Leica Leica M9 vs M240 Resolution Comparison @ lens rentals

[serhan]

Leica M9 vs M240 Resolution Comparison using the Leica 50mm f/1.4 Summilux ASPH lens at lens rentals:
LensRentals.com - Leica M Type 240 Resolution Comparison

That 18-megapixel CCD had more resolution than one would expect from an 18-megapixel camera. The new M-240 resolves at least as well (measured by MTF 50) as the M9 did. It probably is just a bit better. The differences in the center and overall are pretty small and probably of no, or very little, significance. The difference in the corners, though, does appear to be approaching significant difference.

It is only one lens test, but interesting to see that the resolution wise it is not a big jump from 18MP to 24MP.

 

[jonoslack]

HI There
I didn't understand it like that - I assumed he was looking at the pixel per pixel sharpness, not the image as a whole. Could be wrong though!

As far as I can see, the cameras do resolve about the same for the same number of pixels - which means that the 24/18 ratio relates also to the ratio between the acceptable print sizes


all the best
Jono Slack

 

[serhan]

He is using "line pairs / image height" to measure the (raw image) sensor+lens resolution. So with same lens, higher MP sensor should give higher resolution for the same image height, but in this test sensor types are different with Leica's.

LensRentals.com - Why Aren't the Damn Numbers the Same????

Imatest, DxO Analyzer, tests a combination of lens and camera, not just lens.

The Range Chosen for Numbers
Even when presenting MTF 50 results, different testers will present their numbers differently. For example, I use line pairs / image height. Others might use lines / image height (so their number would be exactly double mine). They might also present lines (or line pairs) per mm instead of per image height. This is pretty simple, though. You can get out your calculator and compare line pairs per image height to lines per mm.

Were the Test Images RAW or JPG?
The Imatest results from an in-camera JPG will be significantly higher than if the test is run on a RAW image. How much higher varies depending upon the camera, but 20 to 40% higher is common.

 

[Amin]

He is using "line pairs / image height"

The image height dimension is 3472 pixels for M9 and 4000 pixels for the new M, so assuming ideal lenses and sensors, we'd expect the M9 to provide 87% of the LP/IH. Not too surprising that the new M would do "just a little better" in this regard. While an observed 3% improvement is less than a predicted 15% improvement, both are fairly insignificant to me. The key finding from Roger's perspective and mine is that the new M CMOS isn't a step backwards from the older CCD in terms of resolution.

 

[flash]

The key finding, to me is that the corners seem to be better. That possibly means an improvement in the use of the microlenses and therefore a more significant improvement for landscapes etc.

Then again, I haven't bought a new camera in years just because of the sensor in it. That seems to fairly low on my list of priorities. I had two camera with amazing sensors (NEX7 and a77). Don't have either of them anymore.

Gordon