With AI starting to get more coverage, some of which is a little skewed, I thought I would take a minute to give a brief overview of my typical workflow, for the layman who may not know what exactly is going on behind the scenes. Take this image as an example 1/?

The hardest part to explain is prompt engineering. I primarily use DiscoDiffusion because of its versatility. Unlike some basic text-to-image programs like MidJourney or DALLE there is a lot more to it than just the prompt text. 2/?

Here are a couple screen shots showing just a glimpse of a few of the settings you can use to manipulate outcomes. This gives you an idea of the complexity of the engineering. 3/?

All those aside, the engineering of just the prompt itself, meaning the text you use to tell the AI what you are looking for, is an art and took me about five months to start to feel like I had any mastery over it. 4/?

If I opened DD and wrote "redhead with glasses with neon colors" I would get an output like this. Not exactly the same is it? So yes, you need to do a lot of research and testing into what words in what order with what weight will bring you desired results. + the settings 5/?

After I have my prompt and settings set the way I want. I run it for a few hours and it makes a bunch of images. From these I select which I want to work with. Much like a photographer might take 300 shots and choose the best to bring into photoshop. In this case I chose 6/?

Now that is already a pretty great image, and no doubt DD has done much of the heavy lifting. But I will then take it into PS and do some manipulation and painting to create a more cohesive face. Before and After 7/?

At this point, I usually cut the image into sections and rerun it through DD for more detail, then stitch the sections back together in PS. However, I was feeling a little silly and I decided to run it through FaceApp. I gave her glasses and a dewy glow. 😂 8/?

However, faceapp outputs at very low res. AND I wasn't thrilled with the way the face's style was different than the rest. So I used the image as an initial image in DD to create a cohesive style. 9/?

I then cleaned and sharpened it all up, added a little detail and viola, the final outcome. For me, my process varies a little from image to image, I've never used faceapp b4 for example. and this is just MY process. There is a WIDE range of processes AI artists use. 10/?

I hope this enlightens collectors a little to show there is more than just typing something in and getting a magical image. And I hope it might help artists who are curious about AI and wondering how they could maybe use it. Thanks for the listen! 11/11