Generative

I needed a profile picture for this blog.

In the past, I would have just found some cool image off a site like tumblr. (Note: how I would have casually disregarded any concerns around copyright of the image).

But in this moment of ascension for generative AI, I took the opportunity to explore and make use of tools such as Midjourney, DALL-E, Blue Willow and Leonardo.AI.

Of the systems I was able to test, Midjourney was the easiest to use and produced the best results.

Here was my process for generating the current profile picture of this blog.

First, I used various prompts to generate a series of images. Prompts were along the lines of "black hole stars logo". This produced initial results such as:

Refinement of the prompt yielded different sets of images:

Then I was simply able to select some favourite images, generate variations and finally upscaled versions of them.

This created results such as:

I selected one of the images, and voila there was my profile image generated and set.

The whole process took about 20-30 minutes. The ability to create high quality art work with such little effort, and no exertion of artistic skill is amazing.

I am really excited by the potential of generative AI.