Basic text to video already exists. Considering how fast text to image has grown, it's not unreasonable to have an optimistic timeline. As for perfect text to image, i really don't see it taking more than a year or two from now given the current rate of advancement (compare something like midjourney v1 to V4, which is around 8-9 months of progress). Obviously they could hit a major snag somewhere but i would not bet on it.