12bit or 14bit Compressed or uncompressed

I had always used the 14bit uncompressed mode on my DX cameras but in looking into the pros and cons there seem to be quite a few people that say that if does not matter that much so use the 12bit compressed. One benefit is that sine the files are smaller you can get more shots before the buffers fill.

Can you really see the difference in any of the modes in regular shooting and printing?

I lift the shadows a lot (shooting ISOless) and then you occasionally notice the difference between 14 and 12-bit.

If the buffer or diskspace is an issue, I'd switch to 12-bit and test it. If you don't see any difference during processing, there's probably no reason to use 14-bit.

Here's an article about the issue:

https://photographylife.com/14-bit-vs-12-bit-raw

And here's a reply from someone understanding the technical aspects behind the same issue:

31) martin reiser
MAY 29, 2015 AT 8:37 PM

The article stipulates that there is no (visible) difference between 12 and 14bit raw.

To argue from the total number of colours vs. what the eye is supposed to distinguish misses the point entirely!

The issue is coding of color values in a fixed point format. As a consequence, half of all code points are in zone 10, half again in zone 9. 10 halfings is 1 : 1000, thus in zone 1 are quite few distinguishable code points. That means quantisation noise and banding in the deep shadows, especially if these shadows are lifted.

This is definitively visible, but not with examples like those in this article showing full pictures at a resolution of 600px across.

If you print large (A0) it most definitively shows. Since storage space savings is not an issue, opting for 12 bits is clearly unwise.

Click to expand...

I shot these samples a few years back. My suggestion is to test drive your own camera, there's no blanket statement that covers all cases.


The 12/14 bit issue depends a lot on the camera. My D3x shows a big difference. It is debatable, whether it is because of the badness of the 12 bit read chain or of the goodness of the 14 bit chain. At least the frame rate takes a big hit in 14 bit mode and one of the plausible explanations is that the analog/digital converters run at slower speed causing lower read noise.


In some cameras, particularly Canons, the 14 bit mode is purely a vanity mode, or marketing gimmick, since the cameras have less than 12 stops of dynamic range and thus the extra 2 bits in 14 bit mode are just random noise. But it looks good in an ad.


Test your camera. Perhaps your camera can deliver the same quality in 12 bit and 14 modes, perhaps not. Here are samples from mine. First a reference shot exposed correctly, then all the rest 7 stops underexposed. NEFs available if someone wants to play with them.


Note, that this is only a 7 stop push. In a typical landscape shot there is way more dynamic range than meager 7 stops. What this demonstrates is that 14 bit mode with D3x delivers a lot cleaner shadows.


Reference shot correctly exposed




Test shot 7 stops underexposed OOC




7 stops underexposed shot pushed 7 stops in processing (14 bit)




7 stops underexposed shot pushed 7 stops in processing (12 bit)




100% crop (14 bit)




100% crop (12 bit)




Whether the lossy compression of a camera throws away so much data that it becomes visible is another question altogether. The difference in files size between lossy and lossless compression is almost purely academic, IMHO, and I don't bother with lossy. I shoot JPEG fine when the RAW buffer is too small, which yields virtually endless buffer capacity. In 99% of the time the lossless 14 bit RAW buffer is adequate (78 shots @ D4S and 36 shots @ D3S).

If you want to recover a lot from shadows, then 14 bit will give slightly better results. Another area where 14 bit is better is in colour gradation. There are four times more shades in 14 bits than in 12 bits, that means that there will be less posterization of the blue sky.

That apart for normal shots it makes no perceptible difference. If you want more burst speed then go for 12 bits, other wise stick to 14 bits (you never know when you need those extra 2 bits!), especially with SD card prices going down it makes little sense to skimp on storage.

I shoot the d750 in 14bit RAW. I use highlight metering a lot (which underexposes the shot) and I need to pull shadows. I have however not tested the difference of the buffer between 12 and 14 bit RAW.

pforsell said:

i shot these samples a few years back.

100% crop (14 bit)
100% crop (12 bit)

Click to expand...

this that
WOW!
.....
.....
.....

Blacktop said:

I shoot the d750 in 14bit RAW. I use highlight metering a lot (which underexposes the shot) and I need to pull shadows. I have however not tested the difference of the buffer between 12 and 14 bit RAW.

Click to expand...

I shoot the same way (14-bit RAW, Highlight Protected metering) and while I've never tested the buffer, per se, neither has it ever let me down. Still, it's an interesting question...

On my D7100 the difference might buy me one more shot before the buffer fills, meaning I get 7 instead of 6 on CH. I've only filled a memory card once in my life, and given the cost of a 64GB card I'd much rather have the benefits of the 14-bits as shown above than the hassle of putting a second card in.

I went through the same questions with the D7100 and went with the 14bit uncompressed. But reading the articles recently about this issue make me question it again. The savings in storage are significant and the buffer issue is also a factor I ran into this issue last week when I was shooting a kayak going over a waterfall and I got the final shots of the approach to the falls and only about 2 going down when with the speed of he D750 I should have had 10 or 15 on the face of the falls. Yes I know I should have set it to JPEG Fine Large but I did not think about it at the time.
Guess I really need to do a little testing to see if it does make any difference at all in the kind of shooting I actually do.

I think that the prevailing wisdom is that if you're exposing properly then there is almost no perceptible difference, so if you're confident in your ability to do that then I suspect you'll be more than fine, particularly in the less noisy ISO ranges. But if, like me, you're subject to lazy photographer syndrome and you grab that one shot with absolutely the wrong settings because you didn't bother to check your camera when you grabbed it, well, then it can save your butt. I lean towards the latter, but I'll keep this in my pocket the next time I'm out with the D7100. I don't shoot very often on CH, so it'll be interesting to see what difference it makes on both buffer fill and IQ.

BackdoorHippie said:

I think that the prevailing wisdom is that if you're exposing properly then there is almost no perceptible difference, so if you're confident in your ability to do that then I suspect you'll be more than fine, particularly in the less noisy ISO ranges. But if, like me, you're subject to lazy photographer syndrome and you grab that one shot with absolutely the wrong settings because you didn't bother to check your camera when you grabbed it, well, then it can save your butt. I lean towards the latter, but I'll keep this in my pocket the next time I'm out with the D7100. I don't shoot very often on CH, so it'll be interesting to see what difference it makes on both buffer fill and IQ.

Click to expand...

Here is an interesting chart comparing the buffer capacity of the different modes in both the D610, D700, D750 and D810

https://photographylife.com/nikon-d750-buffer-capacity


So it makes a pretty good difference

of course in JPEG Fine is maxes out at 100 on all the models.

There are some pretty fast cards out there which help too. The faster they write the load, the more flows through the buffer before being filled.

J-see said:

There are some pretty fast cards out there which help too. The faster they write the load, the more flows through the buffer before being filled.

Click to expand...

I have some of the fastest cards made in there now.

I myself rarely have issues with the buffer on the D750 but I also rarely shoot larger than 4 or 5 shots in a burst. When I ran dry it usually was while birding and shooting a couple of bursts too quickly.

What I never tried but what could be an option is lowering to 1.2x. Since I rarely fill the frame during birding, cutting some of it might just give that extra buffer. I end up cropping it afterwards anyways.

Don Kuykendall said:

Here is an interesting chart comparing the buffer capacity of the different modes in both the D610, D700, D750 and D810

https://photographylife.com/nikon-d750-buffer-capacity

NEF (RAW), Compressed, 12-bit in FX the buffer is 33
NEF (RAW), Lossless compressed, 14-bit the buffer is 15

​

So it makes a pretty good difference

of course in JPEG Fine is maxes out at 100 on all the models.

Click to expand...

Only once have I run into issues shooting 14-bit lossless uncompressed on the d750, and it was in a pure pray and spray situation.

The gain in Raw file size (which is truly all I care about) from 14-bit lossless to 12-bit compressed seems to be about 25-33% depending on the camera and image content, which is significant. But in terms of the D7100 that means, perhaps, 1 or 2 add'l frames before the buffer clogs. Again, it's a factor of what you'd rather have in the situation, the extra frames or the recovery? It's enough that it might be worth using a User setting for it for Birds in Flight rather than hunting around in the menus.