Raw 14 Bit

[cupic]

Had a look at the Lossless,Compressed and UnCompressed files sizes
D300s in order 30,45,19
D700 in order 20,23,17
D750 in order 15,21,?

Can anyone confirm that the D750 has or hasn't a 14 Bit Uncompressed file size ?
That or a typo exists



cheers

[arthurking83]

Hi Cupic,

Short answer at the very end of the reply:

Long answer as follows:

I'm a bit unsure of your question, or more specifically the way you've asked it.

You have listed the types of compression in one order, but the file sizes listed for each camera don't appear to line up with the way you have listed them in order of quality.

That is, (on my D300m which is the same as your D300s).

Uncompressed: would be the highest file size value. This is pure 14bit quality with no compression ..
Lossless compressed: this should be the next largest file size, as even tho it has some compression, it's supposed to minimal so as to not lose any quality(bit depth).
Compressed: this should be the smallest file size as it will have lost some info, even tho the lost bit depth info may not be something the human eye can detect.

for the order you have listed "Lossless,Compressed and UnCompressed files sizes"

Your D300 file size order would(should) then be: 30,19,45

(note: your D300s options should be: Lossless compressed, Compressed, Uncompressed .. but Lossless is a good shortcut name for Lossless Compressed)

I just had a look at the D750 manual, and indeed your obsveration is spot on. It doesn't seem to allow a fully uncompressed raw file save!
Well, but the manual, it doesn't list Uncompressed as an option.
The two options are Lossless compressed and Compressed only.

The main reason I actually replied tho is to point out that, if you use Nikon's software to edit/view/save the raw files, the lossless compressed option may actually be exactly as Nikon claim it to be .. lossless(in the true sense of the word .. and not just the marketing speak version where lossless can mean that 'the human eye can't spot the loss of any data'.

If you shoot a raw file with your D300s in lossless compressed mode, and open the file with CaptureNX2 I'm pretty sure(ie. 99.9%) the file is probably uncompressed to a size that is closer to the uncompressed file type option the camera gives you.
That is, Nikon's software must use the compression algorithm that the camera used to uncompress the file now that it's on the computer.
Think of it as a zip file.

I've done this before to see what differences/effect all these settings had on my D300 files wayyyy back when .....
(and because I forgot all the values, I did it again to refresh my memory)

So on the D300, for the exact same scene(which is vital to see what the actual differences are)

in camera:
1/. uncompressed image gave me a 24Mb raw file. The scene is partially colour complex, partially just plain white. A well exposed, full colour image could be closer to 30Mb(or more).
2/. lossless compressed image gave me a 13Mb file of the same scene. So it lost 9Mb in a lossless compression routine. Nikon's zipping seems to be effective.

So I've opened the two images in CaptureNX2 and resaved them in two ways:

1/. saved the uncompressed image with CNX2's option of lossless compressed. Resulting file size is 18Mb
2/ saved the camera's lossless compressed file as another NEF(copy) and CNX2 saved it to an 18Mb file size.

Note that the only 'issue' here is that if the file has been captured with any compression, when saving that compressed raw file in CNX2, the option to alter the compression levels is greyed out(ie. not an option!)

But! CNX2 has inflated the once 13Mb lossless compressed file, by 5Mb to 18Mb now. That 5Mb of extra data is more than 50% of the difference (9Mb) between the compressed and uncompressed files in camera.
So, in a way, Nikon seem to be true to their word in lossless .. but I've never seen any way to fully regain the potential to recover all the lost data when compressing the raw file.
(it probably makes no difference tho).

In summary: No, D750 doesn't seem to allow uncompressed raw files, but it probably makes no difference.
Caveat: I have no idea what this means if you use non Nikon raw software tho. Also another anomaly is that Nikon's latest software(CNX-D) doesn't allow the ability to save a raw file as another copy of the raw file.
This is a severe limitation, and one that is infuriating! ViewNX2 also doesn't allow to save the NEF file as a copy without making some adjustments(which then nulls the ability to directly compare).
Of course CNX2 doesn't recognise D750 raw file tho.
(new summary: Nikon's software system is "up the proverbial effluent waste"

[shaneando]

The D750 does NOT offer uncompressed file options. I think only the D800 series and D4 series currently offer that option.

Having said that, I've read a few very reputable reports saying that Nikon lossless compressed truly is lossless, not just visually lossless, so there is no downside to using that format.

My very amateur understanding is this - let's say an image has 500 pixels all with the same colour and tone data. Rather than record 12 or 14 bits of data 500 times, it records the pixel bit information once, then uses a reference table to tell the decompression software where those 500 pixels lie in the image. So no loss of image data, but less storage data required.

[ameerat42]

My 2 cents of this...

1. No. There is no uncompressed raw recorded.

2. I cannot tell if the compressed data is lossless.

3. This is a Q to Shane: Do you mean that the software interpolates data that the camera has recorded?

4. If "yes", then if not technically "lossless", it is not real data that is being processed.

4. Back to Cupic. Hmm! I'd be thinking about this too!!!

However, it all hinges on what sort of data the program is processing.
Am.

PS: I was teasing and gleaning stuff out of here.
(Not that it was terribly clear)

[shaneando]

2. I cannot tell if the compressed data is lossless.

3. This is a Q to Shane: Do you mean that the software interpolates data that the camera has recorded?

4. If "yes", then if not technically "lossless", it is not real data that is being processed.

No, it simply stores the full original data in a more efficient way. This is a really crude example but I think it shows the principle...

Let's say you had a very small image of 25 pixels, 5 x 5. Of those 25 pixels, 5 have the same colour and tone information i.e. at a bit level (1's and 0's) they are identical. Rather than store this same piece of data five times, the lossless compression algorithm stores the bit data once, then stores some extra information about the location of those pixels i.e. a simple x-y reference. This takes much less space, but the data can be fully restored to its original version with no loss of information, hence "lossless". In this example instead of 25 bytes of data, you might have 22, so you've reduced the file size by 12%.

There is a very good thread on DPReview about it - Nikon uncompressed raw.

There is also another thread suggesting that at higher ISO's 14 bit raw is a waste of space because there isn't enough dynamic range to take advantage of the extra bit depth

[ameerat42]

Thanks for the clarification.
Am.