A year ago I started my own consulting company and I had documented my journey in leveraging AI in the creation of my company - from the name to the logo. What had stood out then was how much the LLMs struggled with generating images with text on them.
Here are six examples and attempts at getting the LLM to generate a logo for my company with my company name.
I finally managed to get it to generate a logo but getting it to accurately add the tagline - Deep Understanding -> Tangible Outcomes was impossible.
So I settled for this logo - btw, this is pretty awesome as it captured what my wife (a writer) and I (a GenAI consultant) both do. It has the book within a gear which I thought was pretty cool.
Yesterday, OpenAI announced 4o Image Generation and I was very excited to see if it would be better at -
Adding my tagline (with the correct spelling)
Add the tagline on the top going around the gear
Add some color to the pages of the book
Most importantly, could I give it my original logo and could it keep all the main elements and just make the tweaks I needed. This had been impossible before.
So I gave it these instructions and this is what it generated
It is very impressive how far we have come and it is another reminder that if AI can't do something today, it is just a matter of time.
I am very excited to see all the new use cases this now opens up for us.