[thxbb12]

[quote name='Brightcolours' timestamp='1307273202' post='9045']

So, while one can say the 450D uses 14 bits and the D90 12 bits, both translate into 24 bits RGB (8-bit formats like JPEG) or 48 bits RGB (16 bit formats like 16-bit TIFF). There may be cameras where the 14 bits show an increase in actual information (like maybe the 1D mk III/IV), but the differences will be small. Between a D90 and 450D, I doubt any differences between the cameras are really impacted by the theoretical advantage of 14 bits over 12 bits.

[/quote]



Until very recently (and excluding Fuji S5 Pro camera which uses a special sensor technology), it was a myth that 14-bit RAW files were bringing anything more than random bits of information (ie. noise) leading to larger files. It was a pure marketing ploy from manufacturers to make people believe they were gaining some more pixel depth information when it wasn't the case at all.

This was true until recently with the release of Sony's 16MP sensor as found in the Nikon D7000, Pentax K5 and Sony cameras. That sensor is able to use, at very low ISO only, the extra 2 bits of information.

Fore more technical details about this, please read Gordon B. Good post below, an extremely competent figure in sensor technology and electronics:



http://forums.dpreview.com/forums/read.asp?forum=1036&message=37014773



I wouldn't worry about 12 vs 14-bit, especially regarding the Canon as the extra 2 bits of "information" you get are nothing more than random noise.

[wim]

[quote name='thxbb12' timestamp='1314721223' post='11171']

Until very recently (and excluding Fuji S5 Pro camera which uses a special sensor technology), it was a myth that 14-bit RAW files were bringing anything more than random bits of information (ie. noise) leading to larger files. It was a pure marketing ploy from manufacturers to make people believe they were gaining some more pixel depth information when it wasn't the case at all.

This was true until recently with the release of Sony's 16MP sensor as found in the Nikon D7000, Pentax K5 and Sony cameras. That sensor is able to use, at very low ISO only, the extra 2 bits of information.

Fore more technical details about this, please read Gordon B. Good post below, an extremely competent figure in sensor technology and electronics:



[url="http://forums.dpreview.com/forums/read.asp?forum=1036&message=37014773"]http://forums.dprevi...essage=37014773[/url]



I wouldn't worry about 12 vs 14-bit, especially regarding the Canon as the extra 2 bits of "information" you get are nothing more than random noise.

[/quote]

Unfortunately, the only real difference is that Canon Raw is raw, whereas Nikon processes raw before saving it to remove shot noise. That also means you lose the last few bits of information, which is one of the reasons why most astronomers prefer Canon to Nikon when shooting stars, and why Nikon bodies score higher in DxOMark figures. Effectively you get almost no noise in a Nikon at the first two bits, at the cost of real image data. With the right tools and processing, you can actually still maintain a large part of the real image data, and still lose as much noise. However, you'd need a proper unprocessed raw in the first place.



As far as I am aware, based on measurements done by some sensor experts, the current low noise champ sensor before any processing whatsoever, measured straight from the sensor, is the Canon 1D MkIV.



Effectively, when processing correctly, with the right tools, even a 1D Mk III outperforms, noisewise, a D3x or D700.



When it comes to displaying a tonal gradation, optimally processed, 14 bits do make a difference, with any camera capable of using an additional 16% of tonal steps, even if DR is the same or less ....



Warm regards, Wim