>In this tutorial, I’ll guide you through the process of upscaling images using advanced AI-driven methods. We’ll explore two primary upscaling approaches and their implications. First, we’ll delve into mathematical algorithms that predict missing pixels based on neighboring pixels, providing slight improvements in image clarity.
However, these methods lack awareness of object details. Next, we’ll explore the generative AI approach, which uses models like Stable Diffusion to fantasize about missing information and create stunning high-resolution images.

I’ll walk you through running these techniques in Google Colab Notebook with Control-net in Automatic1111, discussing pitfalls, optimal settings, and how to adapt these techniques to your specific images. Whether you’re looking to enhance image quality, preserve intricate details, or achieve incredible resolution, this guide has you covered.

0:00 Intro
0:28 Types of upscales
1:48 Google Colab Notebook
2:37 Possible misconceptions
3:38 Upscale during generation Hires.fix
4:31 img2img upscale
5:58 How safe you initial image
7:15 Tiling upscale
8:40 Upscalers comparison
9:58 Combination of upscalers
10:51 The best way to combine upscalers
12:24 Adobe Generative fill

The link for my FREE Google Colab notebook:



Support the Hairy Eyeball

Share this on