8 bit vs 16 bit, and RGB modes

I've been taking photos for about 5 years now and only just noticed that when I open my raw files in Photoshop I've got it set to open them as 8 bit rather than 16 bit. My camera records 14 bits, so obviously by opening my images as 8 bit I'm losing some information, but would it lead to a noticeable loss in image quality?

Also, what colour mode should I be using? sRGB IEC6:1966-2.1? Or the profile for my monitor? Up until now it's been set on adobeRGB (1998)...

Many thanks

You'll loose color gradient resolution, not pixel information. This can sometimes be very noticeable when taking sunset skies or other images where increased color resolution would smooth transitions.

Thanks for the reply. So the banding I've seen in some skies could be worse due to the 8 bit setting then?

jack880 wrote in post #17969530
Thanks for the reply. So the banding I've seen in some skies could be worse due to the 8 bit setting then?

Yes, most definitely.

I feel like a fool...!

jack880 wrote in post #17969546
I feel like a fool...!

Well cheers and congrats for opening up a new realm of image quality!

Consider that some of the banding seen in skies is most likely due to too much data compression, via Quality values like 3 or 4 rather than 8 or 9 when the JPG is created, and not at all due to too few bits for color encoding.

As for color differention limits, due to limitations of the printing equipment, it might be impossible to see a gradient of hues between
100000-->100001-->100002-->100003-->100004, rather than simply seeing the progression of hues 25000-->25002-->25004

Put into terms of a different sense, does your skin detect a temperature change of 0.1°F?! Think of the 8-bit per color 16.7 Million hues like sensing temperature shifts of 1°F, vs. the 16-bit-per-color ability to detect/store values as small as 0.1°F

We drank the coolaid about the aRGB ability to hold an extended range of hues compared to sRGB, yet BOTH still only support 16.7 Million different values! And it is very difficult to find a commercial printer who prints aRGB without converting (and losing data) the file before printing. If we truly had aRGB be superior to sRGB, it should have 33.4 Million hues, so that it can portray all of sRGB -- plus more hues.

So 281 Trillion hues (16 bit) is nothing more than a gleam in daddy's eye today, in the hopes that the future ever expands beyond 16.7 Million hues of 8 bit color.

BUT then you have to stop and remember that the human eye can apparently only detect TEN million colors (and the CIE findings of the 1930s claimed only the ability to detect about 2.8 Million colors.) Do I hear 'overkill'?!...we already can store 6.7 Million more hues than our eyes can see!

Wilt wrote in post #17969680
Put into terms of a different sense, does your skin detect a temperature change of 0.1°F?! Think of the 8-bit per color 16.7 Million hues like sensing temperature shifts of 1°F, vs. the 16-bit-per-color ability to detect/store values as small as 0.1°

Remember, its binary. 16 bit color is not double 8, its 256 times 8 or working your analogy, if each color of a jpeg is 1 degree, in 16 bit it would be .004, or, for each color and its adjacent neighbor in a jpeg you have 1 graduation whereas in 16 bit color, you would have 256.

Most of the time it doesn't make a difference but for images with lots of shades, like sunsets, and you start processing, you can get the gradation effect quite easily. Good rule of thumb I use is if I am working from raw, my work flow stays 16 bit until the last step where I convert to jpeg.

gjl711 wrote in post #17969733
Remember, its binary. 16 bit color is not double 8, its 256 times 8 or working your analogy, if each color of a jpeg is 1 degree, in 16 bit it would be .004, or, for each color and its adjacent neighbor in a jpeg you have 1 graduation whereas in 16 bit color, you would have 256.

Yes, 281 Trillion vs. 16.7 Million. My analogy was not a perfect one, nor was it meant to be.

gjl711 wrote in post #17969733
Most of the time it doesn't make a difference but for images with lots of shades, like sunsets, and you start processing, you can get the gradation effect quite easily. Good rule of thumb I use is if I am working from raw, my work flow stays 16 bit until the last step where I convert to jpeg.

but we need to also remember that at 8-bit per color we already have more than enough to surpass our eyes' ability to detect....just gotta keep JPG compression from defeating us.

I can consistently get '4' wrong on the Xrite color test. Others have scored 3 or 4 or 7. The best score I have heard someone actually achieve was '0' but upon retest he only scored '24'. Many other results in the 25-150 range! so with 8-bit color we don't often perceive differences.

Lots and lots of discussion on this topic--and a LOT of it is not well based. Highly recommend http://www.digitaldog.​net/ as your authoritative resource. Bottom line -- no matter what anyone says, look for actual, physical, proof.

I now stay in ProPhotoRGB all the way through my workflow and printing with 16bit TIF (I do my own printing). For web, I drop to sRGB for a lower res jpg.

Wilt wrote in post #17969740
Yes, 281 Trillion vs. 16.7 Million. My analogy was not a perfect one, nor was it meant to be.

but we need to also remember that at 8-bit per color we already have more than enough to surpass our eyes' ability to detect....just gotta keep JPG compression from defeating us.

I can consistently get '4' wrong on the Xrite color test. Others have scored 3 or 4 or 7. The best score I have heard someone actually achieve was '0' but upon retest he only scored '24'. Many other results in the 25-150 range! so with 8-bit color we don't often perceive differences.

Actually most people are probably doing the Xrite test on a 6 bit output device. So that is only 64 shades per channel, or a total of 262144 colours. I think I scored in the mid 20's but I would have to look the thread up where it was being discussed to be sure.

Even so when working on images in post I will try to keep the image "RAW" for as long as possible. Once it has been converted to an RGB format I will then want to keep it 16 bit for as long as possible to minimise arithmetic errors during processing. Ideally only dropping down to an 8 bit format for the final output JPEG files.

When it comes to colourspaces, I process in LR, and so as we all do when using that program work in it's internal colorspace of ProPhotoRGB with linear primaries. If I have to work in Photoshop, and I know that I will also be processing that file further in LR later I will stick with 16 bit and use ProPhotoRGB for minimum colourspace conversion. Final output will usually be a JPEG in sRGB for posting on the web, or possibly in a printer space for upload to the pro lab that I use for printing, where they have provided a profile, but do not do profile conversion for you.

On the odd occasion that I might go from LR to PS then direct to output, then I would initially process in PS in 16 bit, but I would use whatever my final output colourspace will be from the start.

Alan

Since the OP didn't say whether he is doing his Raw processing in LR or ACR, and to avoid any confusion, it should be noted that the internal workings of LR/Develop and ACR are identical, including the fixed working space. The only difference (and it is confusing) is in the preview data sent to the monitor and used for the histogram. In LR it is in a hybrid space called Melissa RGB (unless soft proofing is on) and in ACR it is in whatever space you have set as PS's working space (unless you change it in the blue link in the center of the bottom margin.)

@jack880

Never your monitor profile. That is unique to your monitor and using it in an image file (rather than a universal ICC space) could cause everybody you share the photo with to see it improperly rendered.

Kumsa wrote in post #17969884
Lots and lots of discussion on this topic--and a LOT of it is not well based. Highly recommend http://www.digitaldog.​net/ as your authoritative resource. Bottom line -- no matter what anyone says, look for actual, physical, proof.

I now stay in ProPhotoRGB all the way through my workflow and printing with 16bit TIF (I do my own printing). For web, I drop to sRGB for a lower res jpg.

Do the majority of your photos contain colors beyond sRGB?

... and, here we go.

So, short answer is, apparently so.

If you want to follow a tortured discussion on the subject, please feel free to read: http://www.dpreview.co​m/forums/thread/396433​7

I've done my printing, and I do see a difference. Is it something that people would come up and say "amazing, I've never seen anything like these colors !" No. But it's there, and I see the difference, and once seen, I don't want to lessen my presentation. My prints are pigment on archival paper and I hope that long, long after I'm gone, they will still be viewed.

http://www.digitaldog.​net/ has test files and explanatory video. Smart enough to have a technical book published on it. http://www.amazon.com …v=glance&s=book​s&n=507846

However, in theory there is no difference between theory and practice, right ? So, print away and see. I've been working on some orchids -- and ProPhotoRGB TIFF to the printer is an improvement. In fact, the color was better than my color on my calibrated monitor. (I was told this would happen with some color regions.) I had to do some calibration test prints to be sure.