We asked AI to create a bunch of cycling-related images – here's what happened

A few years ago at CyclingTips we wrote an article addressing the (somewhat silly) question: could AI be about to steal our jobs as cycling writers? The answer, of course, was “no” (at least not yet), but it was still fun to dive into the intriguing world of neural networks and AI-generated text.

The concept behind the technology is simple: give the AI engine a text prompt, and it will spit out a long passage of text for you, built on your prompt and drawing from the myriad pieces of text it’s been trained on. The results are often fascinating and more than a little entertaining.

Today, we dive back into the world of neural networks with a look at AI-generated imagery. Take note, pro cycling photographers.

In the past couple years we’ve seen an explosion in the number of AI image-generation platforms available online. You might have heard of DALL-E 2, created by the Elon Musk-founded OpenAI lab, or its slimmed-down cousin Craiyon (formerly DALL-E mini). Midjourney has made a huge splash in recent months too, particularly in the world of fantasy art, but these are just some of the many options available.

These platforms all work much like the AI text generators we wrote about last time. Underneath the hood of these sites and apps are neural networks which have been trained on huge libraries of existing images and their descriptions to understand what certain objects look like and how they interact with one another. When you feed these AI routines a text prompt, they do their best to create an image that matches that prompt, based on the learning they’ve done.

Some of the results are truly stunning.

Having watched the growth in this space, it got us thinking: what’s possible in cycling with AI-generated images? What would these AI platforms create if we fed in a bunch of cycling-related prompts? And would any of them be suitable for use on CyclingTips, say?

And so we had a play around with a bunch of free tools: Enstil, Craiyon, Stable Diffusion, Midjourney, and DreamStudio, feeding in a bunch of different cycling-related keywords. The results, we’re sure you’ll agree, are fascinating, entertaining, and, in some cases, more than a little horrifying.

We began with a simple prompt: “a cyclist riding up a mountain”. Here’s what we got from DreamStudio:

There is a lot to take in in this effort from DreamStudio too:

Here are a couple attempts from Stable Diffusion:

And some from Enstil:

Next, we tried the word “peloton”.

Here’s how Enstil interpreted “peloton”:

Alright, say you’re a bike designer and you’re looking for futuristic ideas to help with your next model. Maybe you turn to AI for inspiration? Here’s what you’d get if you dropped “Specialized full suspension gravel bike” into DreamStudio, for example.

Speaking of lots of tubing …

Let’s have a look at what some AI makes of a “state of the art bicycle in a showroom”. Here’s Stable Diffusion first:

Looks like we don’t need to send James or Dave to Eurobike next year.

Enstil had a couple of ideas too:

The showroom DreamStudio envisioned was perhaps a little too crowded.

Here’s Midjourney with the prompt “a show room full of futuristic bicycles”:

Changing gears, what about some close-up portraits of cyclists?

Gravel riding is all the rage at the moment. Let’s see what the AI makes of it.

Here’s “A cyclist celebrating victory in a bike race”, first from Enstil:

Here’s Stable Diffusion’s effort with the same prompt:

Speaking of faces, Craiyon really seems to struggle in that regard.

Here’s a series from Stable Diffusion that turned out surprisingly well from the prompt “black and white, film still, pained face, cyclist, riding bicycle”. Anyone looking for cover art for their new indie rock album?

Alright, let’s rattle through a bunch more images generated from a bunch of different prompts.

Alright, so a lot of this is quite silly and unusable. But there are ways to get the AI to create images that’s a little more useful. Using more feature-rich and well-developed AI platforms is one way (DALL-E 2 seems to be the gold standard but access is limited at the moment). Or you can try asking the AI to be more artistic rather than photorealistic.

Speaking of artistic, it’s possible to generate some truly wonderful stuff with Midjourney. Check out these two terrific images generated from the prompt “sketch of a futuristic road bicycle by Leonardo da Vinci”:

Even better: how about some Pixar-style portraits of some riders?

These are just genuinely impressive images from Midjourney. Which brings us back to one of our original questions: is AI able to generate images good enough to use on a professional website?

For most of the images above, the answer’s a pretty decisive no (as entertaining as they are to look at). But in the case of the more artistic or abstract creations? There’s clear potential there.

Here at CyclingTips we’re often looking for non-racing photos to help illustrate articles about various topics. It can be hard to find a photo that talks about, say, road safety, or the benefits of cycling on mental health. AI could clearly help in this regard, particularly if a more abstract or stylistic image is appropriate. As you can see above, Midjourney already allows you to create terrific content just for this purpose.

There is, of course, an ethical quandary to be navigated here. Using AI to generate imagery like this is directly taking jobs away from those who do this sort of thing for a living. It’s no surprise there’s considerable controversy about the technology for this reason. That and the fact that these platforms effectively generate images from different images that others originally created – itself an ethical grey area.

We’re not going to solve those issues today. What we can say, though, is that race photographers don’t need to worry. Case in point: here’s Enstil’s interpretation of Remco Evenepoel winning at the Vuelta a España.

And here’s a sneak-peek at the final podium of the 2022 Vuelta:

When it comes to racing imagery at least, we might stick with the pros for now.