Questions on RAW format

[Breathe]

As I understand, RAW format is the digital negative, and thus is the original unaltered a.k.a what the sensor sees. I have a Nikon D50, thus the NEF format is relevant to my question.

The question is does setting camera image options affect the RAW image? By image options, I mean things like:
1. WB (i found that WB does affect the RAW file?)
2. image sharpening (does this lead to loss of info in RAW file?)
3. colorspace e.g. SRGB-1 SRGB3a etc.


Thanks in advance.
New to DSLR.

 

[yyD70S]

As I understand, RAW format is the digital negative, and thus is the original unaltered a.k.a what the sensor sees. I have a Nikon D50, thus the NEF format is relevant to my question.

The question is does setting camera image options affect the RAW image? By image options, I mean things like:
1. WB (i found that WB does affect the RAW file?)
2. image sharpening (does this lead to loss of info in RAW file?)
3. colorspace e.g. SRGB-1 SRGB3a etc.


Thanks in advance.
New to DSLR.

The NEF file contains the original image information as it comes off the sensor before in-camera processing so you can do that processing afterwards on your PC with NC, NV, PS, etc.

Note: original image information. IE, yes. The information is attached to the file but can be changed by the appropriate software.

(1) The WB is an option in NC and can be changed at will.
(2) Not that I know of. Experts please confirm this as a yes or no.
(3) Likewise, Color Space is attached and can be changed via the software.

To summarise, many of the camera settings (sharpening, WB, tone settings, etc)which were applied to the raw data can be undone when using the RAW processing software (NC for e.g.).

Just to note that the NEF from D50 is a form of compressed RAW.

 

[solarii]

1. No. WB info (be it metered or user defined) is stored seperately from image data. It's only fixed once to "develop" it into jpeg to tiff

2. No. Sharpening is only applied during the conversion, or even later in a photo editing program if you so wish. Any sharpening setting in the camera menu applies when you shoot in jpeg, not in raw.

3. Nope. You select the colour space during the conversion.

 

[Breathe]

Thanks for all the clarifications. One more question though. For point 2 and 3 (i.e. sharpening and color space), is the information attached to the NEF file? Or does selecting the sharpening have no effect at all on the NEF, i.e. these functions are bypassed when shooting in RAW?

Thanks for the help !

1. No. WB info (be it metered or user defined) is stored seperately from image data. It's only fixed once to "develop" it into jpeg to tiff

2. No. Sharpening is only applied during the conversion, or even later in a photo editing program if you so wish. Any sharpening setting in the camera menu applies when you shoot in jpeg, not in raw.

3. Nope. You select the colour space during the conversion.

 

[espn]

Thanks for all the clarifications. One more question though. For point 2 and 3 (i.e. sharpening and color space), is the information attached to the NEF file? Or does selecting the sharpening have no effect at all on the NEF, i.e. these functions are bypassed when shooting in RAW?

Thanks for the help !

Yes, the information is attached in the file seperately. Changes are not permanent.

However, shooting in aRGB can switch to sRGB, but not vice versa.

Sharpening in camera as long as shot in RAW is totally reversible by Nikon Capture when edited.

 

[UandMe]

Just to note that the NEF from D50 is a form of compressed RAW.

no wonder the D50 raw file's size is not much bigger than its JPEG's...

 

[yyD70S]

The RAW file is still about double that of the highest quality JPEG (12bit ag 8bit) but that's not the point. Unless storage is a concern.

Always shoot in RAW. Especially if you are undertaking very important projects.

 

[dundee]

The NEF file contains the original image information as it comes off the sensor before in-camera processing so you can do that processing afterwards on your PC with NC, NV, PS, etc.

Note: original image information. IE, yes. The information is attached to the file but can be changed by the appropriate software.

(1) The WB is an option in NC and can be changed at will.
(2) Not that I know of. Experts please confirm this as a yes or no.
(3) Likewise, Color Space is attached and can be changed via the software.

To summarise, many of the camera settings (sharpening, WB, tone settings, etc)which were applied to the raw data can be undone when using the RAW processing software (NC for e.g.).

Just to note that the NEF from D50 is a form of compressed RAW.

The sharpening is determined by the raw processor processing the file, not the camera.;p

 

[solarii]

Yes, the information is attached in the file seperately. Changes are not permanent.

However, shooting in aRGB can switch to sRGB, but not vice versa.

Sharpening in camera as long as shot in RAW is totally reversible by Nikon Capture when edited.

I'll have to disagree with you on this. When shooting in RAW what is recorded is the luminance values of the individual pixels, not absolute colour values.

The colour space (adobe or sRGB) is assigned during the conversion process. The RAW converter will assign colour values based on the characteristics of the sensor. So its possible to switch btw sRGB and Adobe RGB, by changing your output colour space in the RAW converter. Your camera's colour space setting has no effect. By the way Adobe RGB isn't the largest colour space availabe... there's also ProPhoto RGB which also can be selected to give an even larger gamut.

But bear in mind that after the conversion process, should you change to a smaller colour space, the additonal colours in the larger gamut will be lost forever.

 

[unseen]

As I understand, RAW format is the digital negative, and thus is the original unaltered a.k.a what the sensor sees. I have a Nikon D50, thus the NEF format is relevant to my question.

Not true. Nikon's compressed NEF format is altered information. It employs lossy compression. It means that some details are lossed. Basically it's just a higher quality JPEG file, that's why size isn't much different.

http://www.nikonians.org/html/resources/nikon_articles/body/d50/d50_2.html

 

[sriram]

Some nikons like the D100 offer compressed and uncompressed NEF's. If I remember correct, the uncompressed nef's were about 10mb and the lossy compressed ones were 5-6mb each. Not sure about the others, since I don't shoot nikon.

 

[lsisaxon]

Not true. Nikon's compressed NEF format is altered information. It employs lossy compression. It means that some details are lossed. Basically it's just a higher quality JPEG file, that's why size isn't much different.

http://www.nikonians.org/html/resources/nikon_articles/body/d50/d50_2.html

The compression which NEF uses and JPEG compression are two different type of compression and should not be confused. JPG uses some mathematical equations to approximate an area but the compression employed in NEF is actually dynamic compression by taking away insignificant bits when representing details at different levels but is still cannot be discerned by the eye.. It's a 12-bit linear to about 9-bit logarithmic dynamic compression which is still better than 8-bit TIFF's 8-bit linear representation or JPEG, which uses 8-bit linear for dynamic in addition to approximation in the spatial domain.
http://www.majid.info/mylos/weblog/2004/05/02-1.html

So you will not lose spatial quality for NEF but you will just lose indiscernible dynamic details since it is still equivalent to 9.4 bits.

 

[unseen]

yeah i know.. I'm just drawing a parallel. Given a high quality jpeg, you lose indiscernible quality too. Most of the time, I doubt anyone can visually tell the difference between a "100%" quality and a "90%" quality jpeg.

If you look at the codes, you'll see that the whites are being discarded. Means it's better to shoot under exposed because slight over, and you'll lose details. It's not really indiscernible dynamic details. it's your highlights. You'll lose your highlights just as long as they're touching the tip of the histogram.
I suppose that's why Nikon intentionally underexposes the images, so that you lose indiscernible details. If you do a proper exposure, you're going to lose discernible details.

Data loss is still data loss.

D200 offers compressed and uncompressed NEF files too..
it's the budget dSLRs like ala D70/s/D50 that doesn't.

 

[lsisaxon]

yeah i know.. I'm just drawing a parallel. Given a high quality jpeg, you lose indiscernible quality too. Most of the time, I doubt anyone can visually tell the difference between a "100%" quality and a "90%" quality jpeg.

If you look at the codes, you'll see that the whites are being discarded. Means it's better to shoot under exposed because slight over, and you'll lose details. It's not really indiscernible dynamic details. it's your highlights. You'll lose your highlights just as long as they're touching the tip of the histogram.
I suppose that's why Nikon intentionally underexposes the images, so that you lose indiscernible details. If you do a proper exposure, you're going to lose discernible details.

Data loss is still data loss.

D200 offers compressed and uncompressed NEF files too..
it's the budget dSLRs like ala D70/s/D50 that doesn't.

No it's not true.. The way we see light is pretty much the same as we hear sound. If we talk about energy, every stop is a doubling of the energy. For digital representation, it is also convenient because an additional bit can represent double. Since the CCD values records the energy of the light linearly, when you have a bright light source, what difference would a tiny variation make? Would it be discernible?

Mathematically, let's say we need to represent the value 2048. which requires the 12th bit. If you drop 3 bits, to become an 9-bit representation, instead of having the step size to be 1, you now have the step size to be 4.

How much of a difference can you perceive when you have a change of 8/2048 (0.004) compared to 1/2048 (0.0005)? It may look significant like that but if you see 1.004 compared to 1.0005 or 0.996 compared to 0.9995, then you will realise that it is not that significant actually. Sure there is a difference but it's not that significant unless you want to expand the histogram for that range only.

But for the midtones and the low light, these less significant bits are still important, so they are not discarded to still be able to give midtone and lowlight variations up to the 12th bit accuracy. But the problem is that there are too many leading zeros in front, so essentially you can encode with less bits for midtone and shadows because there is no information in the bits used to represent highlights.

Also if you think of the number of bits as comparable to the number of stops, 9 bits can encode 9 stops of contrast. That means your brightest and darkest points can be 9 stops of differences, that's already more than film. Give 1 or 2 bits for noise, you still have about 7 bits of information and that's a contrast of 7 stops. Almost on par with film.