Basics - Opening the RAW file in Photoshop - Why 16 Bits?

This is a personal choice, but I almost only use my raw converter (Adobe Lightroom) for its picture library manager.

Before opening

Before opening my RAW in Photoshop CS2, I adjust the white balance and I take care to reset all others default settings to preserve the maximum original data from the sensor and thus the highest dynamic range. I especially pay attention to disable the sharpen setting because I’m about to edit my picture and some treatments give bad results with a sharpened source. Sharpening is, most of the time, the last thing I do on the flatten picture, here is why :

• Spep 1 : On the left the exported picture in photoshop with sharpening, on the right the same picture without sharpening
• Step 2 : I add the same heavy postprocessing on both pictures to bring out the problem
• Etape 3 : We can clearly see that the postprocessing showed up white outlines around the chimney, it also accentuated the noise in the sky
• Etape 4 : I add final sharpening to the right picture, this is perhaps a question of taste but I much prefer this result

 

Editing in 16-bits per channel

My RAW source has a 12-bits precision, so if I export it in 8-bits I cut off 4-bits of precision per channel, in other words I exclude some shades from my picture. So if I want to preserve all data from this source, this is quite evident that I need to work in a space wide enough to host its full depth. Under Photoshop the next bit depth is 16-bits per channel and, because 12-bits is a subset of 16-bits, this is lossless.

Did you already ask yourself “I am working in 16-bits, why the hell do my color picker still show me values from 0 to 255 (2⁸ = 8bits) for each channel instead of 0 to 65535 (2¹⁶ = 16bits)?”, no ? because I really wondering why .

I found three possible answers on the web :

• The first one I found was “Because a histogram that would show a scale with 65536 discreet levels would be too big, you would need a display that is 65536 pixels wide”, I found this one quite disputable because it is just a matter of scale and that’s always possible to add a zoom option to that histogram.
• The second one was “If you display a gradient from blue to black with 256 shades and another with 65536 shades, there is a few chance you can see the difference, so the internal process of Photoshop is indeed in 16-bits but the screen output is rendered in 8-bits”, I am more convinced by this one but I already read that the human eye can discern about 500 shades of a single color, in other words approximately 9-bits
• The third one is very close to the second one, it was something like “Monitors and video cards are working in 8-bits per channel, so it is useless to ask them to display a 16-bits picture since they are not capable of that, the internal process of Photoshop is indeed in 16-bits but the screen output is scaled to 8-bits”, I accepted this answer but now I just wonder how you can take advantage of these famous laCie monitors which claim to display a 12-bits color depth seeing that the screen output of photoshop is 8-bits, video cards are 8-bits (I saw that some of them are able to display a 10-bits precision … but 10 is not 12), so finally is the accuracy of these monitors inexploitable or do I miss something ? If you know more about this subject, please share it to us .

In short, the most important thing is that even if the screen output, the preview is scaled to 8-bits, we really work in a 16-bit space and that’s something easy to verify:

1. I take a gradient from blue to black, no visible difference between 8 and 16-bits

2. Now I select a small portion of this gradient and I resize it to 100% wide to get a new gradient with a very small amplitude, still no visible difference between 8 and 16-bits and that’s normal

3. No I add a levels adjustment layer and I adjust it to contrast the picture and to bring out the quantity of shades recorded in this “mini gradient”

4. The reassuring result, there is many more steps in 16-bits than in 8-bits

I edit my pictures using a non-destructive process, I mean that my original is always the first layer and I stack adjustment layers above it.
In a 8-bits/channel color mode you can quickly have a visible deterioration of your picture with some layers combinations, mostly if you are using multiple layers with gradient masks like I do. Sure this is often our fault like a bad order in layers for example … but in 16-bits, I rarely reach other limitations than those of the source itself or you must do it intentionally , in short, editing in 16-bits per channel offers more flexibility.

The purpose of the exaggerative example below is to show that, despite of an intentional degradation of the source (heavy uncontrast + heavy contrast), the depth of 16-bits allow a much better preservation of the picture.

Share This

This entry was posted on August 3, 2007 and is filed under Digital Photography Postprocessing. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.