After my PhD, I really wanted to do something more creative, and I started creating educational videos about computer science. I felt there was a huge need for engaging, short educational videos to counter all the let's say, "less brain-activating” short videos on social media. I came across Motion Canvas and its use of generator to define animations really resonated with the way I like to code. However, I quickly realized that making high quality videos takes forever! And, unfortunately, existing AI tools cannot generate technical videos that well, they can’t even handle simple text! But then I thought, wait a minute, creating videos with Motion Canvas? This is just code? I turned around and looked at my AI buddy Claude and I told him “you can do it”!
I spent the following months working on that vision of enabling LLMs to generate 2D videos that don't break after just 10 seconds. The result is now called VideoZero and you can try it for free at videozero.ai (paid from my personal credit card). For one, here is the result of visualizing Conway's Game of Life: https://videozero.ai/view/?id=040c6732-38f3-4990-baff-73d03d...
Sometimes the videos are quite good, on other occasions, let's say they need a bit of refinement.
Would love to hear what you think and what you would like to visualize.
Cheers and have a good weekend! Patrick