| The problem is, in theory you can capture any color accurately (if you usa a sensor that has exactly the same spectral response of its channels like the eye), but you can never display all colors accurately, it is mathemathically impossible without an ability to really accurately form any arbitrary spectrum. In real life common standadrs rely on 3 channels (R, G, B; or their recalculated equivalents like YUV or so), so not enough information to form all colors correctly. So the system needs some compromise. It is optimized so at least most natural and common colors are displayed correctly (skin tones, nature,...). The drawback is for certain spectrums the color distortion becomes quite significant, an example being the typical mercury radiation spectrum. So you can in theory tune the system to display the mercury radiation correctly, but then it will suck for other colors, like e.g. skin color tones (notoriously hard to display, as people are especially sensitive on any error there, yet normally of way higher priority for correct display than back alley mercury lamp).
Plus if you take into account what was the historical driving force for picture capturing (the movie industry), the scary looking greenish hue quite matches the mood of the scenes where this lighting uses to prevail, so such color distortion becomes more a feature than a bug...
|
|
|
|
Logged
|