Overview

Learn to generate professional studio-quality portrait images using Kohya LoRA Stable Diffusion training in this comprehensive 45-minute tutorial. Master the process of downloading and installing Kohya's GUI, setting up DreamBooth LoRA training parameters, and utilizing pre-downloaded models. Discover how to optimize training datasets, set VAE files for improved image quality, and generate classification images. Explore advanced techniques such as caption preparation, regularization image selection, and optimal LoRA training settings for GPUs with limited VRAM. Gain insights into saving and loading training states, calculating epoch step counts, and interpreting command line parameters. Apply your newly acquired skills to use generated LoRA safetensors files in Stable Diffusion web UI, perform checkpoint comparisons, and enhance face quality with high-res fixes. Experiment with various training parameters and learn to leverage different types of checkpoints in Automatic1111 Web UI. Conclude by mastering GPU selection for Kohya scripts and installing compatible software versions for optimal performance.

Syllabus

Introduction to Kohya LoRA Training and Studio Quality Realistic AI Photo Generation
How to download and install Kohya’s GUI to do Stable Diffusion training
How to install newer cuDNN dll files to increase training speed
How to upgrade to the latest version previously installed Kohya GUI
How to start Kohya GUI via cmd
How to set DreamBooth LoRA training parameters correctly
How to use previously downloaded models to do Kohya LoRA training
How to download Realistic Vision V2 model
How to do training with Stable Diffusion 2.1 512px and 768px versions
Instance / activation and class prompt settings
What kind of training dataset you should use
Explanation of number of repeats in Kohya DreamBooth LoRA training
How to set best VAE file for better image generation quality
How to generate classification / regularization images via Automatic1111 Web UI
How to prepare captions to images and when you do need image captions
What kind of regularization images I have used
How to set training folders
Best LoRA Training settings for minimum amount of VRAM having GPUs
How to save state of training and continue later
How to save and load Kohya Training settings
How to calculate 1 epoch step count when considering repeating count
How to decide how many epochs when repeating count considered
Explanation of command line parameters displayed during training
Caption extension changing
After when we will get a checkpoint and checkpoints will be saved where
How to use generated LoRA safetensors files in SD Automatic1111 Web UI
How to activate LoRA in Stable Diffusion web UI
How to do x/y/z checkpoint comparison of LoRA checkpoints to find best model
How to improve face quality of generated images with high res fix
18 Different training parameters experiments I have made and their results comparison
How to test 18 different LoRA checkpoints with x/y/z plot
How to properly set number of epochs and save checkpoints when reducing repeating count
How to use checkpoints of Kohya DyLora, LoCon, LyCORIS/LoCon, LoHa in Automatic1111 Web UI
How to install Torch 1.13 instead of 1.12 and newer xFormers compatible with this version
How to make Kohya scripts to use your second GPU instead of your primary GPU

Taught by

Software Engineering Courses - SE Courses

Reviews

Start your review of Generate Studio Quality Realistic Photos With Kohya LoRA Stable Diffusion Training - Full Tutorial

Taught by

How to Do SDXL LoRA Training on RunPod with Kohya SS GUI Trainer and Use LoRAs with Automatic1111 UI

Become a Master of SDXL Training with Kohya SS LoRAs - Combine Power of Automatic and SDXL LoRAs

SDXL Training With Kohya LoRA - Stable Diffusion XL Training Will Replace Older Models

How to Do SDXL Training with Kohya LoRA - Kaggle - No GPU Required

How to Do Stable Diffusion LORA Training by Using Web UI on Different Models

Doubling Stable Diffusion Inference Speed with RTX Acceleration and TensorRT - A Comprehensive Guide

Never Stop Learning.