Full tutorial: https://www.youtube.com/watch?v=ezD6QO14kRc
Z-Image Turbo LoRA training with Ostris AI Toolkit + Z-Image Turbo Fun Controlnet Union + 1-click to download and install the very best Z-Image Turbo presets. In this tutorial, I will explain how to setup Z-Image Turbo model properly in your local PC with SwarmUI and download models and use them with highest quality via ready presets. Moreover, I will show to install Z-Image Turbo Fun Controlnet Union to generate amazing quality images with ControlNet preprocessors. Furthermore, I will show how to 1-click install AI Toolkit from Ostris and train Z-Image Turbo model LoRAs with highest quality configs made for every GPU like 8 GB GPUs, 12 GB GPUs, 24 GB GPUs and so on. I did a massive research to prepare these Z-Image Turbo model training configurations.
Links & Resources Mentioned:
Download SwarmUI & Models: [ [https://www.patreon.com/posts/Download-SwarmUI-Models-114517862](https://www.patreon.com/posts/Download-SwarmUI-Models-114517862) ]
Ostris AI Toolkit (SECourses Version): [ [https://www.patreon.com/posts/Ostris-AI-Toolkit-140089077](https://www.patreon.com/posts/Ostris-AI-Toolkit-140089077) ]
Ultimate Batch Image Processing App: [ [https://www.patreon.com/posts/Ultimate-Batch-Image-Processing-App-120352012](https://www.patreon.com/posts/Ultimate-Batch-Image-Processing-App-120352012) ]
SwarmUI with ComfyUI Backend Windows Tutorial: [ [https://youtu.be/c3gEoAyL2IE](https://youtu.be/c3gEoAyL2IE) ]
SwarmUI with ComfyUI Backend RunPod and Massed Compute Cloud Tutorial: [ [https://youtu.be/bBxgtVD3ek4](https://youtu.be/bBxgtVD3ek4) ]
⏱️ Video Chapters:
Master Z-Image Turbo & LoRA Training: The Ultimate Guide!
In this comprehensive tutorial, I show you how to generate ultra-realistic images in seconds using the lightweight Z-Image Turbo model. We cover everything from 1-click installation on SwarmUI (ComfyUI backend) to mastering ControlNet Union for precise image control.
But that’s not all! I also reveal how to train your own high-quality Z-Image Turbo LoRAs using the Ostris AI Toolkit. I have developed a custom training preset that significantly outperforms the default settings—you have to see the comparison to believe it! Whether you are on a local PC, RunPod, or MassedCompute, this guide has you covered.
What You Will Learn:
Z-Image Turbo Setup: How to run this fast, 6GB model (FP8 included) on almost any GPU.
ControlNet Mastery: Use Canny, Depth, and Lineart to control your generations perfectly.
LoRA Training: Step-by-step guide
- 0:00 Introduction to Z-Image Turbo Model
- 0:54 FP8 Scaled Version 5.7GB for Low VRAM
- 1:10 ControlNet Union with Z-Image Turbo
- 1:30 LoRA Training with Ostris AI Toolkit
- 2:00 Default vs Custom Training Preset Quality Comparison
- 3:00 RunPod Cloud Training Preview
- 3:40 MassedCompute Cloud Training Preview
- 4:16 Downloading Z-Image Models via SwarmUI
- 5:00 Z-Image Turbo Core Bundle & ControlNet Files
- 5:58 FP8 Scaled Model & Musubi Tuner Converter
- 7:13 Updating ComfyUI for Sage & Flash Attention
- 8:13 Updating SwarmUI & ControlNet Preprocessors
- 8:52 Updating & Importing Latest SwarmUI Presets
- 9:20 Generating with Quality 2 Fast Preset
- 10:48 Generating with Quality 1 Upscale Preset
- 11:35 Quality 1 vs Quality 2 Visual Comparison
- 12:13 Setting up ControlNet Input & Aspect Ratio
- 13:41 ControlNet Strength Settings & Canny Test
- 15:26 Using Depth Preprocessor with Z-Image
- 15:58 Coloring Lineart Drawings with ControlNet
- 16:58 Lineart Preprocessing Comparison
- 17:50 Ostris AI Toolkit Installation Prerequisites
- 19:12 Installing Ostris AI Toolkit on Windows
- 20:02 First Time UI Setup & Launching Interface
- 21:04 Loading Custom Training Configs
- 21:38 Creating a New Dataset Structure
- 22:24 Ultimate Batch Image Processing App Install
- 23:17 Dataset Prep Stage 1: Auto-Zooming with SAM2
- 26:08 Dataset Prep Stage 2: Resizing to Exact Resolution
- 28:12 How to Select Best Training Images
- 30:24 Importance of Emotions & Angles in Datasets
- 31:44 Z-Image Resolution & Aspect Ratio Rules
- 33:21 Configuring Training Parameters & Epochs
- 36:52 Resolution Impact on Training Speed
- 37:46 Starting the Training Job on Windows
- 38:39 Monitoring Training Progress & VRAM
- 39:43 Checkpoint Generation Settings
- 40:40 Resuming Training from Last Checkpoint
- 42:09 Training Speeds on RTX 5090 vs 4090 vs 3060
- 43:01 Training Quality: Default vs Custom Preset Comparison
- 44:21 Testing LoRAs with SwarmUI Grid Generator
- 46:04 Fixing ControlNet Error in Grid Generation
- 47:09 Comparing Generated LoRA Checkpoints
- 47:38 Using Trained LoRA with ControlNet Union
- 48:10 RunPod: GPU Selection & Template Setup
- 50:32 RunPod: Port 8675 Config & Initialization
- 51:36 RunPod: Uploading Installation Files
- 52:01 RunPod: One-Click Installation Command
- 54:07 RunPod: Starting AI Toolkit & Proxy Connection
- 54:38 RunPod: Uploading Dataset via Interface
- 55:32 RunPod: Starting the Training Job
- 56:24 RunPod: Speed & Cost Analysis
- 57:28 RunPod: Auto-Stop Command Setup
- 58:24 MassedCompute: GPU Selection & Coupon Code
- 1:00:16 MassedCompute: ThinLinc Client Setup
- 1:01:21 MassedCompute: Transferring Files to Shared Folder
- 1:02:55 MassedCompute: Installation Command
- 1:05:49 MassedCompute: Connecting via Public URL
- 1:06:54 MassedCompute: Starting Training Job
1:08:43 Downloading Checkpoints & Stopping Instance
