FAQ FAQ  Forum Search   Events   Register Register  Login Login

Uncompressed RAW of A700, 850, 900

Author
Andrew R. View Drop Down
Newbie
Newbie

Joined: 02 April 2015
Country: Russia
Status: Offline
Posts: 34
Post Options Post Options   Quote Andrew R. Quote  Post ReplyReply Direct Link To This Post Topic: Uncompressed RAW of A700, 850, 900
    Posted: 14 July 2019 at 07:57
Sometimes it is better not to know how the food was cooked in a restaurant. The same can be true for RAW "cooking".

Recently, I broke this rule by studying how the Image Data Converter reads uncompressed RAW files produced by α700, 850 and 900.

The reason was some unclearity about these cameras. While the compressed RAW file contains 14-bit values, the uncompressed one only 12-bit ones. In several descriptions you will find information that, in general, these cameras support only the 12-bit mode.

And, finally, I am a proud owner of an α700, my newest and best DSLR, which is in intensive service for more than 10 years. (Now, you know why the photo industry is in a crisis...)

The result was a surprising one. After a 12-bit value is read it is shifted to the left at 2 positions, or, what is the same, it is multiplied by four.

                        (hex : bi : dec)
Saved value:            0x00000145 : 0000 0001 0100 0101 : 325
Restored value:          0x00000514 : 0000 0101 0001 0100 : 1300

In my opinion, this means two news, a good one, and a bad one.

The good one: α700, 850, 900 seem to be, at least internally, 14-bit cameras.
The bad one: They do not provide any lossless RAW format. The "uncompressed" 12-bit format is, in fact, a compressed 14-bit one that uses the most simple compression method by cutting off two lowest bits.

I suppose that this "optimization" is used only for saving RAW files. If it is true, then JPEG files are prepared by using full 14-bit values. This means that JPEG files could have some advantages over RAW files in spite to the fact that JPEG data contain only 8-bit values.

This is a question to DSP specialists, what is better: to process RAW data of that kind as "honest" 12-bit values, or, as "would-be" 14-bit values with two lowest bits set to zero.

Edit: the title field does not support Greek letters.

Edited by Andrew R. - 14 July 2019 at 08:02
 



Back to Top
jvandegr View Drop Down
Senior Member
Senior Member

Joined: 07 April 2008
Country: United States
Location: New York
Status: Offline
Posts: 966
Post Options Post Options   Quote jvandegr Quote  Post ReplyReply Direct Link To This Post Posted: 14 July 2019 at 15:06
There is only one test to answer your question: print both of them and see for yourself. If you don't intend to print your photographs, it most likely doesn't matter as long as you're shooting RAW and developing with a quality app such as Capture One Pro, DXO OpticsPro, or Apple Photos.
Jon Van de Grift
Geographer and Documentary Photographer
Back to Top
beautiophile View Drop Down
Senior Member
Senior Member

Joined: 11 December 2014
Country: Vietnam
Location: HCM City
Status: Offline
Posts: 211
Post Options Post Options   Quote beautiophile Quote  Post ReplyReply Direct Link To This Post Posted: 15 July 2019 at 03:46
Originally posted by Andrew R. Andrew R. wrote:


The good one: α700, 850, 900 seem to be, at least internally, 14-bit cameras.
The bad one: They do not provide any lossless RAW format.

AFAIK, those three cameras have 12-bit ADCs on their sensors.

The "uncompressed" 12-bit format is, in fact, a compressed 14-bit one that uses the most simple compression method by cutting off two lowest bits.

I think it's not called "compressed 14-bit". It's just a 14-bit file format but contains only 12-bit data from a camera, like a 100m^3 (metre cube) truck loading 50m^3 of wood.

This is a question to DSP specialists, what is better: to process RAW data of that kind as "honest" 12-bit values, or, as "would-be" 14-bit values with two lowest bits set to zero.

cRAW is a lossy compression. IME, cRAW and RAW make almost the same 8-bit JPEGs after conversion.
D7D, M7D ~ 16-80ZA
Alpha 7, Alpha 807si, Dynax 3, A99 ~ Min 17-35G, 24-85, 28f2, 100Macro; SAL35F14G, SAL50F14. SAL70300G
F20AM, F42AM, F58AM.
Back to Top
Andrew R. View Drop Down
Newbie
Newbie

Joined: 02 April 2015
Country: Russia
Status: Offline
Posts: 34
Post Options Post Options   Quote Andrew R. Quote  Post ReplyReply Direct Link To This Post Posted: 15 July 2019 at 07:47
Originally posted by Andrew R. Andrew R. wrote:

The good one: α700, 850, 900 seem to be, at least internally, 14-bit cameras.
The bad one: They do not provide any lossless RAW format.
Originally posted by beautiophile beautiophile wrote:

AFAIK, those three cameras have 12-bit ADCs on their sensors.

Do we know this for sure?

OK, it seems that α700 uses this 12-bit sensor:

https://www.sony.net/SonyInfo/News/Press/200708/07-072E/index.html

However, Nikon D300 uses a similar but not identical sensor.

Look here:

https://www.dxomark.com/Cameras/Compare/Side-by-side/Sony-Alpha-700-versus-Nikon-D300___562_440

Due to this source α700 is a 12-bit camera while Nikon D300 is a 14-bit camera. However, the sensor performance is almost identical. I would expect more difference between a 12-bit sensor and a 14-bit one.

Well, after some searching I have found a post where this question was discussed (and partially answered):

http://www.photoclubalpha.com/2007/11/08/nikon-d300-and-sony-a700-sensor-similarity/

Originally posted by Andrew R. Andrew R. wrote:

The "uncompressed" 12-bit format is, in fact, a compressed 14-bit one that uses the most simple compression method by cutting off two lowest bits.
Originally posted by beautiophile beautiophile wrote:

I think it's not called "compressed 14-bit". It's just a 14-bit file format but contains only 12-bit data from a camera, like a 100m^3 (metre cube) truck loading 50m^3 of wood.

Has it any advantages (for processing??) to blow up data?

Originally posted by Andrew R. Andrew R. wrote:

This is a question to DSP specialists, what is better: to process RAW data of that kind as "honest" 12-bit values, or, as "would-be" 14-bit values with two lowest bits set to zero.
Originally posted by beautiophile beautiophile wrote:

cRAW is a lossy compression. IME, cRAW and RAW make almost the same 8-bit JPEGs after conversion.

This is probably true. Nonetheless, I have been rather interested in technological or mathematical backgrounds.


Edited by Andrew R. - 15 July 2019 at 07:51
Back to Top
Dyxum main page >  Forum Home > Equipment forums > Digital Darkroom

Forum Jump Forum Permissions View Drop Down



This page was generated in 0.063 seconds.

Monitor calibration strip

Dyxum.com - Home of the alpha system photographer

In memory of Cameron Hill - brettania

Feel free to contact us if needed.