M8 focus shifts and "back focus"

[lars_bergquist]

Advertisement (gone after registration)

Here we go again. Proper experimental procedure calls for a series of data points, so that we will not be led astray by a random sampling error. So we should do at least five exposures, refocusing for each one of them, and then evaluate them all. (This by the way makes the tripod irrelevant—you can handhold.)

 

I did run that test on my v. 4 35 mm Summicron. At .8 m, I found that wide open, the lens placed best focus about 10 mm further away from the camera. The focus point, in my estimate, was still within acceptable d.o.f. When stopping down to 5.6, d.o.f. increased more to the rear than in front, but the focus point was still acceptably sharp. So there was some focus shift, in addition to a basic focusing 'error' (if it was an error; it may well be that arcane things are happening in Solms, and have happened in Wetzlar).

 

Testing my 50 mm Summilux ASPH I found no focusing error wide open at the same distance, and no focus shift when stopping down to 4.

 

Just for the hell of it I then tested my CV 75 mm Heliar. This wide open had about 15 mm back focus (away from the camera) at c. 1.1 m. Stopping down to 5.6 the plane of best focus did not shift.

 

Provisional conclusion: Focus shifts and 'focus errors' exist, but Leica lenses keep within acceptable working tolerances. Perhaps we should be more concerned with practical results than with monitor-gazing.

 

It may well be that M focusing has always been a kind of working compromise, just as when a piano tuner distributes theoretical tuning errors across the entire keyboard, so that we can play not just in one key (say, C sharp) with audibly false tuning at all other keys) but in all keys. Just as with depth of field: only the plane of best focus is maximally sharp, but some things in front of or behind that plane may also be acceptably sharp. We have accepted that reasoning for more than a century now.

 

The old man from the Age of Manual Focusing

[wlaidlaw]

I would greatly appreciate if someone else with a Noctilux could try the following. I checked infinity focus and it is spot on, both with the rangefinder image coincidence and images taken with the lens. If you use the rangefinder to focus on an object 7.5M away, the lens reads 10M and the image is not sharp. If you reset the lens to 7.5M on the focus scale, the image is sharp. The M8 body functions well with other lenses e.g. Elmarit-M 90mm. At infinity and 7.5M, the Elmarit focuses correctly with the rangefinder and gives the correct read out on the focus scale - images are sharp. The Nocitlux was at Leica in December for coding and a check over it is a late one, serial 3814220.

 

Wilson

[carstenw]

Wilson, does it focus okay at the near limit? Do you see any unusual wear marks on the focusing cam on the back of the lens?

[wlaidlaw]

Wilson, does it focus okay at the near limit? Do you see any unusual wear marks on the focusing cam on the back of the lens?

 

Carsten,

 

There are no marks at all on the Noctilux focusing cam it, like the rest of the lens it is pristine.

 

Close focus it is all out a bit. If you set the focus scale to 1M at a measured distance from the target to where I estimate the sensor is in the body (why no focus plane mark on the body?) the image is front focused (clearer near to the camera than further away) by about 30-40mm. If you use the rangefinder, the focus scale reads about 1.05M but the actual image focus is better with now back focus of about 10-20mm - hmmmmm........

 

Wilson

[carstenw]

With an infinity focus which works, and a consistent back-focus in front of that, it almost seems that the focusing cam has the wrong shape. Weird. I have not heard of this problem before.

[wlaidlaw]

With an infinity focus which works, and a consistent back-focus in front of that, it almost seems that the focusing cam has the wrong shape. Weird. I have not heard of this problem before.

 

Perhaps this is just another Leica lens characteristic with which we have to learn to live. The middle distance rangefinder inaccuracy is a surprise. It is way beyond operator error margins.

 

Wilson

[ho_co]

Advertisement (gone after registration)

Peter--

See what your seemingly innocuous question started?

 

There is one more thing to keep in mind, and that is that the lenses' intended output isn't to the screen but to a print.

 

In other words, do the effects you mentioned show up when printed?

 

And that encompasses two aspects: First, you're not going to be printing tape-measure shots. And second, you don't have a mouse in your hand when you look at a print--i.e., you are less likely to be pixel-peeping.

 

 

@ Wilson--I thought you were on the 'do not disclose' list for the Higgs?

 

--HC

[lars_bergquist]

Peter--

See what your seemingly innocuous question started?

 

There is one more thing to keep in mind, and that is that the lenses' intended output isn't to the screen but to a print.

 

In other words, do the effects you mentioned show up when printed?

 

---------------------

 

--HC

 

Is the output by printing alone? Since 1936, when Kodachrome was introduced, output has also been to the projection screen. There are still people who use their Kodak Carousels. Today, it is print – and the electronic screen. Both are legitimate.

 

The old man from the Age of Brovira

[ho_co]

Thanks for the correction, Lars. You're right of course, but you took me literally. What I meant was: Does one see the 'problem' in one's chosen form of output?

 

I didn't see it looking at slides (using Pradovits, not US-made Carousels --the European versions were better).

 

I didn't see it looking at prints from negatives.

 

Now I see it on-screen in Photoshop. But I need to ask further:

 

Do I see it on my home page?

 

Do I see it in my inkjet prints?

 

Do I see it when professionally printed?

 

Do I see it when projected with a good electronic projector?

 

My post was oversimplified for sake of brevity; I wanted simply to amplify what others have said, viz that even though we are now able to see the actual behavior of the lens very easily, the end output may not show it any more now than previously.

 

--HC

[ho_co]

Consider how important Wilson's earlier point is: We've eliminated the enlarging lens.

 

Looking only at resolution: the resolution of a system is given by:

 

(1/system resolution) = (1/lens resolution) + (1/film resolution) + (1/enlarger resolution) + (1/paper resolution)

 

That is, even if all items in the chain are of identical resolution, the system resolution will be lower. On the other hand, eliminating one element of the chain will cause a noticeable jump in the final output resolution.

 

That may be one reason we're seeing details we missed before with film: We've eliminated a smudge factor.

 

The formula is for classical optics. For digital it will need some changes:

 

When viewing the file in Photoshop, the monitor resolution will be the final element, replacing both enlarger lens and paper. (As Guy said, we didn't have 100x loupes with film.)

 

When printing, inkjets replace the enlarging lens--so we're looking at dpi and droplet size and density. We still have the paper, but now it has characteristics of ink absorption and spread. How do we accommodate these changes?

 

When projecting, we still need to consider the resolution both of the screen and of the projection lens; but we must consider in addition the resolution of the image-producing unit of the projector: LCD, DLP, or liquid crystal (D-ILA or LCOS). How does the formula fit here?

 

Anybody have any ideas for approaching inkjets and inkjet paper in this formula? Clearly we're not worried about Airy disks any more.

 

--HC

[colorflow]

Consider how important Wilson's earlier point is: We've eliminated the enlarging lens.

 

....

Anybody have any ideas for approaching inkjets and inkjet paper in this formula? Clearly we're not worried about Airy disks any more.

 

--HC

 

My tcw - In spite of finding focus shift in everyone of my 10 lenses tested, both digital and on film, I really don't think it is a problem in real-life situations for all but perhaps two of them ( the Lux 50 asph is marginal and the Lux 50 v2 is off from last service). I can't see the shift of the others in print or on my monitor without blowing up 1:1. So as you commented before, this characteristic (I am not saying this is a problem or issue) has always been there. There has always been a range of tolerance between the cam profiles of individual lenses and the mechanics of the rangefinder in a particular camera. I saw the same behavior on my M6. Occasionally a lens may be out of whack for whatever reason and need to be adjusted. Our digital systems are becoming more and more sophisticated and critical but the same is not reflected in how we use them in real-life situations. Not yet anyway.

 

The one lesson I did learn is that it is preferrable to have a lens at the verge of front-focusing wide open. Then the backward focus shift when stopping down along with the increasing DOF will bring it into focus. This is not relevant for wider lenses of course.

 

Alan

[pklein]

Peter--

See what your seemingly innocuous question started?

--HC

 

Uh-oh, I've created a monster. "Quick, Igor, the kites!"

 

Seriously, most of what I've found is probably in the realm of pixel-peeping, although I need to shoot my 50mm lenses more to see if it's a problem. I tested two ways:

 

1. Handheld, refocusing with each shot, and taking several shots at each aperture (the Lars method).

 

2. On a tripod, focusing once and observing how the focus shifted as I stopped down without refocusing. This gave me a relative indication of wide open vs. stopped down.

 

I pretty much got the same results with multiple shots using each method. The focus did indeed shift back when stopping down. But even though the point the RF was focused upon was sometimes just outside the near limit of the zone of absolute sharpest focus at middle stops, it was usually only detectable at 1:1. At 2:1, it seemed just inside the zone of acceptable focus. I've found that 2:1 on the screengives a good indication of how a print in the 8x10 - 11x14 range will look, so I'm not too worried.

 

What all this testing does tell me is that I have much more margin of error behind the subject than in front of it, more than the classic 1/3:2/3 ratio many of us learned. So I might focus accordingly at mid-stops. I'll have to see how much this all applies in real shooting.

 

Just so you all know I'm not making this stuff up, here are two representative 100% crops shot with the 35/1.4 Lux ASPH. They were rotated 90% for best display in this format, but apart from that, they are just as they came out of the camera at default settings. I focused on the 28 inch mark, and the camera is to the left of the picture, so the higher the numbers, the farther away it is. These two were handheld and refocused for each shot (they match other, tripod-based shots).

 

The first crop is at f/1.4, the second is at f/4.

 

--Peter

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!

[pklein]

Here's another pair of 100% crops, these taken with a tabbed 50/2 Summicron. These were taken on a tripod, not refocused between shots. Again, focused on the 28" mark. the first picture is at f/2, the second at f/4.

 

--Peter

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!

[wlaidlaw]

If I might be so bold as to sum up much of what has gone before. There are two basic situations of using lenses with digital bodies. One we can live with, the other requires either the lens to be serviced/adjusted or replaced. The following assumes that the rangefinder and linkage in the body itself, are correctly adjusted at infinity, middle and near distances. Situation one is where the focus point remains sharp within the defined COC field of focus at all distances/apertures but the distribution of the depth of focus changes either side of the focus point, as distance or aperture is varied. Situation two is where the rangefinder point of focus is unacceptably out of focus at any selected distance or aperture. Situation one we can learn to live with. Situation two you can't and the lens has to be adjusted or changed.

 

Sadly my Noctilux appears to be situation two and will have to go for another holiday in Solms. I tried playing around with the focus cam in the body yesterday and could either get it right at middle distances but wrong at infinity or vice versa but not both right. When it was right at infinity, that was also the correct cam setting for my Elmarit-M 90 at both infinity and middle distances.

 

Wilson

[lars_bergquist]

----------------------------

Anybody have any ideas for approaching inkjets and inkjet paper in this formula? Clearly we're not worried about Airy disks any more.

 

--HC

 

You are of course perfectly right, and not only in the quoted paragraph! It is largely a new game. This is a point I have made elsewhere: Why are we complaining so much all of a sudden? It is not that Leica lenses have suddenly deteriorated. It is that our eyes have suddenly become much sharper.

 

It is indeed possible that we are actually seeing too much. I have argued before that an A4 or 18 x 24 cm or 8 x 10' print is the realistic criterion in what may be called 'photography for pictures'. This is all we can take in with one look at a normal minimum wiewing or reading distance. Enlarge more, and any normal person (not a pervert who hunts grain or pixels or noise or unsharpness by any means available, and doesn't give a damn about the picture) will try to increase the viewing distance. Meaning that the eye's resolution – not in arc seconds but in microns on the subject – decreases in proportion. So what you don't see in that print, you won't see at all. Except when we use that electronic '100 x loupe' that Guy referred to. But that kind of view is irrelevant to anybody who just wants to enjoy a photograph – on paper, or on a screen.

 

It seems already to be the case that with some lenses, we can see smaller detail in the subject, on paper etc. etc., than we can see with our unaided eyes (or with our specs) before the subject itself. This even if the print does not show the subject larger than we see it with those eyes – as a long tele would normally do. I am specifically thinking of the 90 mm Apo-Summicron.

 

The situation is of course different if we are doing photography for scientific or technical registration, as in photomicrography for instance. There, every pixel counts. But cameras and lenses for general photography are seldom used for such applications nowadays.

 

Even in microphotography, and in microscopy in general, there is something called 'empty magnification'. This is what you get when the magnification is greater than the resolution of the microscope objective: the image gets bigger, but it ceases to show more detail. What the onlooker sees is the criterion.

 

The old man from the Age of the Monocular Microscope

[wlaidlaw]

I have argued before that an A4 or 18 x 24 cm or 8 x 10' print is the realistic criterion in what may be called 'photography for pictures'. This is all we can take in with one look at a normal minimum wiewing or reading distance. Enlarge more, and any normal person (not a pervert who hunts grain or pixels or noise or unsharpness by any means available, and doesn't give a damn about the picture) will try to increase the viewing distance. Meaning that the eye's resolution – not in arc seconds but in microns on the subject – decreases in proportion. So what you don't see in that print, you won't see at all. Except when we use that electronic '100 x loupe' that Guy referred to. But that kind of view is irrelevant to anybody who just wants to enjoy a photograph – on paper, or on a screen.

 

 

The old man from the Age of the Monocular Microscope

 

Lars,

 

I am sorry I am going to completely disagree with you. For photos I want to hang on my walls, I would regard A3+ as the minimum size and they are about right for viewing from 2-3 meters or so. If you go to most photographic exhibitions, the prints are A3+ or bigger. I would assume this was the reason that medium and large format was popular with photographers who exhibited before the arrival of high megapixel digitals, because of its suitability for enlargement to that size.

 

Since I got my HP 9180, I have not printed a single A4 photo on it. If the photo is sharp from the M8, A3+ is no problem at all. I took a picture in Barcelona which had a three masted ship a considerable distance in the background. Printed at A3+, you could cut yourself on the rigging, it is so sharp. Taken with a Biogon 35mm, bi-cubic upscaled to 13' x 19" then printed at 300dpi.

 

Wilson

[ho_co]

It is indeed possible that we are actually seeing too much. I have argued before that an A4 or 18 x 24 cm or 8 x 10' print is the realistic criterion in what may be called 'photography for pictures'. This is all we can take in with one look at a normal minimum wiewing or reading distance. Enlarge more, and any normal person ... will try to increase the viewing distance.

I would regard A3+ as the minimum size and they are about right for viewing from 2-3 meters or so.

Wilson--You and Lars seem actually to be in agreement: We view at 8x10" prints from about twelve to eighteen inches. We stand further back with larger prints. We may view 4x5" prints from only eight to ten inches.

 

Leica said when they introduced the DMR that around 10MP was all that was needed to realize the full benefit of Leica R lenses. The same likely holds with the M lenses.

 

With the M8's sensor, we are probably nearing the lenses' limits. The lenses are just as good as they were on film, but we now have the ability to inspect with Guy's 100x loupe.

 

We are now seeing into the design of the lenses--how the depth of field changes, how the focus plane shifts as we stop down. These effects were always there, but simply part of the image when viewed at a reasonable distance (depending on final image size).

 

As Lars says, we are in many cases being unreasonable to try to judge images at the pixel level onscreen, because we will seldom do that when viewing the print.

 

We may stand ten inches from a large painting to view brushstrokes, or three feet from a digitally projected image to evaluate the pixel patterns--but that is a technical exercise from which we can learn something.

 

To feel something, we view the painting, screen, print as a whole, just as you said--from a greater distance the larger it is.

 

--HC

[billh]

Lars,

 

I am sorry I am going to completely disagree with you. For photos I want to hang on my walls, I would regard A3+ as the minimum size and they are about right for viewing from 2-3 meters or so. If you go to most photographic exhibitions, the prints are A3+ or bigger. I would assume this was the reason that medium and large format was popular with photographers who exhibited before the arrival of high megapixel digitals, because of its suitability for enlargement to that size.

 

Since I got my HP 9180, I have not printed a single A4 photo on it. If the photo is sharp from the M8, A3+ is no problem at all. I took a picture in Barcelona which had a three masted ship a considerable distance in the background. Printed at A3+, you could cut yourself on the rigging, it is so sharp. Taken with a Biogon 35mm, bi-cubic upscaled to 13' x 19" then printed at 300dpi.

 

Wilson

 

Wilson,

 

I agree with you. If I focus on an eye, regardless whether the eye is in a human, dog or bird, and the plane of focus is elsewhere, the photo is not usable. When I use my SLRs or M8 with the APO75, the eyes are almost always perfectly focused. With the 50ASPH, the ears are clearly the sharpest part of the image. This is equally obvious in an 8 x 11 or a 17 x 24 inch print. Perhaps if someone is shooting landscapes or other scenes where the precise location of the plane of focus is not critical, then this is not an issue for the photographer. For those of us typically using 50mm or longer lenses in lower light situations, a focus error of an inch or more at 4 or 5 feet is glaringly obvious and incredibly annoying.

 

This afternoon I made slight counterclockwise adjustments in the rangefinder infinity adjustment, and it clearly improved the focus accuracy of the 50ASPH at a meter - but I am not sure it did not throw some of the other lenses off (the APO75, for example). I will have to do a bit more work tomorrow to see for sure. The moon is in focus with the APO75 and APO90 when the lens is set at the infinity mark - however, the rangerfinder shows the moon as being out of focus. When I focus using the rangefinder, the moon is out of focus, but knowing this is the case, I can live with setting the lens at the infinity mark for distance photography - it is the closer focusing accuracy that worries me. I am also concerned that achieving correct focus in one lens will mean another lens is not focusing accurately. Then what?

 

I wonder why so many M8s are leaving the factory with gross focusing errors? This is my second M8, and both have had this problem.

 

Bill

[wlaidlaw]

Wilson,

 

I am also concerned that achieving correct focus in one lens will mean another lens is not focusing accurately. Then what?

 

I wonder why so many M8s are leaving the factory with gross focusing errors? This is my second M8, and both have had this problem.

 

Bill

 

Bill,

 

I too have become increasingly concerned over the M8 rangefinder since getting my Nocitlux and finding the weird problems it has. I feel that it is probably the lens but I still have a niggling doubt that it could be the body. Further given all the difficulties various people have had with other lenses I think an in depth investigation needs to take place.

 

I think you would need a low mileage M7, MP or maybe even better a Zeiss Ikon with its longer rangefinder base plus 1:1 VF and some gigabit orthographic film. You would then take a selection of lenses and test them at various distances with an MTF type chart, first focused with the rangefinder and then a second series, scale focused . Then perform the same exercise with the M8. If a lens is behaving properly with the film body but not with the M8, it might highlight an issue with the M8 rangefinder. I feel this might be one for Sean.

 

It may be in the rush to get these out of the factory gates, QC fell somewhat below Leica's usual high standards. I know for example, due to the very early serial number of mine, although bought in January (Leica have confirmed that it has the later sensor), that it must have been a return that was re-manufactured. I would guess that this was limited to the sensor/electronics and the mechanics would have been left alone.

 

Wilson

[Advertisement]

Hallo,

Look here: Leica M

[lars_bergquist]

Lars,

 

I am sorry I am going to completely disagree with you. For photos I want to hang on my walls, I would regard A3+ as the minimum size and they are about right for viewing from 2-3 meters or so. If you go to most photographic exhibitions, the prints are A3+ or bigger. I would assume this was the reason that medium and large format was popular with photographers who exhibited before the arrival of high megapixel digitals, because of its suitability for enlargement to that size.

 

Since I got my HP 9180, I have not printed a single A4 photo on it. If the photo is sharp from the M8, A3+ is no problem at all. I took a picture in Barcelona which had a three masted ship a considerable distance in the background. Printed at A3+, you could cut yourself on the rigging, it is so sharp. Taken with a Biogon 35mm, bi-cubic upscaled to 13' x 19" then printed at 300dpi.

 

Wilson

 

Wilson, I'm afraid that you have completely misunderstood me. We do of course make larger prints, at least occasionally. The point however is this: We do spontaneously avoid looking at those larger prints from the same distance we would use if we were scrutinizing a hand-held A4 or 8 x 10", which would probably be some 30 cm or 12 inches. Otherwise, some parts of the picture will increasingly fall outside our peripheral vision. But as we increase the viewing distance, our ability to discern detail decreases in proportion.

 

At 25 or 30 cm, a human eye can resolve two lines around 1/10th of a millimeter apart. If you increase the distance to 60 cm, can you still resolve those lines? No. You can only resolve lines 2/10ths = 1/5th apart. Remember, human visual acuity is measured in measures of arc (seconds of arc), not in millimeters or inches. So, while our visual acuity remains the same, our ability to discern detail (grain, unsharpness) is in inverse proportion to the viewing distance.

 

Three black-and-white enlargements hang on my study wall. Two, made on 30 x 40 cm paper, were made from 6 x 6 medium format, and they are of course tack sharp. (The camera was a 1957 Zeiss Super Ikonta with a Tessar lens, which I used quite a lot up in Lapland.) The third one was made on 35 mm (Olympus OM with 135 mm/2.8) and this was done on 40 x 50 paper. But—when viewed from a normal distance, one where that 50 cm print subtends the same angle as the 24 cm print at normal reading distance, that print does look as sharp and grainless as the medium format pictures. I can crawl over my printer to decrease the viewing distance to 30 cm, and then I would of course see the grain, BUT then I would no longer see the picture of those stately vertical cliffs descending into the Baltic Sea.

 

And you can bet that I could enlarge that neg to two by three feet, or even twenty by thirty, and it would STILL LOOK SHARP becouse you would want to view the print at a correspondingly larger distance; i.e. as long as you were interested in the picture of the subject, and not in grain-peeping. That El-Nikkor is sharp ... Ergo, if it is sharp at A4, it is SHARP.

 

And, Wilson, you say that you view pictures on your walls from 2–3 meters. Now, if you can resolve 1/10th of a millimeter from a distance of 1/3d of a meter, then from 3 meters you can resolve at a maximum one (1) millimeter or 1/25th of an inch. Q.E.D.

 

Best regards from the old man from the Age of the Cooke Triplet.