This week some absolutely-completely-expected data from DXO came out for the M9 and the web forums caught fire. All the usual tripe came out – cheaper cameras push better pixels. Yes! Whatever!
I’ll say my position on this: Leica’s cameras are desirable only because they mount Leica’s lenses. If the M lenses weren’t so exquisite or so small, Leica probably could not sell M cameras above their manufacturing cost. However, support for M lenses without cropping is a significant cause for enthusiasm about the M9.
In my humble opinion, the serious M9 buyer isn’t buying the M9 because it’s the best sensor in the business. To a first order the M9 is only required to not suck eggs because it uses M lenses the way they were intended to be used.
Now, let’s see if we can make more sense of all the predetermined-yet-controversial data.
Here’s a comparison that speaks to me, from first-hand experience:
The D80 was a good camera. Under ISO 1000 the D80 delivered excellent results in my hands, even indoors, and even in crappy light. When I paired it with a Zeiss lens the D80 just roared to life and produced results way above its price tag. I probably came close to extracting maximum image quality from the D80 towards the end of my time with it, and those images still look as good as the best images I’m taking today.
The D700 is a better camera. Purely judging from pictures, upgrading to a D700 brought me 1 extra stop of dynamic range and 2 extra stops of sensitivity with acceptable noise. DXO agrees with both, note. However, these aren’t all the improvements in the D700, and there are plenty of conditions where they are basically moot.
In my review of the D700 (forgive the bad product photography, not my cup of tea) I concluded that the D300 would have been a better choice, because it has all the same non-sensor improvements for half the price. A year later, still with the D700, my final verdict comes down to the crop factor or lack thereof. I’m very happy with the D700’s sensor but basically because of its physical dimensions.
One thing that the D700 did not terribly improve over the D80, and DXO doesn’t take into account much, is resolution. DXO noise results do account for resolution in a way, but they never tackle the subject of image acuity and microcontrast. In this area the D80 and D700 definitely fall within 10-20% of each other and the M9 ought to trounce both (expect ~50 better). If any camera system can actually deliver a perfect 18MP image sample it’s the M system, as opposed to Canon’s APS-C system, say.
Let's combine the DXO data with other M9 facts to draw a bigger picture:
The M9’s color depth is only slightly better than the D80.
The M9’s dynamic range is halfway between the D80 and D700.
Low-light ISO on the M9 is maybe a half-stop better than the D80.
The M9’s image acuity is about 50% better than the D80 and D700.
An M9 with lens is less than half the size and weight as the D700 with lens.
That’s a very good balance – I would gladly give up some high ISO performance for greater acuity and shoulder relief. Especially if I can make up some sensitvity loss by leaning on fast Leica lenses, and here the trade-off gets really interesting.
As you know, I own a Zeiss 1.4/50 Planar ZF. You may not know that I simply do not shoot this lens at f/1.4, ever. At this aperture the bokeh is ugly and distracting, micro-contrast is lost in various fringe effects, and it is extremely difficult to focus closer than 12 feet if ambient light is low. I use this lens starting around f/2.4 - 2.8, where it is much easier to focus and produces beautiful bokeh and micro-contrast. This is basically 1.5 - 2 stops of light-gathering power that I choose to throw away because the D700 can make it up in ISO sensitivity.
Judging by lens availability right now, most M9 owners also carry a 1.4/50 Summilux-M ASPH. This lens has beautiful bokeh at all apertures, is essentially apochromatic, exhibits almost no focus shifts and, best of all, delivers very high micro-contrast right from the maximum aperture. Focusing in low light is easier on a rangefinder than an SLR ground glass. I would shoot this lens wide-open by default, or step down 1 stop to gain some depth of field and forgiveness of focus error.
Some people will cry foul, that I’m excusing sensor limitations by finding lenses that mitigate them. But these lenses in particular can only be used with this sensor, so the two might as well be considered together. They would be right to complain if the Summilux could be mounted on a D700, but alas this is not possible.
So, in my opinion, the M9’s ISO performance being lower is both predictable and partly a moot point. The M9 clearly does not suck and that’s really all it needed to (not) do to succeed.