14 bit lossless vs 12 bit lossy

[jaomul]

Hi all,

I can't really see a difference and I will switch over to the lesser file if I want to increase the buffer (d7100). Is there situations where the bigger file would make a difference to the picture quality I get or is it a case of it being so minute that the eye cannot see a difference?

 

[Trever1t]

I can't give you the # of bits of color information increased by the 2 bits but it's pretty substantial. If you are shooting sports or other activities where speed is more important than I understand and would probably go with the 12 bit but for all other circumstances would always want the most information possible per file.

 

[jaomul]

I agree. I always set up for max info except where I need a bigger buffer, but have not seen a quality difference. Thanks

 

[Tinderbox (UK)]

Not an bad example.

12Bit Vs 14Bit Raw And Compressed Vs Uncompressed? Does It Matter? | DIYPhotography.net

John.

 

[480sparky]

You won't see any difference until you start to engage in some heavy-handed editing.

 

[Solarflare]

Err ... of course you cant see a difference !!!

You cant see a difference between RAW and JPEG, either !

While your eyes would be able to see them, your monitor is simply unable to display any such differences. Its usually only 8 bit (meaning its usually true color, which is 24 bit, which is 8 bit per color channel, while your camera records 14 bit per color channel).

The point of RAW is to give you the full data of the sensor, so you can edit it according to your wishes. And with 14 bit and lossless you'll get even some more room for changes than with 12 bit and lossy.

But if you just look at the result and dont edit anything, you might as well go with the JPEG.

 

[jaomul]

Err ... of course you cant see a difference !!!

You cant see a difference between RAW and JPEG, either !

While your eyes would be able to see them, your monitor is simply unable to display any such differences. Its usually only 8 bit (meaning its usually true color, which is 24 bit, which is 8 bit per color channel, while your camera records 14 bit per color channel).

The point of RAW is to give you the full data of the sensor, so you can edit it according to your wishes. And with 14 bit and lossless you'll get even some more room for changes than with 12 bit and lossy.

But if you just look at the result and dont edit anything, you might as well go with the JPEG.

Ok fair enough, I wont get drawn in by your smart arse first line because to be fair I did not explain myself fully.

I do not see a difference in print between the 12 bit lossy and 14 bit lossless in the sample few pictures edited and printed at similar settings having converted them to max quality jpegs in lightroom. Thanks all for your input

 

[480sparky]

Ok fair enough, I wont get drawn in by your smart arse first line because to be fair I did not explain myself fully.

I do not see a difference in print between the 12 bit lossy and 14 bit lossless in the sample few pictures edited and printed at similar settings having converted them to max quality jpegs in lightroom. Thanks all for your input

It's not a smart-arse reply. It's facts.

The reason you can't see any difference is because a typical human eye & brain cannot perceive the difference. To show you what I mean, can you see the difference between the left and right side of this image?:

I'll give you a hint: The right side is darker.

And it's darker by a factor of one bit.

JPEGS are what is called an 8-bit depth. This means there are 2 ^ 8 possibilities for each color channel. Or, 2x2x2x2x2x2x2x2 , or 256. Each of the 3 color channels (red, green and blue) have the possibility of being 256 shades of that color.

Using the above image, it's hard for most people to even SEE the difference from one shade of color and the next. One side is 183, the other side is 184, and that's out of a possible total of 256.

So imagine trying to perceive the difference between adjacent colors when you have 4,096 possibilities (12-bit color depth). I doubt there's a human in existence that is capable of doing that visually. Upping the bit depth to 14 (2x2x2x2x2x2x2x2x2x2x2x2x2x2) gives you 16,384 possible colors in each channel. An impossible task, I will venture to say, to tell the difference with your eyes alone.

That, and not many people will own a monitor capable of even displaying such a subtle difference.

 

[jaomul]

Ok fair enough, I wont get drawn in by your smart arse first line because to be fair I did not explain myself fully.

I do not see a difference in print between the 12 bit lossy and 14 bit lossless in the sample few pictures edited and printed at similar settings having converted them to max quality jpegs in lightroom. Thanks all for your input

It's not a smart-arse reply. It's facts.

The reason you can't see any difference is because a typical human eye & brain cannot perceive the difference. To show you what I mean, can you see the difference between the left and right side of this image?:

I'll give you a hint: The right side is darker.

And it's darker by a factor of one bit.

JPEGS are what is called an 8-bit depth. This means there are 2 ^ 8 possibilities for each color channel. Or, 2x2x2x2x2x2x2x2 , or 256. Each of the 3 color channels (red, green and blue) have the possibility of being 256 shades of that color.

Using the above image, it's hard for most people to even SEE the difference from one shade of color and the next. One side is 183, the other side is 184, and that's out of a possible total of 256.

So imagine trying to perceive the difference between adjacent colors when you have 4,096 possibilities (12-bit color depth). I doubt there's a human in existence that is capable of doing that visually. Upping the bit depth to 14 (2x2x2x2x2x2x2x2x2x2x2x2x2x2) gives you 16,384 possible colors in each channel. An impossible task, I will venture to say, to tell the difference with your eyes alone.

That, and not many people will own a monitor capable of even displaying such a subtle difference.

Thanks 480sparky, as always your explanations are excellent and easy to follow.

I will disagree with one point though, when someone starts a response with err.. and finish with !!! it is in my opinion being smart ass. The facts are correct as you said and when i ask here i am happy to learn.

 

[bc_steve]

I think if you edited a photo to the point where the sky becomes "posterized" you'd be running into the limits of 12bit.

 

[480sparky]

I think if you edited a photo to the point where the sky becomes "posterized" you'd be running into the limits of 12bit.

The vast majority of shooters never edit an image that far.

 

[Derrel]

I've read so many serious in-depth reviews where the author says he, "Thinks he might see subtle differences," to realize that for me, 14-bit .NEFs are not necessary. I shoot 12 bit, compressed .NEF pretty much all the time. When authors with deep technical understanding and ability think they "might see subtle differences", it's a moot point for most shooting situations I think.

 

[Solarflare]

I will disagree with one point though, when someone starts a response with err.. and finish with !!! it is in my opinion being smart ass.

Okay, noted.

Anyway, I would like to add some more comments about two concepts:

- color depth

- dynamic range

Color depth means how precise your camera measures colors. The human eye can distinguish about 8 bit here (256 values), for each color, for a total of 24 bit. We call that "truecolor" and thats what todays monitors are pretty much always using.

Dynamic range means how many values of bright and dark the camera can distinguish. This is more. The human eye can distinguish about 14 stops here (a stop means double or half the intensity). The human eye can manage to distinguish up to about 14 stops, but also up to 24 stops if it has time to accustom itself to a scene (and if you close your eyes after looking at such a scene, you can still see a residual image, because different areas of your eyes have adopted to different brightness levels).

In both cases, the values of the human eye is also what the best current image sensors are able to manage (except they have no such thing such as variable ISO, thus they really only can manage up to about 14 stops of dynamic range right now).

Compare:

DxOMark: Color Depth of current cameras
DxOMark: Dynamic Range of current cameras

Meaning, unless you have a scene with a high dynamic range, and edit the raw to do a HDR, i.e. brighten the dark areas - you wont benefit from 14 bit too much. Its advantages are really "hidden in the dark" and at the limit what we can see ourselves, with our own eyes.