Thanks for this test Luka, very well executed and presented!
A few comments:
In general, the lack of AA filter in the Leica and different color rendition confuses the heck out of me. I am not sure if I can make anything out of the ultimate sharpness, resolution, contrast and color at f/5.6 other than for the systems as a whole.
Test E:
Left edge crop: the Canon clearly is affected by field curvature.
Test D, and this wil probably get me flamed:
Leica: The smooth bokeh gives to me the look that the Canon L primes are smashed for, flat and uninteresting, sharp foreground and creamy background, two separate layers, no depth.
Canon: The bokeh is more interesting, rendering the details in the background better and still providing separation and overall better sense of depth.
even if the conclusions aren't a surprise, it's very interesting to see these two lenses juxtaposed. thanks for taking the time to post this comparo =)
alundeb wrote:
Test D, and this wil probably get me flamed:
Leica: The smooth bokeh gives to me the look that the Canon L primes are smashed for, flat and uninteresting, sharp foreground and creamy background, two separate layers, no depth.
Canon: The bokeh is more interesting, rendering the details in the background better and still providing separation and overall better sense of depth.
I kind of prefer the Canon rendition in test D as well. It has a bit more 3D and the bokeh while not as smooth is more interesting.
However the photos are not taken from the exact place so could this have affected the result?
millsart wrote:
I've got to agree regarding the Sonnar, mine was certainly a bit soft and dreamy wide open (and hard to focus as it was calibrated for f2.8) but by f5.6 is sharpened up quite nicely and was neck and neck with my Planar at f8
I was checking out Sean Reids site as well with some samples and his copy was really good as well stopped down.
In your test I just don't see any real fine detail at all, which I've found the M9 really can deliver on even with some of the lesser regarded lenses. Just doesn't look like focus was achieved as the image is so soft all over. ...Show more →
´
I can't swear by the Sonnar results as I didn't do any focus bracketing - in case the mechanical infinity is off. I think however the results are fairly consistent with what I've seen before. I can do a more proper test later to confirm it.
edwardkaraa wrote:
I'm going to wait until Leica introduces an EVF body or LV. There are too many issues with focus calibration as far as I understand.
For those who actually own the M8/9, what do you think about the below comments from pebbleplace.com
Yeah, those are pretty accurate, but it's a bit more complicated than that though.
Wide angles are a breeze. I've never missed focus with a lens < 50 mm, including the 35 Summilux ASPH @ f/1.4. With a fast 50 mm lens I get perhaps 1 in 50 OOF at the largest aperture - still generally acceptable. With my 75 Cron that increases to a worrying 1 in 10 wide open - especially at medium distances and if I'm stressed. I've been really unhappy about a couple of shots that I've missed - mostly involving people at medium distances. The 90mm I have to focus with real care and still miss in about 1 in 10 as well.
magiclight wrote:
He found it very difficult to focus with the lenses wide open. I imagine that this would be rather frustrating with you expensive Lux.
Not the 50 Lux - it's still fine for focusing. I have found the 75 APO-Summicron ASPH somewhat frustrating to focus (I've missed critical shots).
alundeb wrote:
Leica: The smooth bokeh gives to me the look that the Canon L primes are smashed for, flat and uninteresting, sharp foreground and creamy background, two separate layers, no depth.
No, no, that's not even close to Canon bokeh. Canon L bokeh is much more creamy than that. This is typical modern Leica bokeh - as perfectly gaussian as it gets. It is very close to what you would get with a photoshop filter - i.e close to a theoretical "ideal" lens model. It's very neutral.
As we all know bokeh preferences are personal so there is no objective standard there. Personally I find the Canon bokeh in test D offensive and the Leica bokeh rather uninteresting. Neither is a lens that I'd use for say closeups of flowers where you want the OOF region to be lively and interesting. The Leica type of bokeh usually works exceptionally well for portraits where it does not detract from the subject.
When I first started using Leica glass, I really disliked the OOF rendering as I had hoped for something Zeiss-like. Over time however it has grown on me and I'm currently using my Leica lenses far more than my Zeiss ZM lenses on the M9.
edwardkaraa wrote:
However the photos are not taken from the exact place so could this have affected the result?
It's taken from the same spot but the Canon distortion + the difference of the height in the cameras makes it look like the vantage point has shifted when in reality it hasn't.
denoir wrote:
No, no, that's not even close to Canon bokeh. Canon L bokeh is much more creamy than that. This is typical modern Leica bokeh - as perfectly gaussian as it gets. It is very close to what you would get with a photoshop filter - i.e close to a theoretical "ideal" lens model. It's very neutral.
Is there a resource comparing the bokeh of the Canon 50 1.2L at f/1.4 (not wide open) and the Leica Summilux-M 50/1.4 ASPH ?
The differences between the Canon 50 1.2 L and 50 1.4 (William Castleman comes to mind) seem to me about the same order of magnitude as the differences seen here, but extrapolation is a risky business.
alundeb wrote:
Is there a resource comparing the bokeh of the Canon 50 1.2L at f/1.4 (not wide open) and the Leica Summilux-M 50/1.4 ASPH ?
The differences between the Canon 50 1.2 L and 50 1.4 (William Castleman comes to mind) seem to me about the same order of magnitude as the differences seen here, but extrapolation is a risky business.
The bokeh of the Canon 50L at f/1.4 is very close, if not identical to the one of the 50/1.4 at f/1.4 (I've had both simultaneously). At f/1.2 it's a different story. The 50L also produces much nicer colors IMO.
As I don't have the 50L any longer I can't perform the test.
denoir wrote:
The bokeh of the Canon 50L at f/1.4 is very close, if not identical to the one of the 50/1.4 at f/1.4 (I've had both simultaneously). At f/1.2 it's a different story.
Are you sure, the 50L seems much closer, if not even past the Leica in smoothness at 1.4, which was you original objection.
Interesting, that's not how I recall it but that test clearly shows it is smoother at f/1.4 than the 50/1.4. At f/2 they seem quite similar though.
But it's definitely not the gaussian type of blur of the Leica - the characteristics of the blur is quite different. You can see it on the sign on the left side where the OOF areas are 'shifted' in the blur and not radially dispersed. The closest I can think of is Minolta type of blur. Personally in many situations I prefer that type of blur as it gives a bit of form and shape to the background.
Great test Luka, thanks for doing this. I wonder why some of these other manufacturers don't get rid of the image robbing AA filter on some of the higher end cameras. I understand the lower end models, but I would bet a 5DMII without an AA filter would sell even better than the current
It's a double edged sword. My NEX-5 has little to no AA filter on it, and moire shows up pretty easily. The NEX has a higher pixel density than the 5Dii, so the 5Dii would show even more moire. While we all get excited about sharpness around here, one has to take a step back and decide how much real improvement we're getting in prints with more sharpness, if the occasional moire can ruin a shot altogether. Granted, if you shoot b&w, it's a great option. With color, I'm on the fence.
Luka,
Thank you for a great comparison . Echoing to what Jeff (Millsart) said about the Sonnar - although optically the same but some early copies of this lens were designed to perform optimally for film cameras. These lenses will show a significant focus shift on digital bodies. They can be adjusted by Zeiss to perform optimally on digital bodies but those lenses will not be optimal anymore on film bodies. Well, if you stick to one recording format and your lens has been adjusted to that then you will get one heck of a lens. I am not sure whether the copy you tested was adjusted by Zeiss or not. A note though that lens is indeed designed to get dreamy results at wide open or slightly wide open apertures but it is sharp when stopped down.
In addition, I am sure you were referring to your copy of the Summilux, which is a silver/chrome version. Yes, a black Summilux is still quite a bit heavier but the weight difference between the two lenses will be less pronounced.
rsolti13 wrote:
Great test Luka, thanks for doing this. I wonder why some of these other manufacturers don't get rid of the image robbing AA filter on some of the higher end cameras. I understand the lower end models, but I would bet a 5DMII without an AA filter would sell even better than the current
I think the moire issues probably would have too many users up in arms that their $2500 AA filterless camera is producing such issues.
It depends on what you shoot of course, but the M8 and M9 can produce a fair amount of moire on specific materials and patterns. I've found C1 pretty good in dealing with it, but sometimes additional PS work is needed, and as a whole, its never been a dealbreaker for me, but I can't say that would hold true for everyone.
I think an AA-less camera would be a bit too much of a niche market to see the likes of Canon/Nikon produce.
There are services out there that will remove the AA filter, for a modest fee, but I've never really found those services too popular. I think if all it took to get M9 like per pixel sharpness was $500 and sending in your 5D mkII everyone would do it.
AGeoJO wrote:
Echoing to what Jeff (Millsart) said about the Sonnar - although optically the same but some early copies of this lens were designed to perform optimally for film cameras. These lenses will show a significant focus shift on digital bodies. They can be adjusted by Zeiss to perform optimally on digital bodies but those lenses will not be optimal anymore on film bodies. Well, if you stick to one recording format and your lens has been adjusted to that then you will get one heck of a lens. I am not sure whether the copy you tested was adjusted by Zeiss or not. A note though that lens is indeed designed to get dreamy results at wide open or slightly wide open apertures but it is sharp when stopped down....Show more →
Joshua, my 50/1.5 Sonnar is optimized for f/2.8 I have been meaning to send it to Zeiss for FFD calibration to f/1.5 but have not gotten around to doing so. At f/2.8 and beyond it's dead on as far as focus goes, but it is possible that the mechanical stop that I used in the example goes beyond infinity. I'll try to shoot a new series and do some focus bracketing.
Having said that the MTF chart ( http://www.zeissimages.com/mtf/zm/CSonnar_1.5_50_ZM.EN.pdf ) shows fairly poor resolution at f/4 and quite a lot of field curvature. The comparison image that I shot was at f/5.6 - it's possible that you have to go to f/8 or smaller to get good results.
douglasf13 wrote:
While we all get excited about sharpness around here, one has to take a step back and decide how much real improvement we're getting in prints with more sharpness, if the occasional moire can ruin a shot altogether.
+1
As long as most of us still downsize for printing, the AA filter can only be beneficial, and its softening effect negated by the downsizing process. I think the AAless cameras would be most beneficial for photographers who print at the limits of the sensor's resolution.
I redid the Sonnar tests now. It was as I suspected - the mechanical infinity stop wasn't quite reliable - I had to pull back a bit. This time I did focus bracketing with four shots per aperture on the Sonnar. On the Lux I used the mechanical stop that I know is reliable.
The Sonnar doesn't really catch up to the Lux when it comes to fine detail until f/16 where diffraction has kicked in on both lenses. Still, solid stopped down performance the Sonnar at f/8 with the exception of the edges which never get quite sharp.
. Echoing to what Jeff (Millsart) said about the Sonnar - although optically the same but some early copies of this lens were designed to perform optimally for film cameras. These lenses will show a significant focus shift on digital bodies. They can be adjusted by Zeiss to perform optimally on digital bodies but those lenses will not be optimal anymore on film bodies. Well, if you stick to one recording format and your lens has been adjusted to that then you will get one heck of a lens. I am not sure whether the copy you tested was adjusted by Zeiss or not. A note though that lens is indeed designed to get dreamy results at wide open or slightly wide open apertures but it is sharp when stopped down.
