Bit Depth and BPP?

[j.weegee]

Okay so I have been looking at camera specs and confusion has arisen when I come across bit depth. Now I know 24bpp is bits per pixel, but when a camera has a bit depth of 14, is it referring to bits per pixel or bits per channel? Is safe to assume that bit depth on a list of specs for a camera just refers to bits per channel unless the suffix bpps follows the number?

 

[Helen B]

Yes, it is bits per channel, but as there is only one channel in the Raw file from a Bayer filter camera it is also bits per pixel in that case.

 

[KmH]

24 BPP means 8-bits per the 3 RGB color channels used to render an image pixel. 3 channels x 8 bits/channel = 24 bits per pixel. JPEGs are often describes as 24-bit images using the same math - 3 channels times 8-bits/channel.

Digital camera image sensors cannot record color. Digital camera image sensors can only record luminosity.
The Bayer Array filter most DSLR cameras have in front of the image sensor is used to infer by interpolation what color of light fell on a pixel or group of pixels.
Info about the Bayer Array and digital camera image sensors - Understanding Digital Camera Sensors

In this group of tutorials is a tutorial that discusses Bit Depth - Tutorials on Color Management & Printing

 

[j.weegee]

That makes sense but what confuses me is when I look at the specs for a camera and it says "bit depth: 14 bit" Am I supposed to just assume that is 14 bits per channel therefore 42bpp? Or is there something I'm missing here?

 

[KmH]

You're missing a lot about how a digital image is made. If you haven't already, see post #21 in this thread - http://www.thephotoforum.com/forum/photography-equipment-products/303010-ccd-analog-2.html

When it says a camera has a 14-bit bit depth, they are referring to the output capacity of the A/D converter (Analog to Digital converter). The A/D converter changes the analog pixels voltages to a digital number that the image processor in the camera can use.

Many DSLR cameras offer a choice of both 12-bit and 14-bit A/D converter output. Usually some variety of both lossless and lossy Raw file compression are also offered.

12-bits can only represent 4096 levels/values of tone. 14-bits can represent 16,384 levels/values of tone.

Bit depth is a major reason for shooting Raw. JPEG only allows an 8-bit depth (256 levels/values of tone). So, if a camera's A/D converter can output 16,384 levels, 16,128 levels per color channel have to be discarded to make a JPEG. As a point of fact, in the conversion to JPEG only color data is discarded, because it is not as needed for human perception. Luminosity data has to be retained to keep a photo looking like what we see.

Raw converters can output 16-bit depth images. However, the additional 2 or 4 bits are for metadata of one type or another, and are not added image data.

Photoshop can do some 16-bit depth edits. Since the extra 2 or 4 bits aren't image data Photoshop doesn't use all 65,536 levels that 16-bits can represent. Photoshop only uses 32,768 levels per channel (0 = black, 32,768 = white)

 

[480sparky]

Photoshop only uses 32,768 levels per channel (0 = black, 32,767 = white)

Fixed it for you.

 

[Helen B]

Raw converters can output 16-bit depth images. However, the additional 2 or 4 bits are for metadata of one type or another, and are not added image

You keep saying that, and I keep trying to point out that it is not true. The metadata is not held in the bits used to pad the file to 16 bits, it is in its own separate area.