The AI-Infographic Wars are heating up. ChatGPT’s new image model is squaring back up to Gemini / Nano Banana Pro’s complete image generation dominance of the last month.
Both models are now incredibly capable, with outstanding text rendering and are able to leverage their thinking modes behind the scenes to guide the direction of the image model.
With no prior context, I asked ChatGPT to go have a look at my node-pixel project in github and just make an infographic that explains how it works to someone who hasn’t played much with hardware and node before. The results are below.
As a one-shot this is pretty good, and is more or less how I describe how it works when I give workshops on it or to people new to it. This probably isn’t surprising because the documentation outlines these concepts to orient new users to the library so it makes sense that the LLM would review these and be steered by them to generate the image content.
The lack of spelling mistakes or blurry text means you could pretty much take this, throw it into a slide for a presentation and talk to it without too much challenge.
The last few months have really demonstrated that the combination of very capable image models paired with high-calibre reasoning models is powerful - and extremely useful.
If you’ve got access to a paid ChatGPT account, it’s definitely worth chucking process docs or particular areas of a code base at it and get it to provide an explanation. I can see this being great for learning assistance for people new to a project, process or codebase.