My Image Workflow

I have many different approaches to how to make an image so here I will talk about my most typical approach. In time this article will be expanded with more details.

Let me do an outline of my process before going into details:

1. Generate lots of images. Often using wildcards. Either with Midjourney or Automatic1111.
2. Upscale images to around 3200 by 3200 resolution using Topaz Gigapixel AI upscaler. Done in batch processing.
3. Pick a good image to bring into Automatic1111 for inpainting.
4. Alternate between inpainting in Automatic1111 and doing corrections, touchups and edits in Affinity Photo
5. Adjust contrast, saturation, hue and other color details in Affinity Photo

Step1: Image Generation

My initial images are made with a Pony XL derived model in Automatic1111, Midjourney or Flux.

I start by experimenting with a prompt to get good something kind of looking like what I want. For prompts I use at example prompts for that model and tweak those. Once I found good prompts I tend to store them as styles.