Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: 16-bit export  (Read 2420 times)
boku
Sr. Member
****
Offline Offline

Posts: 1493



WWW
« on: January 04, 2004, 02:02:38 PM »
ReplyReply

The Canon 300D and 10D produce a raw file, 11-bits, that can be converted in editing software to a 16-bit TIFF, but you will still only have the initial accuracy of 11 bits. Mathematical manipulations after conversion will take place at 16 bits.

Most, if not all, digital sensors render 11 or 12 bits. If they save as RAW you can do what I just described. If they save as a 16-bit TIFF, they are doing the conversion in camera, but not giving you more than 11 or 12 bit of accuracy. If they only save as JPG, they are downsampling to 8-bits and you lose accuracy.
Logged

Bob Kulon

Oh, one more thing...
Play it Straight and Play it True, my Brother.
BJL
Sr. Member
****
Offline Offline

Posts: 5171


« Reply #1 on: January 08, 2004, 10:54:30 AM »
ReplyReply

Please correct me if I am wrong, but I believe that TIFF and JPEG output, direct from cameras or from RAW converters, has had a tone curve and gamma compression applied; the numerical digital levels are no longer in direct proportion to recorded luminosity unless one does something very unusual. With even the smallest common gamma of 1.8 (the pre-press and Macintosh standard), up to fourteen significant bits of sensor dynamic range is compressed into the range covered by 8-bit formats (roughly, bits times gamma = dynamic range in stops).

  If so, with 8-bit formats one loses fineness of tonal graduations within the overall range, but not dynamic range. For example, if, optimistically, the tone curve plus gamma choice divide the 256 available 8-bit levels more or less uniformly between each stop of an 11 stop range, one gets levels spaced at a bit less than 1/23 of a stop.
  How visible are such increments of luminosity? I do not know, but have never heard of anyone working with smaller than 1/12 stop adjustments.
Logged
MatthewCromer
Sr. Member
****
Offline Offline

Posts: 411


« Reply #2 on: January 09, 2004, 09:47:29 PM »
ReplyReply

There is a third-party RAW conversion now for the 828.

It's command-line based, but open source, and I suspect lots of folks will be offering SRF support shortly.

It does support 16-bit per channel linear and non-linear TIFFs as well as 8 bit per channel TIFFs.

http://www.cybercom.net/~dcoffin/dcraw/
Logged
peter
Newbie
*
Offline Offline

Posts: 4


WWW
« Reply #3 on: January 04, 2004, 01:11:42 PM »
ReplyReply

Howdy,
I was distressed to read the Sony DSC-F828 review and realize that it does not export 16-bit tiff. This is beyond stupid (I just emailed Sony). Anyways, does anybody have some comments on cameras that support 16-bit tiff export (uncompressed) for around $1000. My reseach has shown that the Canon EOS 300d is probably a good way to go.

Cheers, Peter
Logged
Digi-T
Full Member
***
Offline Offline

Posts: 149


« Reply #4 on: January 04, 2004, 02:53:18 PM »
ReplyReply

Regarding the F828 I believe it is their RAW conversion software that is limiting the files from being saved at 16 bits not the camera itself. This is really stupid but at least Sony can probably fix this with an upgrade if enough people complain about it. They also need to allow the RAW files to be converted with other third-party programs. Sony really needs to "wake up" regarding this proprietary stuff.

T
Logged
dbarthel
Sr. Member
****
Offline Offline

Posts: 281


« Reply #5 on: January 08, 2004, 11:09:16 AM »
ReplyReply

The two most significant advantages of RAW are:
1) Ability of the converter to recover shadow/highlight detail
2) Extra precision when converting to 16 bit tiff, which means smoother histograms when doing image adjustments.

Quality 3rd party converters like Capture One & PSCS do an amzing job of adjusting exposure, color balance, and contrast during the conversion.  If saving JPEG or TIFF, you're pretty much stuck with what the camera decides is good for you in terms of recovering shadow and highlight detail. Color adjustment and sharpness can certainly be performed on the JPEG or TIFF in PS
Logged
peter
Newbie
*
Offline Offline

Posts: 4


WWW
« Reply #6 on: January 15, 2004, 02:21:46 AM »
ReplyReply

Matthew, thanks for the link for the open source raw conversion software. PERFECT!

  http://www.cybercom.net/~dcoffin/dcraw/

Being a programmer I am very interested in this code. Looks really good.

It would have been a shame to be stuck with 8bit per channel imagery. For anybody that is interested, tonight I wrote some code to count the number of unique DNs (digital numbers) in a scanned medium format black and white negative (scanned with a epson 2450 - awesome scanner!). Almost all of my negatives (16bit tiffs) have between 30,000 and 32,000 DNs. Pretty good for a $400 scanner! I found that what I consider my critical highlight detail had about 6000 levels and my critical shadow detail had about 8000 levels. These number of levels give one a lot to play with, especially if you create separate layers for shadow areas and highlight areas for problem negatives. For those perfectly exposed negatives 256 levels of grey is more than enough.

So, I'm happy that it looks like I will be able to convert the 11bit (10bit?) F828 raw format to 16bit tiffs.
Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad