Which is part of the camera that is responsible for color depth of the images?
Is it the property of the image sensor or the image processor of camera? And why do it differ from cameras?

How much effect do small difference in color depth have on the resultant image?

  • Are you asking about "color depth", which I can only think of as bit depth...or are you really asking about "color fidelity", which would refer to the quality and richness of color produced by a camera? Bit depth is really a choice camera manufacturers make, and they build their ADC's and image processing pipeline to operate on a given bit depth.
    – jrista
    Sep 13 '12 at 17:46

Generally speaking, most DSLR cameras these days are 14-bit. Smaller form factors may have lower bit depths, such as 10 or 12. The bit depth of a camera ultimately refers to the number of distinct levels of luminance a camera is capable of producing. This is a two to the power of factor, which means a 14-bit camera is capable of producing 2^14 or 16384 distinct levels (of tone...irrespective of color). A 12-bit camera is capable of producing 2^12, or 4096 levels.

A difference in the number of tonal levels has a small impact on the number of colors produced, and in most cases one would probably have a tough time telling the difference in color between a 12 and 14 bit camera. An increase in the number of tonal levels can have a significant impact on the overall quality of tone, particularly quality of gradients, in a photograph. A 12-bit camera is going to run a higher risk of producing photos with posterization, an effect that usually shows up in large areas of smooth grades or mostly solid color, due to sharp transitions in luminance.

When it comes to color, this can be more affected by the construction of the sensor, particularly the strength and quality of the color filters over each pixel in a bayer sensor, or the ability of a layered sensor to distinctly discern the number of electrons registered for blue, green, and red readouts from each pixel. A camera with a strong CFA (color filter array) will generally produce richer, more accurate color than a camera with a weaker CFA. The color quality, or color fidelity, of a camera is often more dependent these days on the capabilities of its image processor, or the precision and accuracy of the tone curves (picture or image styles) applied during the processing of the sensor signal. Even with a weaker CFA (which tends to make a sensor a little more "color blind"), color quality can still be tuned and improved with a high quality, accurate image processing algorithm. This applies more to JPEG than RAW, and when it comes to RAW the color fidelity story is again a bit more dependent on the ability of the sensor to sense accurate color.

So, in general, small and even larger changes in bit depth will generally not affect the overall perception of color in a photo. Humans, according to most studies, can only really see discern a few million distinct "colors" (at a constant medium illumination...when factoring in luminance, humans can detect trillions of levels of gradation for a given color...such as red or blue...but we are still only sensitive to a few distinct colors). A 10-bit device can handle over a billion colors (both "chrominance" and "luminance"). A 12-bit device can handle 68 billion colors. A 14-bit device can handle 4 trillion colors. Most people would be hard pressed to tell the difference between an 8-bit and 10-bit computer screen. Even fewer people could tell the difference between a 10-bit and hypothetical 12-bit display, and those who could would likely note the improved quality of tonal gradations more so than any perceived increase in observable colors. The value of a camera that produces higher bit depth photos is in the quality of tone...the quality and fineness of small differences in luminance between pixels. When it comes to color...so long as you have control over the low-level RGB tone curves applied to a RAW image when it is rendered, you could technically achieve any level of color fidelity you wished with just about any camera.


In terms of file performance, the number of bits per pixel limits the color depth of your camera. We talk about color depth here in bits (8-bit, 12-bits, 14-bits, and so on), and the most tangible place to see this is in the RAW files produced by your camera. To the best of my understanding, this is directly related to the bit depth produced by each photosite on the sensor. Each photosite, then, produces a weak signal or a strong signal corresponding to light that falls on it, and this signal is represented digitally in a stair-step function according to a number of bits. One bit would give you "off" or "on", two gives one of {0,1,2,3}, and so on.

This technical capability is only part of the performance envelope, though. The construction of the sensor also determines how sensitive it is to light falling on it, such that you could have one 14-bit sensor that performs better than another. Sensor size is the single biggest contributor to performance here, but technologies such as microlenses also help each photosite gather the maximum possible light.

Finally, optics can contribute to dynamic range performance, as well. Better lenses transmit more light, and they transmit light more neutrally -- across the spectrum of light and across the field of view, so they'll produce better dynamic range performance than a lesser lens.


It is the cmos/ccd and the electronic construction that limits the dynamic range, which is saturation/noise floor.

The bit-depth only tells how many steps are on the staircase from the noise floor to saturation. The in-bred white-balance of the bayer pattern and sensitivity spectrum of the sensor also comes into play, when you have to crop a certain level to keep white highlights as white - this effectively reduces the saturation level. No matter if the maximum pixel value is 255 or 65535 it doesn't change how much real-world light needs to get integrated to overcome the noise floor and how m

uch is needed to saturate the sensor. However, In the conversion from this maximum to making a beautiful 8bit image, it gives you more flexibility to play with.

The lens cannot contribute to dynamic range. Even if you looked through sunglasses, you still get the same DR, just just need a longer exposure to get there which in turn makes it easier to get clear images in a dynamic scene.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.