LensRentals.com

MM MM Good?

Posted by

The first Leica M Monochrome (I think everyone is calling it the Leica M-M; at least everyone around here) showed up today. Lensrentals became a tiny reflection of most online forum discussions regarding the camera, with half the employees saying, “Why would anyone want that?” and the other half drooling. My own opinion has vacilated between ambiguity and indecisiveness since it was announced. Being the only impartial person in the building, I decided I should do the initial evaluation.

Please don’t consider this a review; it’s not. Reviewers spend days carefully evaluating a camera looking for all its strengths and weakness. I spend a few hours with it, run some tests, get bored, and move on to the next thing. I don’t do reviews, I do first impressions.

In this case, it’s the first impression of a guy who shoots with an M9 occasionally, loves the great shots it makes, and hates that I still can’t focus it and miss 80% of the shots I take.

Imatest Data

I’m a gearhead and resolution freak. Which means the first thing I did was take the MM and run it through Imatest with one of the better Leica lenses: the 50mm f/1.4 Summilux ASPH. The results surprised me a bit. I had thought that with no Bayer array there would be an increase in resolution. The Imatest results showed an increased resolution, but it wasn’t as large as I expected.

 

ApertureMM PeakMM AverageM9 PeakM9 Average
Leica 50mm f/1.4 at f/1.4737628600530
Leica 50mm f/1.4 at f/2.811249001025860
Leica 50mm f/1.4 at f/5.6115010281112976

 

The difference between the two cameras was significant, but just barely so. In other words, you should be able to see a difference in a reasonably large print, but you’d probably have to look at them side-by-side to tell which was the MM and which the M9 (Assuming you made the M9 image black and white, of course. Otherwise the color thing would pretty much give it away).

Skipping ahead a bit, though, after we looked at the MM images it was obvious they showed a lot more detail than the M9 images. It didn’t take a side-by-side comparison to tell the difference; it was clear. That didn’t make sense given the numbers above.

So we went back quickly (because our camera turned into a pumpkin and went to shipping at 3 pm) and repeated some Imatest measurements using JPG images instead of RAW images. We’ve always tested RAW images because we want to minimize any artificial sharpening applied in-camera (even with sharpening turned off, most cameras still apply some). Other testers use JPGs because they want their tests to reflect what you actually get from the camera’s JPGs. Neither way is right or wrong, it’s just two different ways to do it.

But it occurred to me that perhaps Imatest’s raw converter was applying a demosaic algorithm for a Bayer filter that wasn’t there that might make the results of the test show a lower resolution. So we repeated Imatest using in-camera JPGs with the Leica 35mm f/1.4 ASPH II.

 

PeakAverage
M9 RAW1050810
M9 JPG1260912
MM RAW1120835
MM JPG20501480

 

As with the previous tests, the MM RAW is a bit better than the M9 RAW. But the MM JPG is hugely better than the M9 JPG. I think that perhaps the DCRAW converter used by Imatest to provide a level playing field for all cameras is messing up the MM files when converting them for testing. It’s also possible that the MM files can be sharpened in-camera incredibly more than M9 files (despite setting everything to not sharpen them).

The bottom line, though, is that this time the non-gearheads are right: throw away the numbers and go take some pictures.

Dynamic Range and High ISO

I won’t bore you with any testing numbers here. We have the “Roger’s favorite bad lighting test” here at Lensrentals. It’s pretty simple — we aim a 500 watt halogen light at a 30 degree angle to a couple of shelves full of antique lenses and books. It creates all kinds of glare and shadow areas with objects ranging from flat white to nearly black.

To summarize things rather simply, the images below (top to bottom) compare 100% crops shot with a Canon 5D II with 50mm f/1.4 at ISO 3200, a Leica M9 with 50mm Summilux f/1.4 at ISO 1600, and a Leica MM with the same lens at ISO 3200.

All were shot at f/5.6 but as you can tell from the closeups that wasn’t enough depth of field to make the vagaries of focus identical. The only point worth taking away (and it is worth taking away) is that the M clearly has less noise and better clarity at ISO 3200 than the M9 does at 1600. Whether it’s better than the Canon or not I don’t think is important. It’s clearly in the same vicinity and maintains at least as much detail. That’s a big change compared to the M9.

Shooting Around

The well-dressed MM wearing the Leica 50mm f/0.95. Just the thing for casual shooting around an industrial park.

 

I would love to have arranged a DigitalRev type swimsuit model session to test the camera out. But they’ve got a lot more free time than us. So Aaron and I walked around the scenic industrial park and warehouse we call home on our lunch hour and took some shots.

 

 

Could we have made the above images with some time in Channel mixer and an M9 file? Maybe. The small online JPGs don’t show it, but the amount of detail is much more than I’ve gotten from M9 images (It was the amount of detail we were seeing made me really question the Imatest results).

And we certainly couldn’t have taken this shot of the top of a cabinet in an unlit corner of a workroom with an M9. Because it’s shot at ISO 10,000.

 

 

If you’d like to pixel peep the noise in this ISO 10,000 image, here’s a 100% crop of a strap with both shadow and white table top so you can see noise at both ends of the luminance range.

 

 

There are some SLRs (not all of them by any means) that can do this well at ISO 10,000. There aren’t any that I can say are just absolutely better, though.

So What Did I Think?

Well, if you’re asking me would I buy one, the answer is no. I’m not buying an M9 either. That’s because I still can’t focus a rangefinder accurately and consistently (It’s not just lack of skill, it’s a vision problem, too. But lack of skill certainly contributes.). If I was buying one of them, though, it would be the MM, no question. I’d give up color for the resolution and high ISO performance in a heartbeat.

For the kind of shooting I do, there will always be something else I’d rather spend $8,000 on.

But will I take it out and play with it? Abso-damn-lutely!!! The images are awesome. I’ll only hit focus on 1 out of 10, but that one will be a good one. Now I can actually take it out for night ┬ástreet shooting, or to a party. Or in broad daylight. Did I mention I really like the images? OK, I guess I did.

There’s a rule that whenever you write about anything Leica, you have to work in the words “it has that Leica look” somewhere. This camera has a unique look, seriously. I don’t have the black and white film background to say it looks like this or that. I’d call it Black and White meets HDR. It’s very attractive, whatever it is. I do a lot of black and white conversions and I can’t say I’ve ever achieved exactly this look. If I had I’d have been bragging about it.

Roger Cicala

Lensrentals.com

August, 2012

14 Responses to “MM MM Good?”

Mark said:

You’ve been writing a lot about m4/3 lately. Would you be interested in a monochrome m4/3 unit (probably CDAF)? I know I would.

Uncle Mike said:

How would you say the image quality compares to a D800E B/W conversion?

LensRentals Employee

Roger Cicala said:

Uncle Mike,

I haven’t done enough D800E B&W to venture a guess. I would say it’s more detailed than any 5DII B&W conversions I’ve done, and I do those quite a bit, agonizing for half an hour in Channel Mixer every time.

Roger

bryan willman said:

For those of us who wonder about such things, how does that 2050peak 1480average number compare to other cameras and lenses? (For the ones I can find on your blog with google, it would appear to dominate the resolution contest. For only $12K or so…)

Also – any thoughts on a quick at home test to sort out the best raw converter available to one? (e.g. Easiest test to sort ACR vs converter-A vs converter-B.) Because for any of these high price cameras, you want to find the best converter you can live with.

Ron said:

Roger, at what distance did you test the 50 Lux ASPH? My impression of the lens is it becomes a bit soft at closer distances and wide open and seems better optimized for farther distances unless stopped down a bit (as reflected by the f/2.8 numbers). But this wouldn’t explain the huge difference in the RAW vs. Jpeg numbers.

Regarding Uncle Mike’s question: Ming Thein did an MM vs. D800E comparison, but take what you will from it, as it’s essentially also a comparison of the new super high resolving 50AA vs. the Nikon 45 f/2.8P. Not sure what your policy is to links, but this is his review: http://blog.mingthein.com/2012/05/27/leica-m-monochrom-vs-d800e/

LensRentals Employee

Roger Cicala said:

Bryan,

It makes for a difficult comparison once you go to jpg: there’s variation about how much sharpening is applied and there’s the possibility of spurious resolution in some cases (although we saw none of the signs of it here). I can say jpgs are generally 25% to 40% higher resolution than raw files and these were way more than that.

Until I hear back from the people at Imatest, and know exactly what’s going on, though, I’m not sure which numbers involving the MM are real. But looking at the image resolution I’d say it’s close to D800 anyway. Since it’s a Leica camera, though, I’m required to say it has a different look. But actually, it does have a different look. I don’t know exactly how to describe it, but I think Black and White HDR is kind of a good description.

Roger

LensRentals Employee

Roger Cicala said:

Ron,

Excellent point about distance. We were at roughly 15 feet with the 50, which is why I believe the 35 was a better choice for comparison (I think it’s a bit better close). At the time I thought perhaps the 50 close up was a resolution limiting factor. I’m still not sure what the jpg / raw difference means, but I can say when I converted in Photoshop it was obvious the conversion was lightning quick compared to color files, which wasn’t the case with Imatest’s conversion in DCRAW, which is what got me thinking about this

And please link away — comments on this blog teach me more than anything else. That’s a great link, if for no other reason than it’s one of the few times I get Lens Envy. I wish I had one of those :-)

Huggs said:

Somebody shops at ~SkyMaaaaall~.

Antoine said:

Hi Roger,
The Lux 50 mm glass is a bit soft at close distance. Run it again with the Cron 50MM / f:2.0…. You should see a difference.
Cheers

Christoph Breitkopf said:

While that may not be a problem for Leica users, what would keep me from going for a BW Nikon (if there was such a beast) is lens quality. If I shoot in color, I can correct some color aberrations easily before converting to BW. If I shoot in BW with a lens that has visible chromatic aberrations, I’m out of luck. And some of my favorite lenses (Nikkor 2.8/180, any version, for example) do have CA.

Frans van den Bergh said:

Hi Roger,

“dcraw -D” should process a raw file without applying any Bayer demosaicing (you can also tell that this is quite fast). Imatest should be able to process the resulting grayscale file without problems.

I would be happy to take a look at the files (just email me some Imatest chart shots).

Some further comments:
1. That value of 2050 for the MM JPEG image sounds a bit high. The MTF50 for a diffraction limted lens at 550 nm wavelength, mounted on a sensor with a 6.8 micron pitch (i.e., the MM), would be more like this:
f/5.6 : 1618 lp/ph
f/2.8 : 1861 lp/ph
f/1.4 : 1981 lp/ph (but obviously the lens is not that good)
I suspect that some sharpening has indeed been applied to the MM JPEG, or something else is amiss.

2. The Bayer colour filter does not prevent you from measuring MTF values near to the expected resolution limit given the pixel size, provided that you measure MTF using the slanted edge method applied to only one channel of a raw file. In other words, on a D800E, you should be able to measure right up to the diffraction limit if your lens is diffraction limited. Note that this is possible because the slanted edge method (used by Imatest) measures resolution over a large number of pixels spanning a step edge — in real world photos the Bayer filter will indeed appear to reduce resolution in certain orientations.

If you demosaic the image first, then you may measure MTF50 values that are quite a bit lower than what the individual channels can give you, which is consistent with what you are seeing on the MM.

Thorsten Overgaard said:

Thanks for a really interesting test served in a enjoyable manner :-)

I think what summs up the Leica M Monochrom is the fun factor: It’s a camera you want to play with and see where will bring you.

If it doesn’t have that factor, then forget it anyways.

It’s interesting that I bought the Leica M9 when it came out as a “fun camera” to play with besides my Leica dSLR cameras. As it turned out, it became my main camera in less than a year.

I don’t see how a black & white camera can become a main camera in a world where clients and all demand color (but do like black & white when they see it), but I think the Leica M Monochrom will open up new ways of viewing photography and black & white photography.

/Thorsten

Ralf C. Kohlrausch said:

Hi Roger,

could you please do a test of the lens on the left side of the cupboard ;-?

Greets
Ralf C.

Lee said:

Ming Thein’s is not very scientific comparison. Lens choices goose it big time in favor of the M-M (although, to be fair, I think he’s pretty up-front about that in the text of the piece, even disclaiming to go by his words more than the images).

But if they’d both been bayer sensors, I doubt even putting a Master Prime on the 18MP and a Samyang on the 36MP would’ve made the former come out on top (well, that might be hyperbole :P ). To me that general point is the interesting take-away.

Leave a Reply