Home · Register · Join Upload & Sell

Moderated by: Fred Miranda
Username  

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Nikon Forum | Join Upload & Sell

1      
2
       3       4       5       end
  

Archive 2012 · D800: 12-bit RAW vs. 14-bit RAW

  
 
taob
Offline
• • •
Upload & Sell: Off
p.2 #1 · p.2 #1 · D800: 12-bit RAW vs. 14-bit RAW


Mark_L wrote:
I guess the (visible) difference may be in working the files hard in post like pulling out shadow detail or big contrast increases.


Yep, and only under certain conditions are some of those effects visible. In most cases, image detail and noise dithering effectively hides posterization effects from an insufficient bit depth. The human eye and brain are also very good at post-processing away those artifacts, so again the argument comes back to "what data is missing" vs "what can you actually see".

I'll post more examples of extreme adjustments later, but it is time to make dinner now.



Apr 08, 2012 at 06:11 PM
lxdesign
Offline
• • • • •
Upload & Sell: Off
p.2 #2 · p.2 #2 · D800: 12-bit RAW vs. 14-bit RAW


Thanks Brian for your research in RAW file compression with the D800. Based on your results, I'll probably shoot 12bit most of the time.... And If I feel its going to be special, I will switch to 14bit RAW. Now.. I just need to get my camera! Hopefully I'll get a call soon.


Apr 08, 2012 at 06:16 PM
Jan Brittenson
Offline
• • • •
Upload & Sell: On
p.2 #3 · p.2 #3 · D800: 12-bit RAW vs. 14-bit RAW


If you look at the histogram of a linear (gamma 1.0) image you'll notice that the highest 1 stop takes up the entire right half of the histogram. The top 3 stops take up 7/8ths of it. The remaining 11 stops, assuming 14 stops of DR are squished into the leftmost 1/8th. In a 16-bit file the lowest stop can have a value of 1 and nothing else. (0 being everything too dark to register and not on the scale.) The next stop can have the value 2. Then 3 or 4. So the 3 lowest stops have exactly 4 different values to differentiate them. The top 1 stop by comparison has 32768 different values.

You can't see squat without applying a gamma greater than 1 to an image. When you do that 1 highest stop is going to lose that extreme amount of differentiation. It might drop to 2000 values or so. Not only can't you tell the difference, but you HAVE TO do this when you process the file. There is no other option this side of HDR. You HAVE TO throw away over 90% of the information in the highest 1 stop to even produce a viewable image.

What NEF (and Leica DNG) compression does is throw away some 80% of this data before even writing the file. It does it by using a lookup table that leaves the vast majority of the scale completely untouched; the bottom 8 or 9 stops or so. From then it slightly reduces the bit depth of the image until the top where it maxes out. It has NO effect on shadow detail. It ONLY throws away what you're forced to throw away when you process the image. There is no way to keep it. The compression just throws it away before the file is written rather than later when it's processed into an image.



Apr 08, 2012 at 08:04 PM
mshi
Offline
• • • •
[X]
p.2 #4 · p.2 #4 · D800: 12-bit RAW vs. 14-bit RAW


You can buy 128GB SD card for less than $200.


Apr 08, 2012 at 11:36 PM
brett maxwell
Offline
• • •
Upload & Sell: Off
p.2 #5 · p.2 #5 · D800: 12-bit RAW vs. 14-bit RAW


I've done controlled tests with a D3 comparing 12 and 14 bit lossless compressed. Even pushing and pulling files by 3 stops in Lightroom I could not perceive a difference, so I shoot 12 bit.




Apr 09, 2012 at 12:15 AM
JimFox
Offline
• • • • • • •
Upload & Sell: Off
p.2 #6 · p.2 #6 · D800: 12-bit RAW vs. 14-bit RAW


taob wrote:
Let me first start off by saying yes, there is a difference between 12-bit vs. 14-bit, and lossy vs. lossless. No question there, since the file sizes suggest this quite clearly. The real question (to be made individually by each photographer) is whether those differences are relevant to their shooting style.

Okay, with that out of the way, let me start with some simple examples. Here are 9 crops at 100% from the D800 at ISO 6400. To make things easier, I've only used the extreme raw compression settings: either 12-bit lossy (about a 35MB NEF) or 14-bit lossless (about a
...Show more

Thanks for this work Brian. I would be interested in seeing your more extreme examples. I shoot primarily Landscape, so getting the most dynamic range I can out of a shot, and being able to pull as much detail from the extremes of a shot are so important to me. I do remember testing the difference between 12 bit and 14 bit before. It was a few years back, not sure now which camera it was, whether it was the D200 or the D700, but yeah, some shots, when the light range fell well within the dynamic range of the camera, the difference was not really noticable. But in those cases, where I was trying to recover more detail from shadows and highlights, especially highlights, is where I found the 14 bit did give me that slight bit more detail I could recover.

So it will be interesting to see what you discover here. I appreciate your time on this.

Jim



Apr 09, 2012 at 01:57 AM
taob
Offline
• • •
Upload & Sell: Off
p.2 #7 · p.2 #7 · D800: 12-bit RAW vs. 14-bit RAW


JimFox wrote:
But in those cases, where I was trying to recover more detail from shadows and highlights, especially highlights, is where I found the 14 bit did give me that slight bit more detail I could recover.


I'm running off to a couple of shoots right now (going to try out the D800 with an Eye-Fi card uploading to my Touchpad!), so I won't be able to post any more examples at the moment.

Yes, the stronger compression options tend to affect only the extremes of an images dynamic range. In a nutshell, this is what you need to remember, at least when it comes to NEFs:

12-bit vs 14-bit: affects shadows
lossy vs lossless: affects highlights

... but only under extreme adjustments!

For landscapes, you might need to pull as much saturation and contrast out of an otherwise uninteresting sky. At the same time, there might be, say, interesting detail in a rock face that is in deep shadow. I would argue that bracketing your exposures is still the best way to go, but if you had to capture everything in a single shot (e.g., you're dealing with moving animals or other features), 14-bit lossless is the way to go. Besides, I'm guessing most landscape folks don't come home at the end of the day with several thousand shots to go through like a wedding or sports photographer might. So whether your files are 30 MB each or 50 MB each won't have as much of an impact.



Apr 09, 2012 at 07:13 AM
Mark_L
Offline
• • • • •
Upload & Sell: Off
p.2 #8 · p.2 #8 · D800: 12-bit RAW vs. 14-bit RAW


I'd be interested to see a 12bit vs 14bit comparison on a low key studio background. As I've mentioned here before, even with 14bit I've had banding with a D700.


Apr 09, 2012 at 08:31 AM
fscherz
Offline
• •
Upload & Sell: Off
p.2 #9 · p.2 #9 · D800: 12-bit RAW vs. 14-bit RAW


brett maxwell wrote:
I've done controlled tests with a D3 comparing 12 and 14 bit lossless compressed. Even pushing and pulling files by 3 stops in Lightroom I could not perceive a difference, so I shoot 12 bit.



I tested as well and came to the same conclusion. It's just 20% less space but also 20% faster in saving/moving/loading. I saw so many 5Dx or D700 shooters doing JPEG, the small difference between 12 & 14 Bit (lossless) compressed will not help me to make better fotos



Apr 09, 2012 at 04:03 PM
taob
Offline
• • •
Upload & Sell: Off
p.2 #10 · p.2 #10 · D800: 12-bit RAW vs. 14-bit RAW


Okay, let's look at lossless vs lossy NEFs first, which mainly affects the highlight regions of an image. I often hear people say that 14-bit images are better at recovering highlight detail, and give smoother tonal gradations in otherwise featureless expanses, such as clear skies.

Here are a bunch of sample images, all shot with the D800, converted to lossless DNG, then processed in ACR 7.0. Aside from the stated adjustments, I will rotate and align candidate images for easier A/B comparisons, and I will get rid of a couple of sensor dust spots.

100% crops can be downloaded here:

http://luxography.ca/Images/FM/hr_tests.zip
http://luxography.ca/Images/FM/sky_tests.zip

This building was shot 2 stops overexposed, then brought back down 2 stops. Before and after:
http://luxography.ca/Images/FM/hr1_orig.jpg

http://luxography.ca/Images/FM/hr1_fix.jpg


For those concerned with flesh tones, here are shots of my hand in sunlight, again shot 2 stops overexposed and then corrected in ACR. Before and after:

http://luxography.ca/Images/FM/hr2_orig.jpg

http://luxography.ca/Images/FM/hr2_fix.jpg


This is an early morning shot out my window, looking to the east. The colours are pretty drab, so I pumped it up with Contrast +100 and Vibrance +100. The second ZIP file above contains 100% crops of three strips across the image so you can examine the colour gradients. Notice any differences between each pair?

http://luxography.ca/Images/FM/sky1_orig.jpg

http://luxography.ca/Images/FM/sky1_fix.jpg


So far, not much difference at all between lossy and lossless NEFs. The only time I've noticed an advantage to shooting lossless NEFs is with pure blue sky, and again only as you start pushing the sliders really hard. Here is another view out my window, later in the afternoon:

http://luxography.ca/Images/FM/sky3_orig.jpg


I cranked up the Contrast and Vibrance both to +100 again. You can see some faint banding in the sky of the second (lossy) image:

http://luxography.ca/Images/FM/sky3_lossless.jpg

http://luxography.ca/Images/FM/sky3_lossy.jpg


Here's another example, late morning with the sun hiding behind an apartment building:

http://luxography.ca/Images/FM/sky4_orig.jpg


Contrast and Vibrance set to +100:

http://luxography.ca/Images/FM/sky4_lossless.jpg

http://luxography.ca/Images/FM/sky4_lossy.jpg



Note that if are shooting under different conditions or in a different location or time of year, you may get results SOOC that match my artificially saturated photos. In that case, a lossy NEF should be fine too, because the sensor is capturing those colours natively. Lossless NEFs make a difference if you start off with an undesirable capture and then really stretch what little colour you have. But if the sky already has that deep blue colour in real life, lossy vs lossless won't make a difference.



Apr 09, 2012 at 10:08 PM
JimFox
Offline
• • • • • • •
Upload & Sell: Off
p.2 #11 · p.2 #11 · D800: 12-bit RAW vs. 14-bit RAW


fscherz wrote:
I tested as well and came to the same conclusion. It's just 20% less space but also 20% faster in saving/moving/loading. I saw so many 5Dx or D700 shooters doing JPEG, the small difference between 12 & 14 Bit (lossless) compressed will not help me to make better fotos


Depends what you shoot. And as far as others shooting jpg's, just because others do it doesn't mean that you should too. For me, the small difference is huge when processing my landscape shots and trying to wring out every last bit of dynamic range. Yes I blend, but in landscape scenes where there is movement, ie leaves or water, being able to process a scene all off of one image, makes things go so much better when you are blending all off of one shot that can be pushed when converting from raw.

Jim



Apr 10, 2012 at 01:23 PM
JimFox
Offline
• • • • • • •
Upload & Sell: Off
p.2 #12 · p.2 #12 · D800: 12-bit RAW vs. 14-bit RAW


Hey Brian,

That is cool to see the difference. What I didn't know was how lossless vs lossy would affect the highlights. And even if it's only slight, there are times where the increase in file size would be worth it. Thanks for these examples!

Jim



Apr 10, 2012 at 01:24 PM
afm901
Offline
• •
Upload & Sell: Off
p.2 #13 · p.2 #13 · D800: 12-bit RAW vs. 14-bit RAW


If there were not a noticeable difference between 12-bit and 14-bit, Nikon would not offer 14-bit as an option. I can't understand the logic of using 12-bit if file size is your issue with it. Storage is relatively inexpensive.

I always shoot RAW. If I need JPEGs fast, I will use the first card for RAW and the second card to save JPEG. I would never ever use lossy compression. Again, storage is relatively inexpensive. I would never use 12-bit unless I needed the speed or the buffer space that setting may provide.

Scott



Apr 10, 2012 at 02:07 PM
mshi
Offline
• • • •
[X]
p.2 #14 · p.2 #14 · D800: 12-bit RAW vs. 14-bit RAW


If you take a serious look at 16-bit raw capture, 14-bit seems like a child's play IMHO

16-bit raw SOOC (H4D-40, 40MP)

http://farm8.staticflickr.com/7127/6882167906_d5055352f1_b.jpg

100% crop

http://farm8.staticflickr.com/7065/6882167904_fb0179128e_b.jpg



Apr 10, 2012 at 02:16 PM
slrl0ver
Offline
• • •
Upload & Sell: Off
p.2 #15 · p.2 #15 · D800: 12-bit RAW vs. 14-bit RAW


OP here.

taob, mshi: Thank you for taking the time to post pictures.

The point of the thread was to try and ascertain, on average, how much one is loosing with 12-bit vs. 14-bit and I think with some of the pictures that's easier to identify. I don't think there is anyone alive who will doubt that 14-bit or 16-bit or 24-bit has the theoretical potential to do "more" than 12-bit. But, practically speaking, what is the difference? In my experience, sometimes "good enough" is enough, and sometimes you want the absolute best with no compromises. It's all relative to what you're doing and what compromises you're willing to make or not make.

- slrl0ver




Apr 10, 2012 at 02:48 PM
mshi
Offline
• • • •
[X]
p.2 #16 · p.2 #16 · D800: 12-bit RAW vs. 14-bit RAW


slrl0ver wrote:
OP here.

taob, mshi: Thank you for taking the time to post pictures.

The point of the thread was to try and ascertain, on average, how much one is loosing with 12-bit vs. 14-bit and I think with some of the pictures that's easier to identify. I don't think there is anyone alive who will doubt that 14-bit or 16-bit or 24-bit has the theoretical potential to do "more" than 12-bit. But, practically speaking, what is the difference? In my experience, sometimes "good enough" is enough, and sometimes you want the absolute best with no compromises. It's all relative to what you're
...Show more

slrl0ver, thanks for your enlightenment. As I have wrote earlier, I wish Nikon could have given me the option of 16-bit on D800. However, I do completely understand those that just need to do 8-bit capture.



Apr 10, 2012 at 03:41 PM
taob
Offline
• • •
Upload & Sell: Off
p.2 #17 · p.2 #17 · D800: 12-bit RAW vs. 14-bit RAW


mshi wrote:
If you take a serious look at 16-bit raw capture, 14-bit seems like a child's play IMHO
16-bit raw SOOC (H4D-40, 40MP)

I'm not sure what you are attempting to show, but it has nothing to do with 12-bit vs 14-bit vs 16-bit. I am guessing the image you show has not suffered through extreme changes of exposure, contrast or saturation. You are also showing a JPEG, which uses 8-bit channels. Even after gamma correction, there will be no difference between 12- and 14-bit capture, much less 16-bit capture.

Your example, along with the 100% crop, is instead good to show other advantages of MFDB, namely high resolution coupled with a large sensor, and the lack of an AA filter (rendering that extra crisp detail in the crop). 16-bit capture offers no advantage in that respect.



Apr 10, 2012 at 06:27 PM
mshi
Offline
• • • •
[X]
p.2 #18 · p.2 #18 · D800: 12-bit RAW vs. 14-bit RAW


14-bit offers 16,384 different possible steps while 16-bit does 65,536 possible steps at sensor capture level. In other words, 16-bit sensor has four times the fidelity (or precision) of the 14-bit sensor. 65,536 tonal values offer much smoother gradient transition and shadows. Whether or not I can tell the difference all the time, I don't know. But that reminds me what Leonardo Da Vinci once famously said, " There are three classes of people: Those who see. Those who see when they are shown. Those who do not see."


Apr 10, 2012 at 07:07 PM
jmcfadden
Offline
• • • • • • •
Upload & Sell: Off
p.2 #19 · p.2 #19 · D800: 12-bit RAW vs. 14-bit RAW


Jan Brittenson wrote:
If you look at the histogram of a linear (gamma 1.0) image you'll notice that the highest 1 stop takes up the entire right half of the histogram. The top 3 stops take up 7/8ths of it. The remaining 11 stops, assuming 14 stops of DR are squished into the leftmost 1/8th. In a 16-bit file the lowest stop can have a value of 1 and nothing else. (0 being everything too dark to register and not on the scale.) The next stop can have the value 2. Then 3 or 4. So the 3 lowest stops have exactly
...Show more


Since you've been away Jan you have become so much "more smarter" , lol

so good to have you back my old friend


J



Apr 10, 2012 at 07:53 PM
wfektar
Offline
• •
Upload & Sell: Off
p.2 #20 · p.2 #20 · D800: 12-bit RAW vs. 14-bit RAW


mshi wrote:
14-bit offers 16,384 different possible steps while 16-bit does 65,536 possible steps at sensor capture level. In other words, 16-bit sensor has four times the fidelity (or precision) of the 14-bit sensor. 65,536 tonal values offer much smoother gradient transition and shadows. Whether or not I can tell the difference all the time, I don't know. But that reminds me what Leonardo Da Vinci once famously said, " There are three classes of people: Those who see. Those who see when they are shown. Those who do not see."


The issue, as ever, is signal to noise, which is what the OP is getting at. At some point all you are getting is finer and finer gradations of noise. Perhaps the D800 has a low enough noise level and a high DR that if you have a contrasty situation using the entire DR and need to lift the shadows, 14 bit will have a perceptible advantage over 12 bit -- but that remains to be demonstrated. (Any D800 owners care to try?). Otherwise if the SNR is too low the image (or relevant part of the image) will be dominated by noise and the greater bit depth is resolving noise.

It also needs to be demonstrated that 16 bit on a D800 will be noticeably different from 14, or even 12. Using a high end MFDB for comparison isn't useful as the MFDB is using something like 2.5x as much information to describe a scene as the D800 (assuming the MFDB is near FF 645).



Apr 10, 2012 at 08:24 PM
1      
2
       3       4       5       end




FM Forums | Nikon Forum | Join Upload & Sell

1      
2
       3       4       5       end
    
 

You are not logged in. Login or Register

Username       Or Reset password



This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.