack thanks @Dr. Furkan Gözükara and @ashleyk for response, from what age do you thi...

Furkan Gözükara SECourses•7/22/23, 8:10 PM

i think below like 15 age

Aashley I wouldn't train minors, Civit AI doesn't even accept models that contain minor ...

Furkan Gözükara SECourses•7/22/23, 8:11 PM

i think he wants to train his personal photo but not sure

Shack•7/22/23, 8:11 PM

thanks guys, my intention is to also create an image of my son as a superhero

monsieurBoutte•7/22/23, 8:32 PM

getting this issue when trying to add the control net extension extension access disabled because of command line flagsextension access disabled because of command line flags

monsieurBoutte•7/22/23, 8:33 PM

--share--share is killing me

monsieurBoutte•7/22/23, 8:34 PM

but I think I need --share--share to run the webUI interface from runpods terminal

monsieurBoutte•7/22/23, 8:36 PM

I guess I'll try --ngrok--ngrok

monsieurBoutte•7/22/23, 8:38 PM

yeah ngrok seems to work

MmonsieurBoutte `--share` is killing me 🤦‍♂️

ashley•7/22/23, 8:43 PM

This is a security feature, otherwise if you publicly expose the Web UI, attackers can install malware on your Web UI instance using the feature to install extensions.

ogkai•7/22/23, 9:17 PM

Loras are crazy with SDXL. I just trained my first lora last night thanks to Furkan Hocam and getting great results.

ogkai•7/22/23, 9:18 PM

Even without prompting drawing or coloring book, it gets the style very easily and nicely. Thanks again

Thavidu•7/22/23, 9:56 PM

In the Lora training video (https://www.youtube.com/watch?v=TpuDOsuKIBo) I found the tabs at the top are different (mine say Dreambooth, Lora, Textual Inversion instead of Dreambooth, Dreambooth Lora, Dreambooth TI as in the video)-- more importantly, my Lora tab doesn't have instance prompt and class prompt in it (pretty much everything else is there). Those two fields only show up under the Dreambooth tab (not the Lora tab)-- I'm guessing my version might be newer and the UI has changed, but I'm not sure how to set the instance and class prompt to the 'ohwx man' thats used in the video

YouTubeSECourses

Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffu...

#Kohya SS web GUI DreamBooth #LoRA training full tutorial. You don't need technical knowledge to follow this tutorial. In this tutorial I have explained how to generate professional photo studio quality portrait / self images for free with Stable Diffusion training.

Our Discord server

https://bit.ly/SECoursesDiscord

If I have been of assis...

Thavidu•7/22/23, 9:56 PM

@Dr. Furkan Gözükara sorry forgot to tag

monsieurBoutte•7/22/23, 9:58 PM

I finally got something working but my results are still pretty rough compared to clipdrop's uncrop

monsieurBoutte•7/22/23, 9:58 PM

any thoughts one what I can try adjusting in my params?

monsieurBoutte•7/22/23, 9:59 PM

I'm also curious if anyone suggests a different sampling method other than Euler aEuler a

TThavidu In the Lora training video (https://www.youtube.com/watch?v=TpuDOsuKIBo) I found...

Furkan Gözükara SECourses•7/22/23, 10:29 PM

hello welcome

Furkan Gözükara SECourses•7/22/23, 10:29 PM

here more up to date interface tutorial

Furkan Gözükara SECourses•7/22/23, 10:29 PM

https://youtu.be/AY6DMBCIZ3A

YouTubeSECourses

First Ever SDXL Training With Kohya LoRA - Stable Diffusion XL Trai...

How to install #Kohya SS GUI trainer and do #LoRA training with Stable Diffusion XL (#SDXL) this is the video you are looking for. I have shown how to install Kohya from scratch. The best parameters to do LoRA training with SDXL. How to use Kohya SDXL LoRAs with ComfyUI. How to do checkpoint comparison with SDXL LoRAs and many more cool stuff.

...

Furkan Gözükara SECourses•7/22/23, 10:29 PM

this one is for sdxl but you will see tabs in this one

Thavidu•7/22/23, 10:29 PM

thanks ill try that

Furkan Gözükara SECourses•7/22/23, 10:30 PM

also sdxl much better than sd 1.5

Furkan Gözükara SECourses•7/22/23, 10:30 PM

but requires more vram

MmonsieurBoutte any thoughts one what I can try adjusting in my params?

Furkan Gözükara SECourses•7/22/23, 10:30 PM

what is it that your doing ? i dont know clicpdrops uncrop :d

Thavidu•7/22/23, 10:30 PM

i think the base models i want to use are on sd1.5 unfortunately, but i'll try use the video to figure out interface

Original message was deleted

Furkan Gözükara SECourses•7/22/23, 10:34 PM

look here : https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Useful-Commands.md#download-civitai-model

GitHub

Stable-Diffusion/Useful-Commands.md at main · FurkanGozukara/Stable...

Stable Diffusion, SDXL, LoRA Training, DreamBooth Training, Automatic1111 Web UI, DeepFake, Deep Fakes, TTS, Animation, Text To Video, Tutorials, Guides, Lectures, Courses, ComfyUI, Google Colab, R...

FFurkan Gözükara SECourses look here : https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Useful-...

xlads•7/22/23, 10:34 PM

thank u

Shack•7/22/23, 11:03 PM

Thanks for the help so far, this is progress for me. Prompt for first 4 images is "ohwx woman, photo portrait, in tomer hanuka style" and for the 2nd 4 images is "a photo of ohwx woman" ,used Automatic1111 locally to generate. The last image shows training settings for 10 images. Is this good or can be improved?

SShack Thanks for the help so far, this is progress for me. Prompt for first 4 images i...

Furkan Gözükara SECourses•7/22/23, 11:03 PM

looking pretty cool. you are using shivam colab?=

Shack•7/22/23, 11:04 PM

ShivamShriraoDreamBooth

Half-Saint•7/22/23, 11:08 PM

hi guys, i am following niko's tutorial for running dreambooth but i am getting trainer is not defined error. any idea what that is?

SShack ShivamShriraoDreamBooth

Furkan Gözükara SECourses•7/22/23, 11:08 PM

reeally hard to improve there

Furkan Gözükara SECourses•7/22/23, 11:08 PM

do you have bigger google drive spacE?

HHalf-Saint hi guys, i am following niko's tutorial for running dreambooth but i am getting ...

Furkan Gözükara SECourses•7/22/23, 11:08 PM

please follow my tutorials :d

Furkan Gözükara SECourses•7/22/23, 11:09 PM

https://www.youtube.com/@SECourses/search?query=dreambooth

YouTube

SECourses

Welcome to Software Engineering Courses (SECourses) – the ultimate destination for skillfully curated insights into state-of-the-art technologies and programming paradigms. We demystify the realms of Artificial Intelligence, Stable Diffusion, DreamBooth, LoRA, ControlNet, Textual Inversion, Software Engineering, Programming, C#, .NET, ASP .NET, ...

Shack•7/22/23, 11:12 PM

@Dr. Furkan Gözükara yes i have space

SShack @Dr. Furkan Gözükara yes i have space

Furkan Gözükara SECourses•7/22/23, 11:20 PM

ok then you can do more frequent checkpoint

Furkan Gözükara SECourses•7/22/23, 11:20 PM

and compare checkpoints

Shack•7/22/23, 11:26 PM

By "more frequent" you mean increasing training steps?

Thavidu•7/22/23, 11:29 PM

This is my first time actually pressing the training button in kohya_ss gui (after following tutorial), but i get "RuntimeError: Error(s) in loading state_dict for AutoencoderKL:
Unexpected key(s) in state_dict: "encoder.mid_block.attentions.0.to_to_k.bias",
"encoder.mid_block.attentions.0.to_to_k.weight"... " any ideas?

Thavidu•7/22/23, 11:35 PM

ah maybe it is because of my base model-- switching to the same one from the tutorial works (realistic vision). The one i was using before fails after loading u-net (or between loading u-net and loading vae)

Thavidu•7/22/23, 11:36 PM

SShack By "more frequent" you mean increasing training steps?

Furkan Gözükara SECourses•7/22/23, 11:47 PM

there is save checkpoints steps count

Furkan Gözükara SECourses•7/22/23, 11:47 PM

you lower it