Software Engineering Courses (SECourses)•3y ago

i have this auto installer

Capture_decran_2023-10-24_a_21.25.50.png

VvKorby Oh ok,im thinkin its my computer also been runnin a bit slow lately so going to ...

mikemenders•10/24/23, 7:26 PM

Sorry, there's no --lowvram-sdxl switch, so it's still slow because you're generating to regular memory, so you have full VRAM. You have 8 GB VRAM, I think. This is very marginal.

mikemenders•10/24/23, 7:27 PM

Check this options:
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Command-Line-Arguments-and-Settings

GitHub

Command Line Arguments and Settings

Stable Diffusion web UI. Contribute to AUTOMATIC1111/stable-diffusion-webui development by creating an account on GitHub.

mikemenders•10/24/23, 7:28 PM

I have an RTX 3060 12 GB card and it is also a bit slow for SDXL generation, but it fits in VRAM. With AI, VRAM is very important, how much you have.

mikemenders•10/24/23, 7:29 PM

If you don't have enough VRAM, you're working in RAM, and even though I have DDR5 RAM with i9, it's definitely slower.

BBarbaraM i have this auto installer

Furkan Gözükara SECourses•10/24/23, 7:30 PM

ye after that start web ui manually

Furkan Gözükara SECourses•10/24/23, 7:30 PM

so kill previous one with

Furkan Gözükara SECourses•10/24/23, 7:30 PM

fuser -k 3000/tcp

Mmikemenders If you don't have enough VRAM, you're working in RAM, and even though I have DDR...

vKorby•10/24/23, 7:34 PM

Yeah I do have 8 GB looks like Imma jus have to deal with that lol, I'm thinking of using this and leonardo ai image 2 image

FFurkan Gözükara SECourses shared disk space

EyeSpyBekoAI•10/24/23, 7:57 PM

Hey there, thanks for this! Network volume seems to be working quite well as I was able to see setup pods pop up after being closed and relaunched. I'm setting up more now. One thing I'm noticing though, is that getting port availability is considerably dodgy. Is there a way to get around this, or is it one of "it is what it is" situations?

Mmikemenders @Jas The large file is also good for styling, and can be reduced to 512 by lora....

Jas•10/24/23, 8:22 PM

Thanks for your recommendations, with the labelling, I am a bit uncertain how to do it as they are quite abstract. I will show you a couple of examples of the images I am training on, maybe you can suggest what to label.

Come to think of it, I am not sure I am really training a 'style' but it is more like a type of artwork / art object.

for the label I have something like "Ceramicblobstyle, pink, green and blue theme"

I don't know how to label it outside of mentioning the colours in the image, because I want to replicate the patterns, the shapes and blobs.

EEyeSpyBekoAI Hey there, thanks for this! Network volume seems to be working quite well as I w...

Furkan Gözükara SECourses•10/24/23, 8:34 PM

when you first time mount

Furkan Gözükara SECourses•10/24/23, 8:34 PM

it snych i think

Furkan Gözükara SECourses•10/24/23, 8:34 PM

thus slower

FFurkan Gözükara SECourses for batch size 1 : rtx 4090

akellio•10/24/23, 8:34 PM

Is it possible to use multiple video cards?

Aakellio Is it possible to use multiple video cards?

Furkan Gözükara SECourses•10/24/23, 8:35 PM

for training yes

Furkan Gözükara SECourses•10/24/23, 8:35 PM

please also write in english

akellio•10/24/23, 8:35 PM

Do I need to change any settings?

Aakellio Do I need to change any settings?

Furkan Gözükara SECourses•10/24/23, 8:37 PM

you need to set accelerate to use all gpus i think

Furkan Gözükara SECourses•10/24/23, 8:37 PM

it is on by default on kaggle

Kallamamran•10/24/23, 8:37 PM

If it's just pattern you want to do. Try Inspire in ComfyUI

FFurkan Gözükara SECourses when you first time mount

EyeSpyBekoAI•10/24/23, 8:38 PM

Nothing left to do but wait and test now. Thanks for the response!

EEyeSpyBekoAI 👍 Nothing left to do but wait and test now. Thanks for the response!

Furkan Gözükara SECourses•10/24/23, 8:38 PM

by the way

Furkan Gözükara SECourses•10/24/23, 8:38 PM

there is also usually visual bug

Furkan Gözükara SECourses•10/24/23, 8:38 PM

so still click them

FFurkan Gözükara SECourses it is on by default on kaggle

akellio•10/24/23, 8:38 PM

Already switched to runpod, kaggle is still quite slow

Aakellio Already switched to runpod, kaggle is still quite slow

Furkan Gözükara SECourses•10/24/23, 8:41 PM

accurate

FFurkan Gözükara SECourses so still click them

EyeSpyBekoAI•10/24/23, 8:42 PM

sounds good. I'll keep that in mind. I'm running training now.

Kallamamran•10/24/23, 8:42 PM

I really wish I had access to the "Inspire" like in ComfyUI to use in A1111

Kallamamran•10/24/23, 8:45 PM

Crazy dream•10/24/23, 8:51 PM

Hello everybody,
I am trying to following this tutorial : https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/No-More-Studio-Photoshoot-Realistic-DreamBooth-Training.md
I am at the beginning where I have to install proper requirements for my extension dreambooth. According to the previous link, on windows, we have to :
"

Open a new cmd inside automatic_web_ui\venv\Scripts

Type activate

Then execute below commands 1 by 1

cd..

cd..

cd "extensions\sd_dreambooth_extension"

pip install -r requirements.txt
"
However, I am not using my PC (because I don't have any GPU) so I am using sagemaker studio lab. All is working until now (SD works fine with the specific model) but I don't have any folder venv. What sould I do to install requirements ?

According to the official git of the extension (https://github.com/d8ahazard/sd_dreambooth_extension), we should use the tab but Dr Furkan is telling us in the video to not use it.

GitHub

Stable-Diffusion/Tutorials/No-More-Studio-Photoshoot-Realistic-Drea...

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...

CCrazy dream Hello everybody, I am trying to following this tutorial : https://github.com/Fur...

Furkan Gözükara SECourses•10/24/23, 8:58 PM

ok for sage maker

Furkan Gözükara SECourses•10/24/23, 8:58 PM

you should follow runpod

Furkan Gözükara SECourses•10/24/23, 8:58 PM

tutorial

Furkan Gözükara SECourses•10/24/23, 8:58 PM

https://youtu.be/c_S2kFAefTQ

YouTubeSECourses

How To Install DreamBooth & Automatic1111 On RunPod & Latest Librar...

I have shown how to install latest version of Automatic1111 Web UI for Stable Diffusion and DreamBooth extension of Auto1111 on RunPod in this video. Moreover, I show how to upgrade to latest Cuda, Torch and cuDNN DLL files. With these upgrades the image generation speed literally doubles.

GitHub Readme File

https://github.com/FurkanGozukara...

How To Install DreamBooth & Automatic1111 On RunPod & Latest Librar...

Marcelo•10/24/23, 8:58 PM

@Dr. Furkan Gözükara, what would be the best way to train a person in different ages? Is it a good idea? Should I use captions?

Furkan Gözükara SECourses•10/24/23, 8:58 PM

also use exact versions

MMarcelo @Dr. Furkan Gözükara, what would be the best way to train a person in different ...

Furkan Gözükara SECourses•10/24/23, 8:59 PM

do you have different age photos?

FFurkan Gözükara SECourses also use exact versions

Crazy dream•10/24/23, 8:59 PM

Thank you for your answer! even if I don't have any issue with the current ones ? (I still don't know if it will work but in the video, I understood you advice us to don't use exact version if it works)

Marcelo•10/24/23, 9:00 PM

Yes, in this dataset I have the same person, but in different epochs, young adult, adult....

Marcelo•10/24/23, 9:01 PM

For example, Christian Bale now and 10 years ago

CCrazy dream Thank you for your answer! even if I don't have any issue with the current ones ...

Furkan Gözükara SECourses•10/24/23, 9:07 PM

use exact versions i used in tutorial

FFurkan Gözükara SECourses use exact versions i used in tutorial

Crazy dream•10/24/23, 9:07 PM

ok, thank you very much

_Captain_•10/24/23, 9:11 PM

@Dr. Furkan Gözükara thanks for your research! I'm one of your patreons now

I have a question about text encoder training and captions.
(Using Koyha Dreambooth training / 3090ti - your sdxl dreambooth workflow works really well!)

To my understanding SDXL text encoder is based on CLIP ViT-L & OpenClip ViT-bigG.
So the model understands only words the text encoder contains. So as long I'm training things the text encoder knows, everything is fine (+ you don't have to train the text encoder).

But let's take a special martial arts pose for example. Some of them have uniqe names and using your "SOTA (The Very Best) Image Captioning..." tutorial, it was easy to prove that the prompt won't contain the correct names for the pose.

When training the model using the dreambooth method, every image has it's textfile with
instance, class, image description -> for example: I want to train a chinese kung fu stance named "pubu" and the model has no knowledge of it ->
pubu, man, in a training hall,

NOW! If I train without the textencoder:
What impact have the captions? Is "man, in a training hall" a guide for the noise generation?
Will the new images enhance the previous knowledge of a man in a training hall?

I assume "pubu" has no impact on the training at all? I have no idea how the unet is working with these captions.
Or does the model somehow learn about "pubu"?

Is this the point where you have to train the text encoder? I tried it once with the same settings and the model (text encoder) was waaay overtrained. EDIT: "just found your settings for training with text encoder, will try that!"

Summery Question:
If I use captions the text encoder doesn't know -> I assume I cannot use them later for generating images.
(So I have to train the text encoder too)

Sorry for the long question, don't know how to shorten all this complexe stuff xD

FFurkan Gözükara SECourses use exact versions i used in tutorial

Crazy dream•10/24/23, 9:11 PM

last question sorry : I have to use exact version used in this tutorial https://www.youtube.com/watch?v=g0wXIcRhkJk right ? (not this one https://www.youtube.com/watch?v=c_S2kFAefTQ)

YouTubeSECourses

The END of Photography - Use AI to Make Your Own Studio Photos, FRE...

Dreambooth is the best training method for Stable Diffusion. In this tutorial, I show how to install the Dreambooth extension of Automatic1111 Web UI from scratch. Additionally, I demonstrate my months of work on the realism workflow, which enables you to produce studio-quality images of yourself through #Dreambooth training. Furthermore, I shar...

YouTubeSECourses

How To Install DreamBooth & Automatic1111 On RunPod & Latest Librar...

https://github.com/FurkanGozukara...

Zono50•10/24/23, 9:23 PM

trying to launch kohya, and getting scipy errors, even after installing both with pip and then conda

Mmikemenders Sorry, there's no --lowvram-sdxl switch, so it's still slow because you're gener...

vKorby•10/24/23, 9:37 PM

For some reason now its running pretty quick lol, with 1024 too

akellio•10/24/23, 9:38 PM

Regularization images in runpod training need to be loaded manually?

akellio•10/24/23, 10:05 PM

outfigured out

Crazy dream•10/24/23, 10:20 PM

I have this error :
"/workspace/venv/lib/python3.10/site-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension: '/workspace/venv/lib/python3.10/site-packages/torchvision/image.so: undefined symbol: _ZN3c104cuda9SetDeviceEi'If you don't plan on using image functionality from torchvision.iotorchvision.io, you can ignore this warning. Otherwise, there might be something wrong with your environment. Did you have libjpeglibjpeg or libpnglibpng installed before building torchvisiontorchvision from source?
[...]
OSError: /workspace/venv/lib/python3.10/site-packages/torchaudio/lib/libtorchaudio.so: undefined symbol: _ZN3c104cuda9SetDeviceEi
Relauncher: Process is ending. Relaunching in 2s..." However I am following the tutorial https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/How-To-Install-DreamBooth-Extension-On-RunPod.md step by step. Somebody would know why ?

GitHub

Stable-Diffusion/Tutorials/How-To-Install-DreamBooth-Extension-On-R...

MMarcelo For example, Christian Bale now and 10 years ago

mikemenders•10/24/23, 10:30 PM

Check Age sliders on civitai. These special loras:
https://civitai.com/search/models?sortBy=models_v2&query=age%20slider

i have this auto installer

Similar Threads

Similar Threads

Similar Threads