AI Video Generation: A Year of Drastic Improvements

What do you think we’ll see in 2025?

Image: author x Midjourney

In May 2023, I created an AI-generated video as a quick test run of some of the models that were around at the time (Zeroscope’s precursors and open source audio generation, if I remember correctly).

Fast forward to July 2024, just over a year later, I re-created my 2023 video with Midjourney (text-to-image), Luma Labs’ Dream Machine (image-to-video) and Suno AI (text-to-music).

The advancements are incredible but not surprising. Everybody in the field saw this coming.

The thing that’s kinda mind-boggling is the fact that this took place in such a short time span. And: The pace of innovation has not just maintained its momentum; it has significantly accelerated.