View Single Post
Old 29th January 2019
Bikie John's Avatar
Bikie John Bikie John is offline
Full member
Join Date: Nov 2008
Location: Wessex
Posts: 4,009
Thanks: 187
Thanked 659 Times in 581 Posts
Likes: 518
Liked 740 Times in 494 Posts
Re: Camera versus The Human Eye

Thanks for the link, Dave, it is an interesting topic.

I think that (unless I missed it of course) the author overlooked something very important, which is our understanding of perspective. I have been thinking about this a lot, prompted partly by Paul's splendid photos of the Flatiron building and the difficulty of getting realistic correction of converging verticals. The article talks about scanning what we are looking at in order to use the very sensitive parts of the retina, but doesn't talk about our comprehension of the wider scene.

I think that we do something similar on the wider scale. When faced with something big like a landscape or a building, we sweep it with our eyes and integrate the result into what we think we are seeing. If we try this with a camera lens we will get all sorts of odd results - this is why unless we are really careful with ultra-wide lenses the results can look plain wrong, and why multi-image stitching is a lot more complex than we think.

Maybe a thought experiment will help put into words what I am thinking. Imagine standing in a flat featureless plain, with a wall of constant height in front of us. If we look directly at the wall, the top looks horizontal and level. If we look at it from an angle, the top and the base will converge towards a vanishing point. So far so simple, and this is what a camera will see. But what we actually do is swivel our eyes around to scan the scene, and our brains make a composite picture. When we swivel out eyes we are changing the angle at which we view the wall, so most of the time we will see converging lines in one direction or the other. Somehow our brain combines all these different views without any problem - but when we try to do it with a camera we have to make some conscious decisions about how to merge the images, and still (usually) none of the options look quite right.

David Hockney thought about this and among other things produced some landscapes based on lots of photos knitted together. At the time I didn't really understand what he was on about but now I think he has a point!

Reply With Quote