Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: allegretto on September 19, 2014, 11:45:01 am
-
Could a manufacturer via software render all pixels monochromatic and then use them all?
Seems easy enough to someone like me who knows nothing of the EE work it would entail, so maybe not
But if it could it would be a strong "feature" for still or video
-
Given the potential advantages in low light performance I must admit I'm quite surprised nobody but Leica and Hassleblad have offered monochrome sensors.
Olympus or Fuji especially seem well placed to go this route, small cameras that appeal to street shooters and offering a monochrome body could combat the disadvantage in noise performance of their smaller formats.
-
Could a manufacturer via software render all pixels monochromatic and then use them all?
Seems easy enough to someone like me who knows nothing of the EE work it would entail, so maybe not
But if it could it would be a strong "feature" for still or video
Maybe you have not heard but Leica has a B&W-only camera:
http://us.leica-camera.com/Photography/Leica-M/Leica-M-Monochrom
-
yes yes
but for me an investment in an M240 and an MM is a bit over the top
not talking about a different body. Talking about re-routing the signal thru the right processor. Could be firmware controlled
-
Given the potential advantages in low light performance I must admit I'm quite surprised nobody but Leica and Hassleblad have offered monochrome sensors.
Don't forget Phase One:
http://www.luminous-landscape.com/reviews/cameras/achromatic.shtml
Jim
-
not talking about a different body. Talking about re-routing the signal thru the right processor. Could be firmware controlled
you can do this during raw conversion/post processing... does not make sense for a manufacturer to spend money for a very niche market unless they can charge a premium like Leica for example
-
you can do this during raw conversion/post processing... does not make sense for a manufacturer to spend money for a very niche market unless they can charge a premium like Leica for example
not the same as recruiting all pixels for IQ
-
yes yes
but for me an investment in an M240 and an MM is a bit over the top
not talking about a different body. Talking about re-routing the signal thru the right processor. Could be firmware controlled
It's not the same. You have color filters on regular sensors. Those are removed in the Leica sensor, and the sensitivity is thus much higher.
-
It's not the same. You have color filters on regular sensors. Those are removed in the Leica sensor, and the sensitivity is thus much higher.
it also uses twice the pixels...
-
it also uses twice the pixels...
Maybe 4x?
-
Maybe 4x?
correct me if I'm wrong but the image info is all green which is 50% of the pixels. No change in sensor size so it's a simple double… yes squared area, but same image size
-
correct me if I'm wrong but the image info is all green which is 50% of the pixels. No change in sensor size so it's a simple double… yes squared area, but same image size
Huh? It takes 4 cells to make 1 pixel in color (Bayer pattern), it takes 1 with the Leica system.
-
Given the potential advantages in low light performance I must admit I'm quite surprised nobody but Leica and Hassleblad have offered monochrome sensors.
Olympus or Fuji especially seem well placed to go this route, small cameras that appeal to street shooters and offering a monochrome body could combat the disadvantage in noise performance of their smaller formats.
I don't know about low light. I got 2 MM's. They are noisy as hell above 2000 to 2500. They are useless for me at much higher levels.
-
Huh? It takes 4 cells to make 1 pixel in color (Bayer pattern), it takes 1 with the Leica system.
yes, but we are talking about number of pixels available for IQ. In Bayer, only the green provide IQ and there are 2X green to every one red and one blue.
If all four are used for IQ that makes 4 not 2, no?
Seems like a double to me...
-
yes, but we are talking about number of pixels available for IQ. In Bayer, only the green provide IQ and there are 2X green to every one red and one blue.
If all four are used for IQ that makes 4 not 2, no?
Seems like a double to me...
Sorry, I don't understand.
-
Could a manufacturer via software render all pixels monochromatic and then use them all?
To do this you have to alter the effect of each R, G & B filter in the Bayer array. Current image processors attempt to do this algorithmically, generally with decent results. But a filter with variable bandwidth and variable peak wavelength would do the job properly. First, though, someone will have to invent it. :)
-Dave-
-
not the same as recruiting all pixels for IQ
you were saying "Talking about re-routing the signal thru the right processor" so you imply that CFA are present, just you want some processing in camera to invert what filters do with spectrum... why 'd you need to bog the camera, when you can do this off camera ?
-
Huh? It takes 4 cells to make 1 pixel in color (Bayer pattern)
No.
It takes the color luminance information from a photoreceptor (AKA pixel) plus that of the surrounding pixel to make a full set of color values at any single pixel's absolute location. With that full RGB value for every X/y coordinate you then do a color to monochrome conversion.
-
What I think Allegreto means is for the manufacturer to implement in firmware the capability for the bayer pattern pixels to go completely color blind. Then proceed to de-mosaic the capture in the monochrom way.
I undestand this would be different to only de-saturating the colors in camera to produce a b&w image or to do it by de-saturation later in post.
If this is true, the gain would be to have the extra resolution of monochrome sensors but without the gain in light gathering because the color filters. So no cleaner images, but better resolved.
Eduardo
-
Huh? It takes 4 cells to make 1 pixel in color (Bayer pattern)
No.
It takes the color luminance information from a photoreceptor (AKA pixel) plus that of the surrounding pixel to make a full set of color values at any single pixel's absolute location. With that full RGB value for every X/y coordinate you then do a color to monochrome conversion.
4 cells to make a complete pixel, at least.
-
What I think Allegreto means is for the manufacturer to implement in firmware the capability for the bayer pattern pixels to go completely color blind.
they can't go color blind unless there is a physical/chemical way to make them equally transparent on demand... everything else can be done off camera during raw conversion/postprocessing, as without making CFA filters truly equally transparent on demand you can only do some math with raw data postexposure (much better then to try to burden the on die circuitry with means to do that in analog domain /pre ADC/ - silicon real estate on sensor surface is a valuable property) ) ;) ... for some interested in OOC JPG -> in camera's firmware.
Then proceed to de-mosaic the capture in the monochrom way.
there is no demosaic there (for a CFA-less sensor - but if you do have CFA then any math will involve analysis of nearby sensels, because you can't properly guess what was the total light hitting the sensel based on just one number /readout from that one sensel/ and filter /on top of that sensel/ spectral transmission profile)