Image/video generation models will be a commodity, so with time we'll have better and better models.

I believe the difference will lie in the interfaces we use to connect/request these models.

I've made this product coolphoto.ai public with AI photo packs created from selfies and curated by humans, but I'm working on a browser-based studio with lot of features to compete with midjourney, sora, etc.