Pompous Moronox
Well-known member
I'll clip lots, as I am too lazy to write much.
re right, but more conservative difference, and the new D7000 and K-5 are able to do it at the base ISO. The new Sony sensor is truly amazing when it comes to read noise - not only it is very very low, but it is also highly symmetric (meaning no banding etc.).
Also you're now mixing diplaying devices and capturing devices - the discussion is not about the properties of paper and ink, or such, but about capturing devices. How the extra DR captured can be viewed is a different topic.
When it comes to human eye, we only see a very tiny area at any time, and even that at rather low quality. It is our brain which computes us high DR imagery - we don't see extreme DR at any given time AFAIK - if we did, our pupils would never change in size. The computer on our should does create a nice illusion of rock solid eyesight though
Also you not only underestimate the demosaicing algorithms in use today, but also forget that every currect Bayer-based DSLR sensor has lots more pixels than any Foveon-based sensor. If one decides, one can do the worst imaginable demosaicing by just combining 4 pixels into one pixel and reducing the pixel count by a factor of 4, halving the original resolution, turning a 16Mp bayer sensor into producing 4Mp images. If we do this, we can do exactly what you describe - move from one colour to another without any transition. This however is extremely suboptimal, literally the worst possible way to do the demosaicing, yet it would still about equal any current Foveon image resolutionwise and be superior in every other department! Doing a proper demosaicing gives vastly superior quality compared to this extreme case.
Of course, if the SD1 comes out, it will create very interesting resolution figures, and local contrast too. A nice tool for black and white photography, I am sure.
I was a bit careless in my wording, and of course you're right. What I meant was that when you store the image into JPG, the image is not the same linear space image the RAW file was, but has gamma curve adjustment and other adjustments, possible for example lifting up the shadows - including more of the dynamic range of the captured scene, than would be if the image were just a straight gamma corrected image. I really should think before I write, sorry about it.
Well, not quite - the S-curve is not to produce a natural image, but to increase contrast in the important area, typically midtones. But gamma correction is just a simple curve with one bend, not an S with two.
It depends on the difference in the light - a coal cellar and bright outdoors, you'
re right, but more conservative difference, and the new D7000 and K-5 are able to do it at the base ISO. The new Sony sensor is truly amazing when it comes to read noise - not only it is very very low, but it is also highly symmetric (meaning no banding etc.).
Also you're now mixing diplaying devices and capturing devices - the discussion is not about the properties of paper and ink, or such, but about capturing devices. How the extra DR captured can be viewed is a different topic.
When it comes to human eye, we only see a very tiny area at any time, and even that at rather low quality. It is our brain which computes us high DR imagery - we don't see extreme DR at any given time AFAIK - if we did, our pupils would never change in size. The computer on our should does create a nice illusion of rock solid eyesight though
11 stops is something the new D7000 and K-5 can handle with very good results (even though DXO rates then as 14 stops or so, for quality output less is realistic, but of course this depends on our quality requirements as well as output size). Also the old Fuji DSLR with the kludge-sensor would probably handle this situatioin all right.
Displaying the images is irrelevant - that is totally different topic and has little to do with cameras or sensors.
This again has nothing to do with DR
Also you not only underestimate the demosaicing algorithms in use today, but also forget that every currect Bayer-based DSLR sensor has lots more pixels than any Foveon-based sensor. If one decides, one can do the worst imaginable demosaicing by just combining 4 pixels into one pixel and reducing the pixel count by a factor of 4, halving the original resolution, turning a 16Mp bayer sensor into producing 4Mp images. If we do this, we can do exactly what you describe - move from one colour to another without any transition. This however is extremely suboptimal, literally the worst possible way to do the demosaicing, yet it would still about equal any current Foveon image resolutionwise and be superior in every other department! Doing a proper demosaicing gives vastly superior quality compared to this extreme case.
Of course, if the SD1 comes out, it will create very interesting resolution figures, and local contrast too. A nice tool for black and white photography, I am sure.
You're talking about contrast or local contrast, not about dynamic range
