Focussing - when your lens hits 'infinity' what happens to the rangefinder?

[Julian Thompson]

Advertisement (gone after registration)

I am trying to understand this so please forgive me if this has been previously explained!

 

Let me set the scene - M8.2 viewfinder with 35mm summicron so you've got rangefinder focussing from 0.7m to 10m marked on the lens. Then, rotate the lens a touch more and you get the infinity mark.

 

So - OK - imagine several sailing ship masts in a line, with you stood at one end looking along them.

 

You focus on the nearest mast, at 2m - no problem, then the second mast, at 6m, twisting the barrel of the lens the rangefinder window 'virtual' mast moves in from the right hand side of the viewfinder towards the 'real' mast in the viewfinder. The third mast is at 10m - again as you rotate the barrel of the lens it continues to physically compress and get shorter.

 

The fourth mast is sat at 20m - so you twist the lens barrel more and hit the infinity stop.

 

The fifth mast is 30m.

 

The sixth at 40m.

 

The seventh at 50m.

 

The eighth mast at 200m.

 

The ninth mast way in the distance at 500m.

 

You can't twist your lens any more once you have focussed at mast 4.

 

When you look at mast 9 and probably 8 it is clear that the effective metering base of the camera means that the two masts will likely look pretty darned aligned.

 

BUT!

 

So - my question (finally!)

 

When I look at masts 5, 6, 7 - what should I see?

 

Should they line up perfectly?

[Nicoleica]

Hi,

 

The simple answer is yes. I'm sure that someone will be able to provide a far more complex and detailed answer, as there are a myriad of variables to consider, including the effects of temperature on the infinity position of the lens and rangefinder.

But, for all practical purposes, once you've reached the infinity stop, that's it.

[Advertisement]

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

[Julian Thompson]

Hmmn - thing is, they dont - at least on my camera!

 

When you think about it, the rangefinder makes a triangle I guess - so once you run out of focussing 'twist' that means the lens will now be at 'infinity' but the rangefinder window doesn't know this, and as a result I am not convinced that these masts (5/6/7) would perfectly align in the window. I can see that once you get to big distances like masts 8 and 9 then of course a rangefinder with a metering base of just a few inches and a human eye would not be able to work, but at these mid distances (30/40/50m) I can sort of see that it might be right to see the masts misaligned in the viewfinder but un-important because the lens is now at infinity and therefore the resulting focus will be as good as possible for that lens/aperture setting.

 

Does that make sense or am I in danger of turning into 'professor of complete nonsense' :-))

[SJP]

A rangefinder can focus with say 5 cm margin at 5 metres, that corresponds to an angle change of roughly 5/500 - 5/505 = 1/100 - 1/101 = (101-100)/(100*101) = 1E-4 = 0.0001 radians. Here I am assuming a 5 cm basis for triangulation. Note this is an amazing accuracy !

 

If the lens&rangefinder are set at infinite the rangefinder 'beams' are parallel, so you should be able to see a rangefinder offset if an object is at 0.05/0.0001 = 500 metres. None of the masts line up if the rangefinder is at infinite.

[Julian Thompson]

Thanks Stephen - but I do not understand how the rangefinder can be parallel at infinite?

 

There is a chimney here at just the exact distance at which the 35 'cron goes to 'infinite' - surely the fact that this lines up perfectly in the rangefinder means that there must be some triangulation going on?

 

...and that in my masts example, if the 'infinite' setting were indeed parallel then that by definition means that as I unwind the focus there would always be a 'setting' for every mast?

 

Can you elaborate on it for me so I can understand better?

 

Thanks for your time!

[SJP]

The rangefinder superimposes two images that are collected from two points on the base of a triangle (the baselength of the triangulation). On the camera these points are the viewfinder itself and the small square opening more to the centre of the camera, these images are superimposed in the "rangefinder patch".

 

If you focus the rangefinder on a close object the lines emanating from the base of the rangefinder cross each other at the focussing distance so for 1 metre you should imagine an equilateral triangle of 1 m high and 5 cm base as the path of the triangulation rays. Similarly for focus at 10 m, the triangle is 5 cm wide and 10 m high. In that case the angle in the upper corner becomes 0.05/10 = 0.005 radians which is already a very small angle, i.e. the two rays entering the rangefinder are nearly - but not quite - parallel.

 

As you extend the point of focus further and further away the trangulation rays become increaingly closer to being completely parallel, reaching that point when the focus is at infinity.

 

In your example if the lens and rangefinder are set to infinite then all the masts should not line up perfectly in the rangefinder, although 500m and infinity should be very barely distiguishable. Something at 1000 m is at infinity within the accuracy of the rangefinder.

Edited July 4, 2009 by SJP

[j. borger]

Advertisement (gone after registration)

In your example if the lens and rangefinder are set to infinite then all the masts should not line up perfectly in the rangefinder, although 500m and infinity should be very barely distiguishable. Something at 1000 m is at infinity within the accuracy of the rangefinder.

 

Interesting topic!

Well in my case i find 100m and infinity barely distinguistable with mu 50 lux asph.

Does this mean i have a problem with my lens or my eyes??

Consequence of this is that it is about impossible to focus a lens accurately wide open over longer distances. Even objects at 30m have its risks for misfocussing with for instance a 50 lux wide open.

Or is the length of the focusthrow of a lens also of relevance here??

My rigid rummicron had much longer focussthrow and was much better to focus on mid-distance subjects!

[jaapv]

Mathematics will teach you that true infinity does not exist so engineers will find their own approximation The whole system is calibrated so that infinity is 1000x the focal length of the lens.

A 50 mm lens should hit infinity in the viewfinder at 500m. Of course anything beyond that will have rangefinder patches that do not coincide, but that shift is so minute that it is invisible to the human eye and within the tolerance span of the system anyway. Having said that, a very small discrepancy at infinity does not affect focussing accuracy. Even temperature changes will bring that on.

With your lens you should be able to focus up to mast 9.

There is a thing about the rangefinder patch that is mentioned too seldom:

When you have lined up your patches, you are not yet at maximum focus.

When you have reached exact focus, the patch, which seemed lined up before, will suddenly jump into maximum contrast. That is your focus point.

Edited July 5, 2009 by jaapv

[Steve Ash]

Even with the new Noctilux wide open the dof is huge (85,64m- ∞) at infinity.

 

Regards

Steve

[odyocu]

Just a few thoughts:

 

Well in your specific case of a certain camera body and a certain lens, no body knows exactly what happens when the setting of your lens is at infinity point, because real world results overcome theoretical considerations. What matters is what you record on your sensor.

 

Firstly, 1000Xfocal length may as well be be 10000Xfocal length to have a 500m assumption.

 

In the old days, those who had to make their own rangefinder calibration, without proper calibration equipment, usually checked the result at night using the moon to guarantee sure that it is done correctly. Of course, the equipment (and the eye) wouldn't be able to differentiate beyond a certain point, but the above method was good enough. As good may be as the 10000X mentioned, or more.

 

In the real world, it is even more complicated. You have to first be sure about the distance between the lens mounting flange and the film surface (or sensor) to start with. Theoretically that's not perfect. Don't forget that the lens is not perfect either. Hopefully the mistakes may not add up. Or even in the case these add up, the engineering tolerances are small enough not to made the misfit obvious. Therefore, even with a reasonably good aligned viewfinder, the result may not be perfect on the film plane. (Note: My M8 had to be serviced for this distance between the CCD and the flange)

 

If there was a perfect camera and a perfect lens, at infinity setting of the lens, you should be getting a perfect picture of the objects at infinity position (however you may define infinity), which is rarely the case. Then comes the lens properties of back/front focusing, the focus points change with aparture, curvature of the image produced, and other lens defects. For example Hasselblad has a system in their H3DII cameras, which makes more precise images by adjusting for the aparture difference of the focus point of the lens, because it will be slightly different from the AF point calculated by the camera's AF system taken at full open aparture. The focus point is shifted as necessary just at the last moment before taking the picture.

 

To cut it short, engineeing tolerances chosen and the produced values become more important provided these are made reasonably well in the first place. This is why one lens performing excellent on one body show focusing problems in the other, if you are unlucky for having the tolerances adding up. Therefore, for a good product, most of the time the real world conditions may become more important that the theoretical considerations alone.

 

In the past, the precision in design and of manufacturing tolerances was the main reason differentiating an excellent camera like Leica from many others. At present, everything is much better fortunately.

[SJP]

Even with the new Noctilux wide open the dof is huge (85,64m- ∞) at infinity.

 

Regards

Steve

That is for a FF sensor, on the M8 it will be less, and if you do a 100%crop even less.

 

50 mm, f/0.95 at infinite, using my universal caculator

FF: 75m - ∞

M8: 100m - ∞

100% crop, M8: 200m - ∞

 

Where did you get the 85,64 - ∞ from?

 

Anyways, bottom line: in normal prints the difference between 500m and ∞ is of no importance, if you print on A0 or wallpaper size and use a loupe then they will be different. The point in the OT was however about the accuracy of the rangefinder itself, not on the lens you are using.

Edited July 5, 2009 by SJP

[el.nino]

1m = 100cm = 1000mm

 

1000 x 50mm = 50m !

[SJP]

(101-100) = 1, 100*101 = 1E4 (roughly) so it is 5E-2*1E4 = 5E2 = 500 m

 

qed

[vertekijker]

Mathematics will teach you that true infinity does not exist so engineers will find their own approximation The whole system is calibrated so that infinity is 1000x the focal length of the lens.

A 50 mm lens should hit infinity in the viewfinder at 500m. Of course anything beyond that will have rangefinder patches that do not coincide, but that shift is so minute that it is invisible to the human eye and within the tolerance span of the system anyway. Having said that, a very small discrepancy at infinity does not affect focussing accuracy. Even temperature changes will bring that on.

With your lens you should be able to focus up to mast 9.

There is a thing about the rangefinder patch that is mentioned too seldom:

When you have lined up your patches, you are not yet at maximum focus.

When you have reached exact focus, the patch, which seemed lined up before, will suddenly jump into maximum contrast. That is your focus point.

 

Very interesting and helpful, Jaap. A few years ago, I noticed the 'problem' when focusing as well, but as I couldn't see it in the pictures, I didn't bother. Still nice to know that it is completely normal.

Thank you,

Frans

[Steve Ash]

Where did you get the 85,64 - ∞ from?

 

I got it from the technical spec of the new noctilux which can be downloaded from Leica's website.

 

What I do not understand:

If I set the focusing distance at or beyond the hyperfocal distance anything should be within my dof at infinty. The only thing that will change is the starting point of the dof. So in my point of view the rangefinder accuracy at the infinty setting is of no meaning at all, isn't it?

 

Regards

Steve

Edited July 5, 2009 by Steve Ash

[SJP]

Ah, I will check to see what they are up to

 

It is true that by using a hyperfocal setting you can bring the closest "sharp" point foreward, under standard viewing conditions this looks fine but if you start pixel peeping the results may be less than desired. It is mainly something to decide for yourself.

 

As a firm believer in the empiricist school of natural philosphy I tried determining the rangefinder accuracy empirically and the difference between 100 m and ∞ is easily verified. I could not see the difference between 500m and ∞ but admittely my eyes are far from perfect.

Edited July 5, 2009 by SJP

[el.nino]

I got it from the technical spec of the new noctilux which can be downloaded from Leica's website.

 

What I do not understand:

If I set the focusing distance at or beyond the hyperfocal distance anything should be within my dof at infinty. The only thing that will change is the starting point of the dof. So in my point of view the rangefinder accuracy at the infinty setting is of no meaning at all, isn't it?

 

Regards

Steve

 

In theory you are right, BUT:

 

1. dof- and hyperfocal-concepts are not made for infinity!

 

2. there is no sharpnes in dof. the whole dof-concept only talks about ACCEPTABLE sharpness. and details at infinity (e.g. leafs on a tree) are so small that "acceptable sharpness" just is not good enough. that only works for not so small details in the foreground.

 

3. if you want any sharpness at infinity you have to focus on infinity (on a correct calibrated lens you have infinity at the infinity-stop - so no need to focus with the rangefinder)!

 

further reading:

Digital Focusing Part Two

http://www.trenholm.org/hmmerk/TIAOOFe.pdf

[SJP]

1. dof- and hyperfocal-concepts are not made for infinity!

If you actually read the Merklinger book (reference 2) you will find that they are, in fact that is how the hyperfocal distance is defined, see page 14 and 15. If D=H then the back DOF is at infinity. Merklinger makes the point that you can use a different criterion for trading off near and farfield "blur" that he finds more pleasing esthetically. The traditional approach nevertheless remains valid and useful.

 

I agree that objects at infinity are not sharp enough using the standard hyperfocal setting.

Edited July 5, 2009 by SJP

[el.nino]

The traditional approach nevertheless remains valid and useful.

 

yes, i did read it and of course the traditional approach is valid at infinity, but once you adjust your CoC to the needs of very fine resultion you automatically come to the infinity-only-approach. your traditional dof-calculator will tell you to focus at e.g. 200m (with a 50mm-lens) and since infinity starts at about 50-100m (with this focallength) you already are in the infinity-area.

 

Focallength: 50mm

f-Stop: f8

 

Subject distance 10 m

 

Depth of field

Near limit 9.69 m

Far limit 10.3 m

Total 0.64 m

 

In front of subject 0.31 m (48%)

Behind subject 0.33 m (52%)

 

Hyperfocal distance 312.6 m

Circle of confusion 0.001 mm

just an example with a custom (very strict) CoC.

[Advertisement]

Hallo,

Look here: Leica M

[jaapv]

Correct . Dropped a zero. Doesn't change the argument,