M9 + Lens Focus Setting

[ho_co]

Advertisement (gone after registration)

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

How did you figure the location of the sensor? Leica has only been able to give me an approximation, since the M8 (and I assume, the M9) doesn't have a focal plane mark.

 

 

In a focus test, though, the plane of best focus is usually taken to be in the center of the depth of field. ...

I've never heard that before. I would think that "center of the depth of field" wouldn't even be calculable ad hoc, since DoF doesn't exist as a physical thing, but is created by the brain.

 

And how do you compare the point of highest resolution with the point of highest contrast if you're making such a calculation? I would be more willing to grant that "best focus" were centered between them.

 

Very interesting idea. Tell me more. Can you cite a source?

[Guest trond]

 

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.

 

Dear Hoppyman,

 

To try to clarify further, I have included the focus test image from the Summilux 50mm shot wide open at f 1.4, as shown below. The focus point is the central "siemens star" with he red dot. The rangfinder square covers just a little more that the central star in this test. The image is a 100% crop (center) of the original file

 

As you can see the lens is perfectly focused, and the lens and rangefinder is in perfect agreement.

 

The focus target is calibrated in such a way that when the five red dots covers the hole when the target i shot at a distance of 50x the focal length, the target is parallel with the camera. At shorter distance, as in this case, the dot will not cover the holes completely, but a symmetrical pattern should be obtained.

 

Parallel alignment can also be verified by the black center dot in the center of the different siemens stars, have the same diameter. The pattern in the center of the stars goes beyond the resolution of the M9 sensor, so the center is rendered as a uniform gray dot. When all dots are the same size the target is parallel with the sensor as shown in this case.

 

Remember also that the DOF of the 50mm at f 1.4 is less than 10mm in this case, if the circle of confusion is defined as the pixel size of the M9 sensor.

 

This test is very critical, and it is not possible to achieve good focus if not everything is perfectly aligned. It is thus very easy to verify that the rangefinder and lens agrees on the set distance, any error is immediately and easy visible using zoom view on the M9 screen.

 

All I care about is that my M9 and all lenses agrees on the distance, and they don´t.

Of my 7 Leica lenses, the 21 lux, 50 lux and WATE are perfectly aligned. The WATE is difficult to test because the DOF is so wide that it is very forgiving on focus error.

 

A tolerance of 2% between several mechanical devices produced at different time and place, is very difficult to achive in serial production no matter how strict the quality control, as even the temperature variation of lens and camera will have big influence on the focus accuracy.

 

Leica production and quality control is probably as good as it gets on a finite budget, as shown by the high price on the lenses. Lenses are meticulously controlled and rechecked until tolerances are as specified by Leica quality proceidures.

 

My point was that the resolution of the lenses and the M9 sensor is so high that it can not be realized unless camera body and all lenses are calibrated together.

 

DSLR manufacturers have realized that, and given the users a means of performing their own calibration of camera and lenses in a straightforward and easy way, without shipping every item to the factory for calibration.

 

As discussed on numerous other threads on this forum, I think Leica needs to rethink this, and come up with a way that it will be possible for the user to utilize the full potential of the wonderful expensive Leica glass.

 

Best regards

 

Trond

 

PS: The problem was at least as bad on my Nikon D700, Canon 5D2 and 1Ds3, in some cases much worse, as with the EF50mm 1.2L and 5D2, which barley could be calibrated sufficiently. Focus out of the box is all over the place, but it can be calibrated at home by the user.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[jaapv]

 

And how do you compare the point of highest resolution with the point of highest contrast if you're making such a calculation? I would be more willing to grant that "best focus" were centered between them.

 

If I recall optical theory, MTF and frequency response vs microcontrast correctly, those points are identical or should be on a properly corrected lens.

[Guest mc_k]

How did you figure the location of the sensor? Leica has only been able to give me an approximation, since the M8 (and I assume, the M9) doesn't have a focal plane mark.

...

 

I think he just meant he did the tests at 0.7m, 1m, etc.

 

If he's just checking the rangefinder, and not the scale, the test distances don't have to be exact.

 

To mark the sensor position, I think some people just use the flange-to-film standard from the other M cameras (sorry if my terminology is off) and measure accordingly.

[Guest mc_k]

...

I've never heard that before. I would think that "center of the depth of field" wouldn't even be calculable ad hoc, since DoF doesn't exist as a physical thing, but is created by the brain...Very interesting idea. Tell me more. Can you cite a source?

 

post was in a context of focus-testing by ruler, chart etc. When you (more or less subjectively) read the result, you either assign the plane of best focus and you're done, or the two depth-of-field limits. I'm saying it's conventional to take the average of the two DOF limits, which I think is what most people usually do.

 

If you want to calculate the near and far limit from formula, that's fine, but I don't think that is the custom, and I'm not sure you want to erode the custom.

[hoppyman]

Hi Trond. That was interesting to see your calibration device in action.

Focus calibration for AF cameras is very different of course, as are the mechanisms. For a start the tolerances are very likely higher and individual hand assembly does not happen. That's one of the things that contribute to the cost of Leica M's. That is a different topic for a different forum though.

I have watched the calibration being performed at Solms on their rig by a skilled technician. You can see aspects of the process in some publicly shared video from Solms if you would like to search.

As mentioned I was in Solms and had my M8 upgraded. Customer Service volunteered to calibrate my two lenses as well and I left them to be done later. (they were done free of charge and returned to me). I asked the very helpful technician about the processes. He explained that the M8 rangefinder and the sensor were checked and calibrated (since the camera was disassembled) as part of the upgrade and my camera was then very close to their benchmark ideal camera. This calibration is done with a standard lens that is known to be nominally perfect.They did not want to retain my M8 to match my lenses to it. If lenses were to be adjusted to one particular camera without that camera being calibrated you could clearly potentially end up with a mis-match in another combination.

 

I maintain that a 20mm error at one metre is definitely not acceptable in the most critical applications. Both my M8 and M9 do much better than that. This is from the APO 75 Summicron ASPH. at f/2.0 and minimum focus distance (.7m). Focus was on the stem (which puts the plane of best sharpness just behind the centre of the flower in this case).

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Guest trond]

Advertisement (gone after registration)

Dear Hoppyman,

 

I think we fully agree, we just speak a different "language".

 

I have taken my M9 to be the "standard" in this test, and assumed it is made according to Leica´s tolerance specification. The same goes for the 50mm Summilux.

 

Both camera and the lens was received at the same time from Leica, and both where factory sealed.

 

The test shows that both lens and camera are perfectly aligned.

 

So far so good.

 

Then, to the two Summilux 35 I tested in the same way. One brand new, the other three years old, but in "like new" condition.

 

Both perform differently, one front focus, the other, back focus.

 

Again, I assume that at least the new one to be according to Leica´s acceptable production tolerances. If not, the lens should never have left the factory, and Leica has a quality control problem.

 

As for the Summicron 50, "golden lens" used in the production calibration of M9 cameras. This lens has its own tolerances, as do the calibration process, the camera beeing calibrated, and the operator that performs the calibration.

 

They all have tolerances, nothing is perfect, no matter how long the operator takes to perform the calibration, in the end, the result will only be within certain limits, set by Leica as acceptable.

 

And I am sure Leica sets these limits as narrow as practically feasible in volume production.

 

I don´t know the exact number of these tolerances, I can only observe that I have one brand new M9 and one brand new Summilux 35 which are off by approx 20mm.

 

I assume both the M9 and the Summilux 35 to be according to Leica´s acceptable limits, or the lens should never have left the factory. Thus, the limit is probably in the range of +/- 10-20mm.

 

It is important to remember that the DOF of the M9/Summilux 35 at f1.4 and 1 meter distance, is just +/- 20mm.

 

This means that the focus error falls completely within the DOF of the combination, and can only be observed by looking at the location of the DOF in front of, and behind the subject

 

Now, let´s go out and take some photographs!

 

Best regards

 

Trond

[hoppyman]

I'm not explaining myself well then

 

Never mind. I look forward to some posted pictures.

[Guest trond]

Hoppyman, it is probably I that don´t explain myself well!

 

However, I an enjoying a week in San Francisco with my M9 and new WATE.

 

I am loving it!

 

Canon had to stay at home on this trip

 

All the best!

 

Trond

[Guest mc_k]

...This is from the APO 75 Summicron ASPH. at f/2.0 and minimum focus distance (.7m). Focus was on the stem (which puts the plane of best sharpness just behind the centre of the flower in this case).

 

If the camera and lens are adjusted to account for near and far limits of DOF, and you focused on the stem...best focus would be on the stem. If the DOF is averaged, best focus would be in front of the stem by an insignificant amount. In any case, in this example the difference is only 0.039mm (near limit is 694.812mm and far limit is 705.266mm; CoC=0.03mm).

 

I have always assumed that for practical reasons, when adjusting lenses/cameras, best focus is taken to be the midway point of the DOF. This is what I have heard second-hand about Leica practices. In most cases it wouldn't matter; in the rather extreme case of the poster's 21 at f/1.4, 1.0m, the difference between the two adjustments is about 9mm. So the question is not altogether unimportant.

[Guest trond]

If the camera and lens are adjusted to account for near and far limits of DOF, and you focused on the stem...best focus would be on the stem. If the DOF is averaged, best focus would be in front of the stem by an insignificant amount. In any case, in this example the difference is only 0.039mm (near limit is 694.812mm and far limit is 705.266mm; CoC=0.03mm)..

 

The CoC on the M9 should be set as 0.0084mm for B&W, or 0.017 for color details.

 

One time the pixel size (8.4 micron) or two times the pixel size, respectively, due to the interpolation of color information from the Bayer pattern.

 

The old rule of thumb of CoC of 0.03mm is equivalent to that of a 1 megapixel camera.

 

This rule of thumb is completely out of date in today's era of 20+ megapixel cameras.

 

Best regards

 

Trond

[Guest mc_k]

The argument runs the same for a 1mp camera or a 20mp camera.

 

You can calculate your own CoC value from the depth of field you observe and it will be just as "official" as the values quoted.

[jaapv]

No, because the value you observe is subjective and thus not fixed. Trond is talking about the native DOF of the sensor, which is determined by the pixel size, with an addition for colour. I doubt the difference because the Bayer interpolation will be the same regardless of working in B&W or colour as it will be written into the RAW file anyway. But it does mean that th COC must be twice the pixel size in all cases instead on the single value assumed in general.

[SJP]

Using the pixel pitch as the CoC criterian is wrong and confusing.

 

The pixel pitch governs the maximum resolution of the sensor and using 2x the pixel size is the Nyquist criterion for determining the minimum detail size that can be resolved by the senor. So far so good.

 

The Circle of Confusion criterion is a properly standardised (gentlemens) agreement on what a human eye can resolve when viewing a standard size image at a standard reading distance (distance = 1x the diagonal is good enough). Back calculating this to standard 35mm (135) film format then gives CoC of 30 micron. That has not changed in the digital age (by definition). What has changed is an increased desire for sharpness induced by pixel-peeping. This fine by me but it does not change the CoC definition, it only changes your own private criterion which is not standardised.

 

CoC does not depend on sensor resolution just like it does not depend on film grain & what ASA/ISO/DIN value you are using.

 

CoC does depend on sensor size, but that is a different story - see M8 subforum where this is discussed ad nauseum.

[SJP]

If the camera and lens are adjusted to account for near and far limits of DOF, and you focused on the stem...best focus would be on the stem. If the DOF is averaged, best focus would be in front of the stem by an insignificant amount. In any case, in this example the difference is only 0.039mm (near limit is 694.812mm and far limit is 705.266mm; CoC=0.03mm).

 

I have always assumed that for practical reasons, when adjusting lenses/cameras, best focus is taken to be the midway point of the DOF. This is what I have heard second-hand about Leica practices. In most cases it wouldn't matter; in the rather extreme case of the poster's 21 at f/1.4, 1.0m, the difference between the two adjustments is about 9mm. So the question is not altogether unimportant.

Agree, the depth of field is nearly symmetric for APO 75 Summicron ASPH. at f/2.0 and minimum focus distance (.7m) and somewhat asymmetric for 21 at f/1.4, 1.0m (giving 0.91 and 1.11 m respectively). However the point of highest sharpness remains on the plane of focus & I guess lenses are calibrated by determining that point.

[Guest trond]

Using the pixel pitch as the CoC criterian is wrong and confusing.

 

The pixel pitch governs the maximum resolution of the sensor and using 2x the pixel size is the Nyquist criterion for determining the minimum detail size that can be resolved by the senor. So far so good.

 

The Circle of Confusion criterion is a properly standardised (gentlemens) agreement on what a human eye can resolve when viewing a standard size image at a standard reading distance (distance = 1x the diagonal is good enough). Back calculating this to standard 35mm (135) film format then gives CoC of 30 micron. That has not changed in the digital age (by definition). What has changed is an increased desire for sharpness induced by pixel-peeping. This fine by me but it does not change the CoC definition, it only changes your own private criterion which is not standardised.

 

CoC does not depend on sensor resolution just like it does not depend on film grain & what ASA/ISO/DIN value you are using.

 

CoC does depend on sensor size, but that is a different story - see M8 subforum where this is discussed ad nauseum.

 

Dear SJP,

 

I am not debating the common definition of CoC.

 

The common definition is not scientifically very accurate, but guides what used to be acceptable sharpness in a 8X10 print at normal viewing distance for a person with normal eye sight.

 

I was discussion the maximal sharpness possible by Leica lenses and the M9 sensor, the Nyquist limit.

 

To achieve this, the lens needs to be focused so the CoC i equal to two times the pixel pitch.

 

I agree this is pixel peeping, but nevertheless, this is the maximul resolution possible, and it requires very accurate focusing.

 

In the case of the Summilux 50mm and the M9, my combo is actually able to reach this level of sharpness because the rangefinder and lens is perfectly matched.

 

While my Summilux 35 and APO-Summicron 90, can only reach this level of resolution by guessing the correct focus, due to miscalibration of the two lenses.

 

Best regards

 

Trond

[jaapv]

You beat me to it, Trond. Rarely I disagree with Stephen.After all, I would hate to be electrocuted The convention is not disputed. It simply exists and is adopted by the whole industry, whatever the rights and wrongs of it. But DOF is governed by enlargement, so it is perfectly valid to calculate a DOF at the maximum resolution (i.e. maximum enlargement) a sensor will allow. In fact, it is a more objective definition than one that is based on the average resolving power of the human eye, an enlargement to a 6x9 cm final print and a film of an ancient thickness. We only have to read the forum to see the confusion the old convention ( I hesitate to call it a definition) creates in the minds of photographers.

[Guest trond]

Yes, this can also easily be verified by taking a photo using the DOF scale on the lens, and I guarantee that the image will not look sharp!

 

The M9 and Leica glass is capable of shapness much grater than what is governed by the common definition of CoC of 1/30mm.

 

This definition was made long time ago in the era of poor film quality and uncoated lenses.

 

Norman Koren and Harold Merklinger has an excellent discussion of the topic in the links below:

 

Depth of field and diffraction

 

DOF Revisited

 

If the standard of sharpness is CoC of 30-35 micron, we don't need Leica glass, a cheap plastic lens can do it!

 

Best regards

 

Trond

[Guest mc_k]

No, because the value you observe is subjective and thus not fixed. Trond is talking about the native DOF of the sensor, which is determined by the pixel size, with an addition for colour...

 

Reading a ruler focus test is already subjective.

 

My point to the poster was to check the value himself and see if it is reasonable, instead of relying on whatever number is currently fashionable. The CoC value in context is just a constant in the depth-of-field equation, nothing more and nothing less (with certain given conditions.) It doesn't matter in the end how you choose it; what matters is how accurate a description it provides. But you are right of course on the usefulness of a fixed value.

[Advertisement]

Hallo,

Look here: Leica M

[SJP]

Dear SJP,

 

I am not debating the common definition of CoC.

 

The common -snip- due to miscalibration of the two lenses.

 

Best regards

 

Trond

Hi Trond,

 

OK if that is your aim it is a valid point. Are you stating that some lenses can surpass the sensor resolution while others can not? On paper this is only governed by aperture but lenses are not quite ideal in many cases due to chromatic abberation etc. The focal point of a decent lens can be close to the diffraction limit which would be a few micrometers at most at largish f/values. It would be interesting to know how close Leica glass is to the ideal.

 

best regards Stephen

 

Edit, concerning the remark "If the standard of sharpness is CoC of 30-35 micron, we don't need Leica glass, a cheap plastic lens can do it!" true on paper but Leica glass is not about absolute sharpness it is about the overall package, mico and macrocontrast, lack of abberations, lack of distortion, small size and whole pile of other subtleties. Making a sharp lens is easy enough, just like making 12% wine is easy enough, there are other more important aspects. Moreover a photo's is not better if it is sharper or at least not necessarily.