A toolbox with cogs, AI generated with Stability Matrix

Introduction

As you may have guessed now, if you are reading this blog regularly, I am not an artist. I am an engineer, developer, PhD in Computer Science, gamer, and many other things. But I am definitely not an artist (not yet at least). I released this year, in February, my first Google Play app, called « My Device Info ». But the logos and « art » were quickly made with MS Paint to have the app published on Google Play. Now, in order to appear more professional, or at least to have a proper visual art, I had two options. First, I could learn how to do art, how to use software like Inkscape or Gimp, how to draw… Second, I could try using AI image generation for a quick way to get logos for my app. And for the next apps coming (my Unity game for instance).

Free online tools for AI image generation

So, where to start? As I knew nothing about AI image generation, I googled « ai image generation free », and stumbled across Canva’s Dream Lab. I had 20 free prompts to try to do something. My goal here was to test the tool, and not really to generate a final logo.

I mostly used the same prompt, with some variations in the wording, or the words order. The results were mixed: some interesting ideas, but most of them complete crap.

I had exhausted all the free tries, so I needed a new way to generate logos and if possible, freely. Then I thought about running an AI image generator directly on my PC. From conversations with colleagues, I knew it required some powerful PC.

Luckily, I have a gaming rig:

  • CPU: Ryzen 7 3700X
  • GPU: RTX 3080 12 GB
  • RAM: 32 GB

So I should be fine.

Stable Diffusion and Stability Matrix

My new Google search was: « ai image generation locally windows », and I first read about Stable Diffusion. First, I did a few tests with NMKD SD GUI and got horrible results – see below. Then, I ran across Automatic 1111 Web UI and Stability Matrix. Basically, Stability Matrix allows to install several AI image generation tools with a GUI. It also handles all the process of installing dependencies with just a few clicks. So this solution is perfect for the beginner I am.

I also found several resources to learn how to use these tools:

So first I trained by replicating the examples from the tutorials (and succeeding to some extent). Then I began working on the logo for my app « My Device Info« . The prupose of the app is to lists technical info from a mobile phone (sensors, SDF, etc). So the main idea for the logo was a toolbox, a cog, and to do something with them.

Gallery of AI generated images

See below a gallery of fails and successes while trying prompts. You can open the images in a new tab to see them at full scale.

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *