So, revisiting the question. Why does modern photogrammetry take so many computational resources?
The reason why is because our present-day computer systems are designed for mathematical and visual perfection, whereas our brain is literally nothing but a bunch of cheap hacks on vision processing. And that’s why present-day computer vision solutions are so computationally intensive. It’s only what’s technically required to get that kind of degree of perfection. Human vision systems, by contrast, get by with far less perfection. “Perfect” images in our mind are progressively constructed by repeated exposure to certain objects and careful visual scanning of the object in question. Over time, our brain builds a more detailed image of the object in question. Well, some people can perform this process faster than others. Often times, in common language, we say these people have “photographic memories,” but in truth, they don’t, strictly speaking, because no human alive can actually see the individual “pixels” from their retina. The human brain only “sees” the post-processed information that their retina passes upward, which is a description of lines and patterns, not pixels. That is, not unless the pixels are large enough can the conscious brain actually see and measure in pixels passed upward from the retina.
The fallacy of photographic memory, that’s simply just not the way the human vision system operates.
Every single pixel from that camera sensor goes straight to the CPU. It’s like a dream that humans only wish they could live, but no matter how hard you try, your conscious brain cannot accumulate the pixel data from multiple dark image exposures to get a bright image. Why not? Because your brain doesn’t see pixels, of course!
Your brain never actually gets to see the data originally captured by your eyes.
The very expensive equipment used by astrophotographers. Of course it exceeds the capabilities of human vision.
Your brain’s memory circuits are very much the same way too. They operate as fixed function structures at the subconscious level. “Subconscious” is perhaps the word we use for all such fixed-function structures in the human brain that operate by simple, non-reprogrammable algorithms. At least not in our current understanding of computers where you can pass a program directly to a functional unit and have that functional unit execute that program.
Though that would be pretty amazing if our brains could be fully reprogrammable like computers. This might be good to include in Tour de Force of the Operation of Computers.