Leica Compressed or Uncompressed DNGs

flash

Veteran
Name
Gordon
What's everyone using? Compressed or uncompressed?

I've read a lot where users say that there's no noticable difference between compressed and uncompressed DNG files. Well on my M9's I rekon there's a noticable difference most of the time. Not on every shot, but on a majority of the subjects I shoot. I have jet to do any "scientific" testing but I'm comfortable to say that I'll be shooting only uncompressed from now on. I'm seeing less artifacts in the areas with no detail and smoother transitions in tone. I don't know about DR but there's an extra something going on.

Or am I seeing something I just want to see?

Gordon
 
I shoot uncompressed, and with 8GByte cards- why not...

The M9 output needs 14 bits to capture the full range of the detector. So you are not wasting bits by storing uncompressed.

Shadow detail benefits most from uncompressed, and that is worth the effort.
 
What little I understand about the technology is that the M9 files do lose data if you shoot compressed. The new M, however, has lossless compression when you choose compressed DNG.

So I too shoot my M-E in uncompressed DNG.

On another note, as I've been reorganizing all my past images into a Lightroom managed system I'm finding I really wish I had shot all my past files as RAW and not just JPG. The latest image processing software is really incredible and I'm not happy with myself that I'm missing data to work with in my old files.
 
Another good reason to shoot uncompressed. LR4 improves noise reduction, and some of my older ISO2500 shots look even better now than before.

There are good lossless compression algorithms, usually get between 2:1 and 3:1 compression. BUT- the better ones take more specialized processing. The M Monochrom would lend itself to store running differences between samples. Store a 16-bit value, then all of the differences as 8-bit values, -127:127. The "positive 128" would be used as an "escape" that the difference exceeded 8-bits, and a new 16-bit value is present. Worked well in the 1980s... Wish I had the code and compiler for the DSP in the camera. The main advantage would be speeding up store to the SD card.
 
Another good reason to shoot uncompressed. LR4 improves noise reduction, and some of my older ISO2500 shots look even better now than before.

There are good lossless compression algorithms, usually get between 2:1 and 3:1 compression. BUT- the better ones take more specialized processing. The M Monochrom would lend itself to store running differences between samples. Store a 16-bit value, then all of the differences as 8-bit values, -127:127. The "positive 128" would be used as an "escape" that the difference exceeded 8-bits, and a new 16-bit value is present. Worked well in the 1980s... Wish I had the code and compiler for the DSP in the camera. The main advantage would be speeding up store to the SD card.

This is something I hadn't thought about. I'm a fairly low ISO shooter. I start to sweat at about ISO 640. It really may be time for a few controlled tests.

Gordon
 
Back
Top