Don’t Give Up, Ribbon Variations.

by See-ming Lee
Using letterforms to drive compositions with Stable Diffusion.
Don’t Give Up, Ribbon Variations.

I have previously explained that the brand colors for the SML identity was intended to show the full spectrum of colors — that, like the rainbow in nature, it’s intended to express the diversity of life.

Recently, a friend reminded me how the phrase “Don’t Give Up” needs to be “periodically injected”, so I thought why not make some elaborations on this simple typographic form that I made so many years ago, and which I have iterated on for so many years.

So here I present — Don’t Give Up, the Ribbon Variations, made with Stable Diffusion.

Tech

I have been using Stable Diffusion to do some experiments with design. There are many AI image rendering engines out there, but only Stable Diffusion gives me the best range of possibilities. The learning curve for Stable Diffusion is steep, but if you’re willing to do the work to figure it out, then it will allow an artist or a designer the maximum control that you seek to realize a creative vision.

At the moment, there are AI engines like Midjourney that most creative professionals use. It is a very powerful engine, but it lacks controllability — there are some things that simply can’t be easily done — such as the designs I have here — where I supplied the typographic forms that I drew inside Adobe Illustrator to control the flow and composition of the ribbons.

I want to show these experiments because I have tried very hard to convince artists who work with Midjourney to try out Stable Diffusion in order to give them the freedom to create anything they want. Yet many of them give up as soon as they played with Stable Diffusion briefly, usually discouraged by its “ugly UI”.

The beauty of Stable Diffusion is that it’s not only available in one user interface. It’s open source. If you don’t like the UI, write a better version! That’s the beauty of open source. It‘s something that I have been working on — to make better interfaces for it so maybe more designers would be willing to give it a go.

Until then, here are some sketches that I have been working on.

If it’s not so very clear about what’s going on. Here’s what I did:

  1. In Adobe Illustrator, I turned my typographic design into black and white.
  2. I exported that vector drawing into a PNG.
  3. I fed the PNG into the Canny control net.
  4. I described the ribbons in txt2img, and the control net drove the compositions by following the letterforms of Bickham Script Pro.

Images

Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.

Source

This is the original design that I make, which I put on top of the generations from Stable Diffusion.

Additional Images

Images below are from the same session. They use the same workflow, but are not visually as interesting. I’m including them here so you can see the range of results if I didn’t cherry pick the best ones.

Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.
Don’t Give Up, Ribbon Variations.