Creating AI images using text prompts is great, but what if you don’t quite know how to describe the image you want to create? That’s where Scribble Diffusion comes in.
As the name suggests, Scribble Diffusion is a web app that lets you conjure up AI art by sketching what you want to see. And even if you’re not a natural Van Gogh, there’s still a prompt box so you can tell the AI what your scrawl is supposed to be.
When you launch the app, you’re presented with a square canvas that’s populated by a rough demo drawing (see, I told you you don’t have to be good at art,) accompanied by a prompt box that has some example text. These instructive items are all you need to show you how to get started.
Pressing Clear under the drawing box allows you to get started with your own blank canvas. You can draw with a mouse, a stylus, or even your finger. In my case, I used my iPad and Apple Pencil, but again, it’s just about creating a shape for the AI to work with.
I drew a crude hen, and then told the AI I wanted an image of a chicken in a farmyard. I set the AI to work, and a few moments later I had a somewhat stylized image of a feathery, brown and white bird, standing in a wooden barn strewn with straw.
What makes Scribble Diffusion great is perhaps also what holds it back; it’s easy to understand and simple to use, but I couldn’t help feeling like I wanted to do more with the tool. For example, beyond the prompt box, there is no customization. You simply have to hope that the AI comes back with something usable. And while the results are broadly quite good, there were a few occasions where I hankered for the option to tweak the settings.
Scribble Diffusion is an open-source project offered by Replicate, and so while it can feel like somewhat of a basic parlor trick at times, that shouldn’t detract from what it’s actually offering. After all, it doesn’t cost anything to use, and the results are watermark-free, so at the very least, it’s worth checking out.