Full Stable Diffusion SD & XL Fine Tuning Tutorial With OneTrainer On Windows & Cloud - Zero To Hero

In this tutorial, I am going to show you how to install OneTrainer from scratch on your computer and do a Stable Diffusion SDXL (Full Fine-Tuning 10.3 GB VRAM) and SD 1.5 (Full Fine-Tuning 7GB VRAM) based models training on your computer and also do the same training on a very cheap cloud machine from MassedCompute if you don't have such computer.

SDXL Configs Updated to Better Version After 20 New Experiment ⤵️

Tutorial Readme File ⤵️

Register Massed Compute From Below Link (could be necessary to use our Special Coupon for A6000 GPU for 31 cents per hour) ⤵️

Coupon Code for A6000 GPU is : SECourses

0:00 Introduction
3:54 Intro to instructions GitHub readme
4:32 How to register Massed Compute (MC) and start virtual machine (VM)
5:48 Which template to choose on MC
6:36 How to apply MC coupon
8:41 How to install OT on your computer to train
9:15 How to verify your Python, Git, FFmpeg and Git installation
12:00 How to install ThinLinc and start using your MC VM
12:26 How to setup folder synchronization and file sharing computer and MC VM
13:56 End existing session in ThinClient
14:06 How to turn off MC VM
14:24 How to connect and start using VM
14:41 When use end existing session
16:38 How to download very best OT preset training configuration for SD 1.5 & SDXL models
18:00 How to load configuration preset
18:38 Full explanation of OT configuration and best hyper params for SDXL
24:10 How to setup training concepts accurately in OT
24:52 How to caption images for SD training
30:17 Why my training images dataset is not great and what is a better dataset
31:41 How to make DreamBooth effect in OT with reg images concept
32:44 Effect of using ground truth regularization images dataset
34:41 How to set regularization images repeating
35:55 Explanation of training tab configuration and parameters
41:58 What does masked training do and how to do masked training and generate masks
44:53 Generate samples during training setup
46:05 How to save checkpoints during training to compare and find best one later
47:11 How to save your configuration in OT
47:22 How to install and utilize nvitop to see VRAM usage
48:06 Why super slow training happens due to shared VRAM and how to fix it
48:40 How to reduce VRAM usage before starting training
49:01 Start training on Windows
49:11 Starting to setup everything on MC same as on Windows
49:37 Upload data to MC
51:11 Update OT on MC
52:33 How to download regularization images
53:42 How to minimize all windows on MC
54:00 Start OT on MC
54:20 Setting everything on MC same as Windows
55:22 How to set folders on MC VM
56:31 How to properly crop and resize your training images
57:47 Accurate Auto1111 Models folder on MC
58:05 Copy file & folder path on MC
58:54 All of the rest of the config on MC
1:03:29 How to utilize second GPU if you have
1:05:45 Checking back again our Windows training
1:06:06 How to use Automatic1111 (A1111) SD Web UI on MC and Windows
1:11:35 How to use default Python on MC
1:11:55 Checking training speed and explaining what it means
1:12:13 How many steps we are going to train explanation
1:13:40 First checkpoint and howe checkpoints named
1:14:15 How to fix A1111 errors
1:15:44 How to start A1111 Web UI and use it with Gradio Live share and locally
1:17:45 What to do if model loading takes forever on Gradio and how to fix it
1:19:01 Where to see status of the training of OT
1:19:43 How to upload checkpoints / anything into Hugging Face for permanently saving
1:26:21 How to auto upgrade A1111 and install ADetailer and ControlNet extensions
1:29:10 How to use trained model checkpoints on Massed Compute
1:30:08 How to test checkpoints to find best one
1:32:15 Why you should use After Detailer (adetailer) and how to use it properly
1:34:48 How to do proper highres fix upscale
1:36:19 Why anatomy inaccuracy happens
1:37:07 How to generate images forever in A1111
1:38:02 Where the generated images are saved and download them
1:40:30 Super Important
1:45:16 Analyzing x/y/z checkpoint comparison results to find best checkpoint
1:48:20 How to understand if model is overtrained
1:52:27 How to generate different expressions having photos
1:54:53 How to do inpainting in Stable Diffusion A1111
1:56:34 How to generate LoRA from your trained checkpoint
1:58:03 Windows OneTrainer training completed so how to use them on your computer
2:00:24 Best SD 1.5 models Fine-Tuning / DreamBooth training configuration / hyper-parameters
2:03:50 How can you know you have sufficient VRAM?
2:05:36 What to do before terminating MC VM
2:06:55 How to terminate your VM to not spend anymore money
2:08:35 How to do style, object, etc training
2:09:47 What to do if your thin client don't synch

[alexmazurek]

Your contribution to the broader AI community cannot be understated. Thank you so much for your education efforts!

[mehmetalirende]

Mr. Furkan's dedication to open-source AI education deserves the highest recognition. His contributions are truly valuable, and his Patreon page is a prove of his humbleness and devotion to education in-how he welcomes anyone for a cost of a cup of coffee, meanwhile he keeps updating scripts, as well as maintaining access to valuable datasets. 👏👏👏👏👏👏

[GKChandlerBooks]

Just signed up to your Patreon and it's worth every penny. I thought I was doing pretty well after watching your video, but now everything has changed. I've gone from 330MB LoRAs to 88MB. Quality is excellent. It takes longer, but it's worth it. Still fine tuning but this has given me an enormous leg up on what I was doing previously with OneTrainer. Thanks, Doc!

[akairas]

Hello Furkhan!, I must say that I was a little tired of seeing your posts everywhere. BUT I also have to say that this tutorial is awesome and you clearly put a lot of effort into it. I just followed your tutorial and the results were AMAZING. I appreciate the work you've done on this and keep it up!

[Siliciomusic]

I just wanted to please you since I know everything I have learned from your tutorials, I have learned Python, I work at runpod and I have been following you since you started this adventure with AI. I don't have the money to pay for your courses on Patreon but despite that you have helped me a lot. thank you

[xyfhkod]

Dear tutor Furkan (Computer Engineer Phd.) has an effort and a vision that far beyond his time. Mentoring for AI tools, libraries and easy to understand workflows anyone who interests AI-Compuer Vision field. I highly recommend his patreon just because of always updating scripts, reading every messages to reply and his own helper tools and various pixel-size training reference image data-sets. 👍👋👋👋

[graphicstooge]

I want to dedicate a little of my limited free time to do a well-deserved review of Furkan's latest tutorial on Onetraining on Massed Compute: it's a truly wonderful thing! And I'm not just talking about the technical aspect, but above all the narrative plot with which it was created... The parts that fascinated me the most are precisely those little stumbles that he encountered (and that we all encounter in our amateur approach to generation of open source artificial intelligence), like the simple copy and paste of a simple script... it was really fun, and I really appreciate that those parts were not edited by Furkan... For the rest, high quality content quality as always, well explained by a person who has an evident strong passion in what he does... I'm definitely happy to be part of this small community, and I definitely want to upgrade my level of support to Furkan in the way I deem most appropriate in the shortest possible time! Keep on Furkan!

[ForTheEraOfLove]

OneTrainer was the only working local training software with my Pop OS 22.04. I tried Kohyas and Lora Easy Training Scripts. I appreciate you for co-creating a magnificent future 🙏

[YalnizKurtMK]

Discord'dan geliyorum. Video şaheser olmuş hocam, sonuna kadar sardırmadan izledim. İzlerken bir ara sinyaller bende koptu ama genel bir özetle aslında çok pratik bir anlatım yapmışsınız. Videoda gösterdiğiniz veri seti ile çıkan görüntüler neredeyse muazzam. Sanırım ilk denememde sizin gibi bir veri seti ayarlayıp deneyeceğim. Bahsettiğiniz şekilde; farklı ortam, farklı kıyafet vb. göze uğraştırıcı geliyor. Öğrenci olmasam çok daha fazla katkıda bulunmak isterdim. Konunun training ve stable diffusion olmasına rağmen, her karesini böyle detaylı anlatan sizden başka İngilizce kaynak görmedim. Elinize sağlık.

[reapicus]

Thank you for making this tutorial! I was sort of lost in the dark, so to speak, going from tutorial to tutorial. Your videos and resources have been excellent, and you have helped me tremendously. :D

[UHDking]

Very smart guy. Thanks. You got my like I appreciate step-by-step.

[CendioThinLinc]

Thanks for sharing your knowledge about ThinLinc with your subscribers. We've already added your video in our Community Videos playlist!

[KnowtHurts]

Just finished the tutorial and HOLY CRAP MAN! THE RESULTS ARE AMAZING!!! Thank you, man! You ROCK!

[ymgxbjo]

Great Video, Learned a lot of things, hope can implement these in my project. Thanks!

[Vyviel]

Thanks for posting such a detailed tutorial it's going to take me awhile to watch it all but I am steadily working through it in my free time

[hundredfireify]

Great work. I've been following your work since the very early days of dreambooth training. Some of the most precise tutorials out there.
May I suggest that you nailed the realism training. I think it would be great if you could experiment with stylized training (training optimized to turn yourself into an pop/manga/anime etc).
It's how many people use this technology for nowadays, given that realistic models are not very flexible.
Cheers !

[princesonnenberg]

Thank you very much sir for open sourcing your material in this manner. There are no words to describe this level of generosity in an age of each one for himself and god for the rest of us.... I wish you continued success and happiness.

[PierrickBignet]

Thank you for this excellent tutorial. It takes so much time and effort to put something like this together, you're awesome.

[glassmarble]

ok. i was angry cause its 2 hours video but testing 100 different setting for this video big time consumption.
yes i forgive you :) thx

[ihsasss]

I have watched it about 1 hour watching the reat tomorrow