CCD vs CMOS: Can you tell which is which?{merged}

[Lenshacker]

Advertisement (gone after registration)

Uncompressed DNG, ISO2500, -1ev, pushed 1-stop in LR, no sharpening, no noise reduction:

 

L1012652_DNGcrop by fiftyonepointsix, on Flickr

 

Same exposure, same settings, waited until the buffer was flushed, DNG-8,

 

L1012654_DNG8_crop by fiftyonepointsix, on Flickr

 

My M9 has very low noise, and I have used it at ISO2500 -1ev, uncompressed DNG. The Noise level is less than 1/2 of the M8.

 

I believe the DNG-8 compression routine increases noise where there are "boundaries between steps", but decreases noise once in the step. With the M8, more noise- more chances for bouncing between steps. The M9 has less noise, so less fluctuations.

 

If you look at the DNG-8, the noise seems to be "correlated", I think it is following the steps. The uncompressed DNG looks more random to me. Sit back from the computer screen 3ft or so, look at each. I am still learning how to read DNG files, I have the old IEEE "mixed-Radix FFT" routine in the code. I will try it out to see what is going on with the frequency makeup of the image.

 

I always thought the M9 could have added an ISO5000 setting.

Edited April 2, 2015 by Lenshacker

[friedeye]

Thanks for this, David. As an M9 user,  I think I hit about 90% - but who cares, really? The CMOS sensor, as you proved, is capable of… whatever one wants.  I love my M9 and its capabilities - they seem to synch very well with my film cameras; I switch from my M9 to my M2 or 3 a lot and without a hitch. And I find now, shooting in manual, that I rarely have to use a meter. But that said, the 240 is an amazing camera, capable of so much more than the M9 (low light, LV, and lenses) -- One day I'll own one - or, perhaps its successor. 

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama