M9 + Lens Focus Setting

[k-hawinkler]

Advertisement (gone after registration)

In order to assess the focus setting of my M9+50 mm Summilux-M ASPH combination, I setup my 30" Apple Cinema Display with a screen orientation of roughly 45º with regards to the line of sight to the M9. I displayed 2 full MS Word pages in Full Screen mode, filled with Monaco 12 size characters, and hand held photographed the screen from close focus distance at f=1.4, focusing on the black 00 in front of the row of blue 00s. This is what I get:

 

http://winklers.smugmug.com/photos/760328043_XjTf7-O.jpg

 

On my screen characters are about 5 mm in height, spacing between characters is about 3 mm. First column, to me it looks like the DOF extends from about slightly below 00 to about the middle of 03. On screen I measure about 21 mm. That translates into a line of sight distance of about 15 mm, consistent with Leica's "Depth of field table" value in http://en.leica-camera.com/assets/file/download.php?filename=file_1769.pdf for close focus at 0.7m.

 

Another interesting detail is the pattern of faint greenish and reddish rectangular lines. This seems to coincide with the DOF, presumably caused by the lack of an AA filter and the sharp focus of M9+lens on the display screen's underlying ultra-fine pixel structure, and is reminiscent of a Moiré pattern.

 

As rick_dykstra has pointed out in the M9 Delivery Register thread, page 44, this front focus can be quite useful. I imagine taking a portrait near close focus distance with focus on a subject's pupil. Then everything from the pupil till 15 mm foward should be in focus. Nice. I had never thought about this. Thanks to Rick for pointing this out.

 

Provided my focus in the exercise above wasn't too far off and based on rick_dykstra's guidance, I now have to conclude that my M9 + 50 lux are just fine as they are. If it ain't broken don't fix it! However, additional advice is always welcome.

 

Now I wonder what Leica's policy is with regards to lens calibration?

I don't know the answer - but I can check out some lenses I got used from KEH.

 

First the 75 mm APO Summicron-M ASPH:

 

http://winklers.smugmug.com/photos/760454078_qxjxX-O.jpg

 

then 90 mm Elmarit-M:

 

http://winklers.smugmug.com/photos/760454100_ZWbME-O.jpg

 

And finally 90 mm Tele-Elmarit-M:

 

http://winklers.smugmug.com/photos/760454116_27pYz-O.jpg

 

It certainly looks like these 3 lenses are calibrated similar to the first one. Please, let me know if these lens close focus settings are the way they are supposed to be.

 

Thanks for all your help. I greatly appreciate it. With best regards.

[lars_bergquist]

I can add only two comments. First, a computer display is not a good target because the thing has depth. Were is the thing actually? Not on the front surface of the monitor's covering glass, at least. I would start worrying if you repeated the test with printed-out text on paper, and obtained the same result.

 

Also, you do not have an unequivocal focus point. Which part of the numeral do you focus on? The best way is to focus on a narrow line which runs in a plane perpendicular to the optical axis, and parallel to the short side of the focusing patch, i.e. at a right angle to the focusing movement in the rangefinder.

 

One pertinent question is what experience you have with rangefinder focusing. Another is whether a different camera, or a different lens, will give the same result. There are three possible scenarios:

 

• Only the 50mm misbehaves: Lens calibration error.

• All lenses misfocus on this body: Rangefinder caibration error.

• Results all over the place: Probable pilot error -- and this can also be relevant for the first two points.

 

Second, remember that a computer or TV screen is a mosaic of red, green and blue pixels. The camera sensor is ALSO a mosaic of red, green and blue pixels. If the resolution of the picture approaches that of the pixel pattern, moiré effects are to be expected.

 

The old man from the Age of Letterpress Printing

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Jager]

I raised a similar question last month...

 

http://www.l-camera-forum.com/leica-forum/leica-m9-forum/109385-lenses-cameras-warning-long.html

 

In addition to agreeing with Lars that a computer monitor is not a very good subject for assessing lens focus calibration, I'd also suggest that you really need to do these kind of tests off a tripod. You also ought to consider focus-bracketing to try and mimimize the variability inherent in manual focus.

 

You're certainly not the first to worry about this sort of stuff. My Noct and 90 Elmarit are going back to NJ this week for recalibration.

 

Good luck!

[cliffp]

I took delivery of my M9 a couple of weeks ago and noticed that some photos of swans I took at a distance of just over a metre (50mm Summicron at f2 and 1/400s) were not completely sharp. I photographed a test chart at 45 deg from a distance of around 1.2 metres (f2 and 1/125) and could see that it was front focussing by around 20mm (I took two test shots which did not differ by more than 5mm). I repeated the experiment with a 90mm Summicron Apo at f2 and a distance of 1.2m. The error was the same.

 

Should I send the camera for calibration or try and correct for it? The only way I can envisage correcting the error is by trying to focus on a feature 20mm further back than the intended focus point - not always easy to find such a feature.

[okram]

Computer screen is excellent for testing lenses, or even flat panel tv.

You dont need anything showing on the screen- what you need is a piece of tape that you stick on the screen surface and later observe the unsharpness moving on the screen pixels.

 

I might add, that this focusing error is with 50 so small that it is smaller than real world focusing capabilities of this camera. You will make much bigger mistakes by your own- I all tend to lean forward while shooting, and often after focusing...and with all cameras...

[cliffp]

Okram

 

Your leaning forward comment has made me think of a possible solution - focus on the eyes of your subject and move the camera forward 20mm!

[Guest trond]

Advertisement (gone after registration)

I have just performed a carefully calibrated focus test of my M9 and my Leica lenses.

 

The test was done at 0.7 meter, 1.0 meter and 1.5 meter distance from camera sensor to the focus target.

 

As the focus target I used the LensAlign PRO calibration system, which allows careful mesurment of exact focus plane and depth of field.

 

The LensAlign PRO is laser calibrated to make sure that the camera sensor and focus target are absolutely parallel during the calibration.

 

These are the results @ 1 meter:

 

21mm Summilux: approx 10mm back focus

35mm Summilux Sample A: 20mm front focus

35mm Summilux Sample B: 10mm back focus

50mm Summilux: 0mm focus error

90mm APO-Summicron: 20mm back focus

 

In the case of the two 35mm Summilux lenses and the 90mm APO-Summicron, the focus error is grater than the depth of field, with resulting loss of sharpness (significant loss).

 

The combined tolerance between the lenses and the camera body is in the order of 2% in my case.

 

I do not know what Leica's spec is for their allowed tolerance in rangefinder and lens calibration, but 2% does not seem unreasonable to me.

 

My conclusion of this test, in my case, is that the sharpness of the lenses and the M9 sensor, outperforms the rangefinder accuracy by a significant margin.

 

The M9 rangefinder is just not up to the task of extracting the performance possible by Leica glass and the 18Mpix sensor.

 

Remember also, that the test was performed using a stationary well lit high contrast target which is very easy to focus accurately.

 

Best regards

 

Trond

[okram]

In my experience with M8 (same pixel size- smaller sensor-even more demanding) even with the 1.25 magnifier, the human error (or RF-human collaboration) at 1-2meter wide open on f 1.4-2, is about 2cm with 35 and up.

 

In practice, it is more often something else...the movement of the subject or photographer, disregarding the shutter speed...but whats important to be aware of is that, with my 1Ds2, or 5d2 with AF, (with 50 and 85/1.2- lenses and apertures) the failure rate in practice is about the same...

[cliffp]

I have tried taking a few shots on the 90mm at f2 using a technique of focussing, then moving forward an estimated 20mm and comparing with just focussing and I do get better results. The subjects in each case were not moving and I took my time.

[okram]

I one sent my 3 lenses and a camera altogether to Leica, and they returned it perfect- of course for free. It takes about 20 days altogether with shipping.

If you are not satisfied, do so.

[pascal_meheut]

The M9 rangefinder is just not up to the task of extracting the performance possible by Leica glass and the 18Mpix sensor.

 

I had my M9 rangefinder perfectly calibrated by a technician in Paris. Took a few days, costed me nothing.

 

I then tested my M9 with 24mm/2.8, 28mm/2, 35mm/2 ASPH, 50mm/1.4 ASPH, 75mm/2 ASPH, 90mm/2.8 Tele-Elmarit and 90mm/2.8 Elmarit.

 

All these lenses were spot-on except the 90mm Elmarit and I know the problem is with him.

A friend also tested tenth of lenses using a DxO chart and their software, including the 50mm/0.95.

 

The conclusion is that the rangefinder is perfect and results are excellent.

 

So maybe you should not extrapolate your experience. The M9 rangefinder is perfectly up to the task if your lenses and your body are well calibrated.

[Guest trond]

In my experience with M8 (same pixel size- smaller sensor-even more demanding) even with the 1.25 magnifier, the human error (or RF-human collaboration) at 1-2meter wide open on f 1.4-2, is about 2cm with 35 and up.

 

In practice, it is more often something else...the movement of the subject or photographer, disregarding the shutter speed...but whats important to be aware of is that, with my 1Ds2, or 5d2 with AF, (with 50 and 85/1.2- lenses and apertures) the failure rate in practice is about the same...

 

I totally agree with you!

 

I own both the 1Ds3, 5D2 and the two f1.2 lenses. The 1Ds3 had pretty accurate focusing right out of the box, but the 5D2, the focus was "all over the place".

 

However, both cameras had consitent repeatable focus error, differet for each camera/ lens combination.

 

Both mentioned cameras have the possibility of focus calibration by the user. Calibration can be done in the matter of minutes, using the above mentioned method.

 

Once calibrated, focus is only rarely missed on portrait head shots with focus on the models closest eye, even in dim light where manual focus would be difficult.

 

High performance lenses like the mentioned Canon L-Glass, and even more so Leica lenses, does not allow any focus error on 20+/- pixel sensors, if optimum sharpness is called for.

 

I have found the Leica rangefinder to be very consistent, but the tolerances in the assembly and calibration of lenses leads to too much focus error, just as with autofocus came/lenses.

 

I just hate the idea of packing all my Leica stuff in a large box and ship it to Solms for five-six weeks for calibration at a cost of 280 Euros for each lens!

 

Best regards

 

Trond

[okram]

But you can post them under warranty as M9 malfunction- and then it should be free.

[Guest trond]

Good point Okram!

 

I will try that.

 

Best regards

 

Trond

[richfx]

I've had my M9 for two weeks now (aving taken appr. 1,700 shots). and it has focused perfectly with a new 28mm Elmarit and a 90mm Tele Elmarit M built in 1982. I have been able, with both lenses, to get razor sharp image focusing through the viewfinder and while viewing through Aperture 2.1.4. I've been shooting primarily in shutter priority mode, with uncompressed DNG and JPEG fine file formats.

I took delivery of a new 50mm Summilux ASPH yesterday (amazingly lucky to find one through Allen's Camera Shop in PA after my local CT shop gave up on getting one a week ago), and am very troubled. First off, even when the focus appears sharp in the view finder (which is much more difficult to obtain than with the 28mm and 90mm - just can't get that focus "lock" with the 50mm), the image is out of focus when viewed in Aperture.

Secondly, and more troubling the moving RF metering field is not parallel with the fixed RF metering field. It shows a slightly angled image as it moves into focus with the fixed field - perhaps 5 or so degrees off vertical and the top is farther to the right than the bottom, a la Leaning Tower of Piza.

I haven't dropped the body or lens; I have removed and reattached the 50mm Lux and turned the M9 on and off, including removing the battery, to eliminate those possible causes.

Any ideas or suggestions other than sending the lens back to Leica in NJ (I live in CT, so the trip would be easy).

Many thanks,

Rich

[jaapv]

The positioning of the rangefinder window cannot be influenced by the lens mounted, so we must take that out of the equation. Having said that, my new Summilux 50 asph frontfocussed out of the box. CS Solms took one look at it and kept it for a few weeks, now it is perfect.

Edited February 10, 2010 by jaapv

[Guest mc_k]

...the moving RF metering field is not parallel with the fixed RF metering field...

 

Apologies if you've used a rangefinder before, but rotate the camera a little and watch how that changes things in the patch. Keep your fingers away from the little window on the shutter button side, or you won't be able to focus. You should go back and try another lens, and then try the 50 again, before you send anything back. Your 50 pictures may have been out-of--focus because you weren't using the rangefinder properly.

[richfx]

Thanks, Gents. Very little experience with RFs before the M9, but focusing with the 28m and 90mm have been a breeze and shots with them have been superb in terms of clarity and IQ. I shot the same images with all three lenses today and it seems that the 50mm is front or back focus shifting. I e-mailed Leica USA today and await their suggestions. If they can calibrate the 50mm in NJ, that would be great, as I'm only an hour or so drive away. If not, I guess it will take a trip to Solms. Wonder how long that RT will be.

Rich

[hoppyman]

Hello Trond, this kind of technical experimenting can be appealing. In my opinion you need to be careful about which conclusions you arrive at though.

Firstly, what are you testing here (i.e. rangefinder adjustment of your specific M9 and/or the calibration of your specific lens samples. The results will only be valid (within your limitations) for those specific combinations on that specific test setup.

 

My personal experience was that Solms calibrated my M8 body (on upgrade) to nominal ideal and then later my two most critical lenses as a separate process. The lenses are not matched to a specific body; each item is calibrated to nominal ideal.

 

You are perhaps aware that Leica performs more elaborate versions of this testing after the camera body has been assembled (for rangefinder and sensor positioning also). Of course Solms has much more sophisticated test rigs and measuring equipment. Although the LensAlign folks may well use some laser measuring system when fabricating their target, that is not the same as you being able to apply the same accuracy to your setup. I don't know to what extent the viewfinder/lens axis displacement affects the accuracy of your test (designed for SLRs) either. Logically any error would be more prominent at closer distances.

 

Here are some considerations for you to explore if interested.

At what aperture settings did you perform your testing?

It is not only the amount of DoF at each specific aperture but where the best focus is positioned within that acceptable area. (i.e. it's not necessarily in the centre). The acceptable area of course is a value judgement and the result is also dependant on your measurement method/criteria there.

Consider also where the lens is best optimised in design(aperture/distance). It will always be a compromise and you cannot have best accuracy at every combination concurrently. That characteristic won't be the same across a variety of designs/samples either.

Are you aware that faster lenses in general are more prone to focus shift as the aperture varies? For example, the 35 Summilux ASPH. is documented as being especially susceptible. Lenses such as the 50 Summilux ASPH. and the APO 75 Summicron ASPH are documented to be much better designs in that respect (and others!)

 

I don't consider that a 2% tolerance is acceptable in the most critical applications. 20mm error is significant at 1m. The classic application is focus on the near eye with the lens wide open. However whether that tolerance falls within your technique variability in use may well be a different issue.

 

Leica's engineers disagree with your conclusion that the rangefinder is not sufficiently accurate to allow for best performance with the best M lenses and the M9 sensor. Anecdotally many users do as well. My personal experience is that the system accuracy exceeds my ability to make use of it in operation.

 

In a controlled test such as this, I would expect that your judgement of the focus point using this rangefinder would be more precise than using any dSLR focus screen. I'l leave that to those with more expertise/experience to make comparisons.

 

As a practical matter a good follow up to this type of testing is to shoot a lot and share the images if inclined. It helps with maintaining perspective.

[Advertisement]

Hallo,

Look here: Leica M

[Guest mc_k]

...It is not only the amount of DoF at each specific aperture but where the best focus is positioned within that acceptable area. (i.e. it's not necessarily in the centre)...

 

In a focus test, though, the plane of best focus is usually taken to be in the center of the depth of field. I am not aware that Leica does otherwise.

 

For the focal lengths and apertures you would be testing, the approximation is usually good. The "one-thirds, two-thirds" rule is maybe not so good.