This is not a big scientific test. I don’t plan on
expanding on it. It reflects my experience with the Canon
5D (over the past year) and Leica M8 (over the past few
weeks). I do shoot in poor light, and value that capability
in better digital cameras.
This test was prompted by Tina Manley’s Canon 5D/Leica
M8 test. Tina’s test shows a big edge for the 5D in low
light. I don’t dispute Tina’s results in her situation, and
agree that the 5D is amazing in low light, but her really
poor results with the M8 don’t match my experience. So I
did my own quick-and-dirty version of her test. It’s
interesting how people can do similar things and get such
different results.
So I grabbed my 5D and M8 off the shelf. They had my
most used lenses mounted already---50 f1.4 Canon, and 35
f1.4 Summilux ASPH (not coded). I set the Canon on 1600,
and shot one photo of a wall, then upped the ISO to 3200
and shot another. I shot a photo of the same wall with the
Leica. I set both lenses to f 1.4, as I nearly always do
when shooting at ISO that high (why bother with a fast lens
if you don’t use it fast when you need to?).
Here are my results. I shot RAW on both cameras, and
converted using Raw Developer. I had to set white balance
for all the photos, they all were way off (in retrospect,
it looks like I clicked a slightly different spot for
white, but perfection doesn’t matter for a noise test
anyway). I had to add just a smidgen of exposure to the
Canon shots; it appears the matrix meter missed ideal
exposure by a little, but not by enough to matter.
When I push the ISO with either camera, I generally
convert to black and white (I prefer black and white in
general, and when things get a little noisy, I prefer it
very much). So I’ve included some Raw Developer black and
white conversions. I capture sharpened all the Canon shots,
because it is clearly needed, and that’s my normal Canon
workflow. I provide both sharpened and unsharpened versions
of the Leica shots because I find that I often like its
look better with minimal (or no) sharpening.
I included some Canon edge crops with the vignetting
(very roughly) compensated near the end.
Anyhow, this is the way I really use these cameras and
lenses, and these results are representative of what I
typically see. And after this test I still think a
reasonable person could prefer either camera.