Missing Bits?

[superhornet59]

Hey everyone, I know the D80 is supposed to capture 12 bits of data per channel, but to my dismay I looked through the exif data of one of my RAW photos on irfanview, and it's saying the *original* file (untouched RAW, not a post-conversion jpeg or anything) has a bit depth of 24 bits per pixel, which obviously means (and also explicitly is shown in another part of the exif data list) 8 bits per channel.

Am I missing something here? I should be getting 12 bits per channel and a total of 36 bits per pixel. I don't know of any in-camera settings that can lower that. Where are those missing 4-bits per channel? That's alot of information that's getting thrown away!

 

[KmH]

What is the Document Type, the File Size, and the Color Mode?

 

[superhornet59]

Document Type: NEF (RAW)
File Size: 7.75 MB
Color Mode: Tried Ia, II, and IIIa along with the various camera presets, all with the same result.

Is this an irfanview thing, or a problem with my camera? Can anyone test the same with their own RAWs? All the photos I have tested show 24 bit (8 bits per channel) when they should be 36 (12 per channel) according to Nikon...

what the heck is going on here?

 

[KmH]

Might you have NEF + JPEG selected?

It seems to me a .nef file should be more than 7.75MB, but image content effects file size too.

I have never used Irfanview and can't help you out there.

I would suggest trying a different application than Irfanview as a comparison.

 

[superhornet59]

Nope, RAW alone.

Just popped one into Adobe Camera RAW 6.0 and it's saying the same thing: 8-bit .... (and curiously enough, 10.0 MP when I know my camera is 10.2).

I'm getting pretty freaked out here.. can anyone with a D80 try the same test in either program?

By the way, when in the 'Image Quality' section of the shooting menu, and I press the '?' button, it describes the different setting.. and it specifically says the RAW is 12-bit.

This is really weird. I was hoping their would be a quick answer, but I feel I may need to contact Nikon soon.

 

[Garbz]

Who cares about ExIF data, it's not guaranteed to be right or consistent between camera makes and models. The important bit is that your NEF file comes up as a D80 RAW. RAW is the true sensor data, if your D80 files are readable then they are 12bit files. Otherwise the software wouldn't know how to decode the data.

Also careful with RAW compression. The NEF compression mathematically sacrifices 2bits of data without any notice.

 

[superhornet59]

Well that's what's so puzzling about all this.. that it's a RAW. Yes it does read D80 in the exif. What do I make of the 'kind of small for a raw' comment earlier? The RAWs in my D80 are the only ones I've ever worked with, so I have nothing else to compare them too. I see Nikon states the RAW's are saved in a lossless compression format. Doesn't sound so lossless when I got 4 bits per channel missing... this was my first DSRL and I've been very happy with it, and have nothing else at the moment to compare it to, and would not notice all the missing data (which reduces dynamic range and gradient tone smoothness.) Nevertheless if my camera still has more capability in it, I wouldn't mind making the most of it. I'd check this in NX but my trial expired a while ago and I have no plans to purchase.

 

[KmH]

The point is, you're not likely to actually be missing any image data.

Have you tried a differnt EXIF reader yet? There's only about 300 free ones online.

How about copying and posting the EXIF data? :thumbup:

 

[Garbz]

Kind of small for a RAW is easily explained. The Nikon D80 has RAW compression that is always on with no option for turning it off. The same compression is employed in the D200 and I will also generate pictures between 7-12MB. Whereas actual RAW uncompressed for a 10.2mpx camera is 15.3MB

You are in essence missing a few bits, but by the way the compression works the bits are sacrificed only in the highlights. Clever form of lossless compression given that the purpose of those bits are usually to brighten an image, and therefore dark pixels will retain the full sensor data exactly to aid in the post production process. I don't know where it is off hand but I remember there was a blog somewhere written by the guy who reverse engineered NEF compression for the D200 which had all the juicy maths in it.

 

[superhornet59]

Filename - DSC_0742.NEF
ImageWidth - 160
ImageLength - 120
BitsPerSample - 8 8 8
Compression - 1 (None)
PhotometricInterpretation - 2
Make - NIKON CORPORATION
Model - NIKON D80
StripOffset - 128846
Orientation - Right top
SamplesPerPixel - 3
RowsPerStrip - 120
StripByteCount - 57600
XResolution - 300
YResolution - 300
PlanarConfiguration - 1
ResolutionUnit - Inch
Software - Ver.1.01
DateTime - 2010:07:23 12:40:02
ReferenceBlackWhite - 0
ExifOffset - 468
DateTimeOriginal - 2010:07:23 12:40:02
ExposureTime - 1/80 seconds
FNumber - 2.00
ExposureProgram - Aperture priority
ISOSpeedRatings - 125
DateTimeOriginal - 2010:07:23 12:40:02
DateTimeDigitized - 2010:07:23 12:40:02
ExposureBiasValue - 0.00
MaxApertureValue - F 1.74
MeteringMode - Multi-segment
LightSource - Auto
Flash - Not fired
FocalLength - 50.00 mm
UserComment - (C)Matt ------------
SubsecTime - 50
SubsecTimeOriginal - 50
SubsecTimeDigitized - 50
SensingMethod - One-chip color area sensor
FileSource - Other
SceneType - Other
CustomRendered - Normal process
ExposureMode - Auto
White Balance - Auto
DigitalZoomRatio - 1 x
FocalLengthIn35mmFilm - 75 mm
SceneCaptureType - Standard
GainControl - None
Contrast - Soft
Saturation - High
Sharpness - Hard
SubjectDistanceRange - Unknown

Maker Note (Vendor): -
Data version - 0210 (808595760)
ISO Setting - 125
Color Mode - COLOR
Image Quality - RAW
White Balance - AUTO
Image Sharpening - MED.H
Focus Mode - MANUAL
Flash Setting - NORMAL
Flash Mode -
White Balance Adjustment - 0
White Balance RB - 746
Exposure Adjustment - 68608
Thumbnail IFD offset - 12434
Flash Compensation - -301922816
ISO 2 - 125
Tone Compensation - MED.L
Lens type - AF
Lens - 830
Flash Used - Not fired
AF Focus Position - Center
Bracketing - 131072
Contrast Curve - I #(,159=BGLPUY^bglquz„ˆ’—› ¥ª¯´¹¾ÃÈÍÒ×Üáæëðõúÿ






#
(
-
2
8
=
B
G
M
R
W
\
b
g
l
q
w
|

†
Œ
‘
—
œ
¡
¦
¬
±
·
¼
Á
Æ
Ì
Ñ
×
Ü
â

Color Mode - MODE3a
Light Type - NATURAL
Hue Adjustment - 0
Noise Reduction - OFF
Total pictures - 24228
Optimization - CUSTOM


That is rather clever indeed Garbz, but unless the highlights are fully blow on out I'd be pretty conservative on compression. bits are an exponential unit of measurement (2^#bits). Having 8 bits of data instead of 12 means only 1/16th of the tonal detail (colors and luminosity levels) is recorded. That is ALOT.

I understand why this compression exists as NEF's were big and memory cards small, but with today's cheap monster capacity cards, you would think Nikon would take a hint and remove/create an option for this in new firmware upgrades. 'Lossless' as it may be, there must be a reason why the D200 and D2x have a toggle setting for it.

 

[Garbz]

Yep and it's a lot of absolutely useless information that isn't relevant since most post processing relates to boosting the dark information which essentially retains all the data anyway. Remember it's not just the storage that matters but also the interpretation. Sensor data is linear, pictures and human vision follow a gamma curve. The image you see has far more detail in the highlights than it does in the shadows. In the name of compression knocking off a few of the least significant bits in the highlights is more than perfectly acceptable for most demanding photography.

Also you have a 4 year old camera. Memory has only started getting cheap in the past year or so. If this is such a concern for you then it's time to upgrade your camera, but my suggestion is look less at the numbers, realise that you have a fantastic camera in your hands, and focus on taking better pictures.

By the way my NEFs also say 8bit even though each file is exactly 15.8MB

 

[superhornet59]

Okay, then at least I'm not alone. I was hoping it was just the way the editor interprets the file. Tone mapping my NEF's and my Jpegs are two completely different scenarios, with the Jpeg's having significantly less shadow detail and significantly more posterization. I figured the NEF was keeping more data, but just needed a little reassurance.

Don't get me wrong, I LOVE my camera (although.. i've seen some of these HD videos the D90 can put out.. and ahhhh it's so hard to resist), I just wanted to make sure I was squeezing every bit of 'pro' out of it, starting with max IQ. I've taken some beautiful pictures in jpeg that I was unable to sell later because it wasn't a HQ TIF or NEF... IQ is as important to me now as the pictures themselves.

Anyway, thanks!