AI image generators are becoming more and more realistic, but they have their limits. Recently, the internet was sent into a frenzy when a highly realistic image of Pope Francis wearing a puffer coat was created using a text-to-image program called Midjourney. With AI art generators becoming more capable of realistic depictions, there is a greater possibility of the spread of misinformation. However, when it comes to unicycles, these same AI programs are coming up short.
Luke Bailey, head of digital at The i Paper, recently wrote a Twitter thread addressing the difficulty AI image generators face when trying to depict unicycles. Prompts tested with the text-to-image generator Stable Diffusion 2 yielded sub-par results, with the technology having issues both with depicting the one-wheeled vehicle itself and how humans interact with it. In some cases, the unicycle seemed to be replaced with a bicycle.
The trouble with AI programs depends on the type of source material used in training them. Human hands, for example, posed a challenge for AI due to the focus of the source images which largely being on faces instead. However, Creative.ai’s Alex Champandard noted to The New Yorker that if source imagery was more focused on hands, then generative models would be better at portraying human extremities. This idea has recently been implemented with Midjourney, which underwent an update that has seemingly allowed AI to better depict human hands, according to The Washington Post.
As more money is poured into the industry, AI is likely to make some significant strides in the near future. Unicycles too may soon become much easier to depict. In the meantime, it is important to be wary of AI-rendered images as they could be misleading or lacking in accuracy.
Midjourney is an AI-centered technology company focused on leveraging artificial intelligence to explore the creative possibilities of visuals. Founded in 2017, the company has become a leader in image synthesis and text to image process. With the immense power of AI, they have changed the way people create and share visual content, giving them the ability to create realistic images quickly and efficiently.
Alex Champandard is a co-founder of Creative.ai, an AI firm that specializes in creative solutions powered by AI. He is a recognized expert in the field of AI, specializing in creating algorithms that allow developers to create beautiful 3D images and animations quickly and efficiently. He has been featured in The New Yorker, Wired, BBC, and other leading publications for his work in the AI space.
