Originally posted by Trickortreat Anyway - still hope real camera manufacturers will start implementing some kind of computational photography algorithms sooner than later
Originally posted by Trickortreat even the best sensor and best lens in the world are nowhere near what your eyes can see. with computational photography they just might get there or even beyond...
Each to his or her own, I guess...
I
can see the potential power of computational photography in various applications, and some uses for it... but it doesn't hold any attraction for me in terms of my photographic hobby. I'm not interested in having the camera emulate what my eye sees (or what I think it sees) through computational algorithms. I want to record how a lens captures the scene I've observed and composed. To me, there's so much character in each lens, and that adds something unique to every photo I take, warts and all. Of course, computational photography could potentially emulate the output from those lenses to some degree, but it would never be precisely the same as shooting with the original lens on a particular film or sensor format.
I used to play guitar a lot. Less so these days, but I still own a few nice instruments. The most unusual of those is a Line6 Variax Acoustic - actually a solid-bodied electric guitar (shaped like an acoustic) with an under-bridge pickup and electronic sound modelling that lets the player emulate the sounds of different acoustic guitar bodies and microphone / pickup positions. With simple adjustments, I can get a sound quite similar to a large-bodied Gibson J-200 or smaller 000-sized Martin, and anything in-between. It's clever, and has its uses for live work. But it's not nearly the same thing as playing a
real Gibson or Martin acoustic, mic'd up with proper instrumental amplification microphones. The guitar doesn't feel the same. The player's experience and connection with the guitar is completely different... easier, in many respects, but bland. And while the broad sound is generally convincing in a mix, all the little nuances of the real thing - nuances that make such a difference - are missing, or lost in generalisation. Similarly, when playing electric guitars, I've used rack-mounted and software-based signal processors that try to emulate the sound of a particular guitar, amplifier, speaker and environmental combination. They're great fun, and they can be very useful - but they don't perfectly reproduce those conditions by any means. Instead, they give a broad approximation that's missing all the wonderful nuances of the real thing.
My concern with computational photography is that, however clever it may be, it would be computing and generating significant aspects of my photographs rather than simply capturing, as faithfully as possible, the output of a real lens and film or sensor combination. For some folks, I'm sure that's fine, but it would take much of the romance and validity out of the photographic process for me. And I wouldn't feel like I was wholly responsible for my photographs