M8-Focus on focussing

[lct]

Advertisement (gone after registration)

Japp, the formula you're after if perhaps b' = (e * f^2) / (k * z) where b' is the base length, e the visual acuity (0.0003mm at approx. 1 arcmin), f the focal length, k the aperture and z the circle of confusion (CoC).

If you choose a CoC value of 0.023mm for the M8 by example, this would make a base length of about 60mm to focus accurately a 135 lens at f/4. Smaller CoC values would lead to higher figures on the grounds of the same formula. But again i may be wrong as i'm no techie at all.

[jonoslack]

If you choose a CoC value of 0.023mm for the M8 by example, this would make a base length of about 60mm to focus accurately a 135 lens at f/4. Smaller CoC values would lead to higher figures on the grounds of the same formula. But again i may be wrong as i'm no techie at all.

 

sounds like it:)

[Advertisement]

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

[Rolo]

The point about focus-recompose is an important one, Rolo.

 

I agree.

 

A couple years ago I did the maths and TBH the difference in focal plane didn't seem to amount to much at all in terms of fractions of an inch.

 

I suspect that, when you recompose there's an opportunity to move the camera a fraction, or the subject to move. Wide open at minimum close distance it all matters a great deal.

[lxlim]

Interesting post Jaap. I enjoyed reading it.

 

In my experience, critical focus from a high performance DSLR AF system is much easier to operate in a dynamic environment than the RF. It's easy to split a vertical of a building with an RF for a super sharp image, but follow a human subject through six hours of a wedding, or street work, and those verticals are difficult to find.

 

No doubt, it's for the reasons you mentioned and one you didn't - speed of operation. An RF takes time to focus (presume there is no room for zone focus in critical focus situations) and that time slows the work down.

 

I take thousands of photos of people in a month and would love to switch to RF for all it's other advantages, but easily pinning down critical focus won't allow the switch.

 

As evidence, I've many hundreds of images I could post where the eyelashes, or even the veins in the eye can be clearly seen, not in a single photo, but in a sequence of ten different images, say, in a four minute window.

 

I'm not suggesting that all DSLR AF systems are equal, but mine has a separate computer dedicated to the AF system that will track minute movements in the subject, or the shooter. It's absolutely accurate, very quick, quite reliable and works well in low light. I'm confident that RF focussing can get nowhere near it in the applications I use it for and that's a real pain because I'd love to dispose of a complete system.

 

I cherish my 3 M's, but if I need critical focus, it's not the right camera for me to take .

 

I want to learn more!

 

I use a 1DmkIII which works very well low but contrasty light (ISO 3200 - 6400 at about f2 1/250s) but nailing the focus in flat shadowy low light is pretty much 80-90% of the time only. It takes about 3/4 to 1 seconds to get it right but not always. Sometimes thats too long. Am I being too demanding in these situations.

 

I find the converse is true in some situations, RF provides better critical reliable focus and much more quickly than my DSLRs. I glad to have both, so I have a choice.

Edited August 18, 2009 by lxlim
Added details

[Rolo]

I want to learn more!

 

I use a 1DmkIII which works very well low but contrasty light (ISO 3200 - 6400 at about f2 1/250s) but nailing the focus in flat shadowy low light is pretty much 80-90% of the time only. It takes about 3/4 to 1 seconds to get it right but not always. Sometimes thats too long. Am I being too demanding in these situations.

 

I find the converse is true in some situations, RF provides better critical reliable focus and much more quickly than my DSLRs. I glad to have both, so I have a choice.

 

I'm using the 1D MkII N and have less high ISO capability and I stay below 1250, but I'll go down to 1/60s so my light conditions are not too far away from yours. I generally shoot on Ai Servo focus, but when I get down to these levels I'll switch to standard. I'll add a ttl flash at this level if at all possible as I'm doubtful if many pictures I'd shoot at these levels of illumination are actually worth having. If the face is grey & flat - why shoot it ?

 

Ai Servo does take a moment to lock-on in low light, but usually my subjects are not moving fast, otherwise a flash would be pretty essential. Not sure if it's 3/4 second to focus, but it maybe. That would compare with several seconds with the M I suspect.

 

I have an MP, M7 and an M8 plus six Leica lenses, so I'm well committed.

 

Where do you find that the M's can outpace the 1D ?

[jaapv]

Well, I won't compare to any AF, but I would say light would have to be pretty desperate to need 3-4 seconds to focus. In all circumstances where I can still walk without tripping I would say focussing should be well within the second, and normally just one twist of the focus ring. The culprit - and I am not meaning you, Rolo,as an old RF hand is often that people start to "hunt" SLR fashion when the going gets tough.

[markowich]

Advertisement (gone after registration)

Since M went digital we have had dozens of threads on rangefinder accuracy, diopters, magnifiers, etc., undoubtedly set off by the pixel-peeping possibilities. There have been some disinformative opinions on RF accuracy, recently there have been doubts about the suitability of the venerable mechansm for full format sensors, a real snakepit without hard facts.

So here are the numbers:

 

Focussing acuracy is in any system is determined by triangulation. The measuring base is the determining parameter for accuracy.

On a rangefinder it is quite simple. The measuring base is the distance between the windows of the camera, multiplied by the magnification.

The values are:

 

M8 47.1 mm

0.72 VF 49.9 mm

0.58 VF 40.2 mm

 

On an SLR it is different. The measuring base is determined mainly by the focal length.

The values are (for an R series camera):

 

21 mm 1.63 mm

35 mm 4.43 mm

50 mm 9.82 mm

90 mm 29.20 mm

135mm 65.70 mm

 

The cross-over point between the systems is approx. 110 mm. (0.72)

 

 

Now for magnifiers:

M8: 1.25x = 58.4 mm

1.4x = 65.9 mm

 

0.72: 1.25x = 62.0 mm

1.4x = 79.8 mm

 

This clearly shows that there is no reason to change the rangefinder mechanism for anything "better" to compete with a DSLR. At 35 mm it is a full magnitude more accurate.

 

Often the tolerances in the roller/cam mechanism are quoted as a drawback. That may be true in an absolute sense, but it must be offset against the tolerances in mirror placement and screen positioning in an SLR. I would be hard pushed to say which system introduces more inaccuracy.

Focus shift is just as disturbing on an SLR as on an RF because an SLR will measure the distance wide open.

 

Now to Autofocus.

The same remark about tolerances in the mirror and AF sensor apply, added to which the AF motor in the lens adds more inaccuracy. The accuracy is measured at optimal contrast and falls below the factory specification at low contrast and/or low light.

 

The best AF systems have a tolerance of "within 1/3rd DOF".

That means on an 1.4 50 mm lens plus or minus 5 cm at 3 meter.

For an f 4.0 zoom at 50 it would be a whopping plus or minus 15 cm at 3 m.

Values that would have us return our M8 to Solms with the first UPS and set off an angry thread here.

 

The conclusions are:

 

A. If you feel you M8 is less accurate at 90 mm or lower than your SLR there are just three possibilities:

1. the system is out of calibration and needs to be adjusted

2. there is an eyesight problem. Consider a diopter.

3. (sorry) simple user error.

 

B. The M8 is acurate enough to beat any other system.

 

C. A full-format DRF with a 0.72 viewfinder will be more accurate than the M8 - no need to change anything in the rangefinder

 

D. To equal SLR and AF accuracy on a 135 lens one may need a magnifier. 1.25x on an 0.72 viewfinder, 1.4x on an M8.

 

E. The optical RF is not stressed by modern technology and does not need to be improved as far as measuring ability is concerned.

 

source for the numbers: Günther Osterloh - Leica M - Advanced Photo School and Johnathan Eastland - Leica M Compendium

 

interesting and accurate analysis for focusing a still object with nice vertical high contrast lines under the best possible conditions. unfortunately completely useless for 95% of all practical cases, where

 

1)the objects moves (even slowly is enough to offset the comparism)

2) you do not find good enough verticals

3)your eyes get tired after hours of shooting

.......

 

i just bought a nikon D5000, which is a prosumer DSLR with a rather lowly autofocus engine. and actually, it is pretty hard to get out-of-focus files out of it.

but scientifically speaking you (or rather your source) are (is) right.

peter

[jaapv]

Misconception: If you are focussing by lines only you are using less than half of the potential of a rangefinder. Coincidence focussing needs no more than a structure- any structure. Contrast focussing even less, although it is the most accurate way. Moving subjects? It works just as well - it is just a matter of learning the technique, as are many skills.

[Rolo]

The culprit - and I am not meaning you, Rolo,as an old RF hand is often that people start to "hunt" SLR fashion when the going gets tough.

 

Jaap, shouldn't make any assumptions about me () on this subject because I've had a hell of a struggle with M8 focus. I might have sorted it now with an Allen key, but not 100% sure. It might also be an eyesight issue as I tend not to use glasses and perhaps I should.

 

Bearing in mind I'm very aware of the risk of error, I do take more than the minimum amount of time in low light and wide open, but it's mainly due to having a suitable spot to focus on.

 

I've also noted it's easier to hit the target with my 24/50/'s/90mm lenses than with the 35/75mm Lux's when under pressure. All of them are fine in good light with zero pressure, but that's not what I do.

 

As an aside, I suspect Auto-ISO has been a big contributor to my recent focus experience. I've abandoned it for the time being. My Megapearls x1.4 magnifier helps in good light, but may be robbing the viewfinder of necessary light in dim surroundings.

 

Please understand, I have no issues outside between dawn and dusk !

Edited August 18, 2009 by Rolo

[markowich]

Misconception: If you are focussing by lines only you are using less than half of the potential of a rangefinder. Coincidence focussing needs no more than a structure- any structure. Contrast focussing even less, although it is the most accurate way. Moving subjects? It works just as well - it is just a matter of learning the technique, as are many skills.

 

agreed, a 'dominating' structure is enough. as long as it is stil and not a periodic pattern. did you ever try to rangefinderfocus a rice paddy? pure nightmare. i also -respectfully of course- disagree on moving objects-they become guesstimation with rangefinders. i have not seen many M's in soccer stadions recently, but maybe only because the sports photographers didn't learn the techniques yet.

peter

[jaapv]

Don't use the Allen key - I've been told off pretty severely by Will van Manen for that. It may correct a misalignment on infinity, but it upsets the two other checkpoints for focus. Better to get a qualified technician to do the adjusting. It is a skilled job and one really needs a reference lens and a collimator.

[Jamie Roberts]

Rolo: I actually find the opposite to you. I also shoot many thousands of pictures per week. I use a Nikon D3, which arguably has the best AF on the planet. In great light and with subjects moving quickly, it's hard to beat (the M8 consistently beat my 5d and 1ds2, and the D3 is an order of magnitude better than they are).

 

But honestly, I've never really missed a shot with my M8 (or the DMR for that matter!) at a wedding. It's true I'm not shooting past 90mm with it (and not that much past 50), and the Nikon is excellent with a 135 and longer. People at weddings don't move like quarterbacks or hockey players (and if they did, the D3 would be / is my tool of choice). Bird releases are probably the trickiest things to get a closeup; the D3 is great for them.

 

But in low-light my eyes are better (still) than the D3 by far.

 

Unless I want to use the focus assist on the SB900 (which I don't / won't for my work), the M8 (with a magnifier) consistently gives me much easier to focus results in very low light.

 

And guys, no-one has full cross-type AF sensors around the whole display yet to my knowledge (the d3 has 51 focus points, but not all are high resolution), so you're looking for areas of contrast and verticals--and focusing and recomposing--as well with your AF cameras in the dark.

 

So I have no trouble using a rangefinder for most shots, though there are times I rely on AF. But the difference can be far less than many people think.

 

YMMV of course.

[jaapv]

agreed, a 'dominating' structure is enough. as long as it is stil and not a periodic pattern. did you ever try to rangefinderfocus a rice paddy? pure nightmare. i also -respectfully of course- disagree on moving objects-they become guesstimation with rangefinders. i have not seen many M's in soccer stadions recently, but maybe only because the sports photographers didn't learn the techniques yet.

peter

Go to the photoforums here. I won't contaminate this thread with images, but I can show you dozens if not hundreds of shots of flying birds, jumping dolphins, running dogs, playing kids, etc. All focussed confidently with the M8 and wide open and long lenses.. And so can many other users. It is really just a matter of technique.

Btw, how did your autofocus do on the rice paddy? I bet it is still hunting for focus

[Jamie Roberts]

{snipped} i have not seen many M's in soccer stadions recently, but maybe only because the sports photographers didn't learn the techniques yet.

peter

 

Peter, really now... you won't find a rangefinder at a sporting event but that's not because it won't focus well.

 

I would argue that it's primarily a matter of reach: the market for 28mm sports shots from the press bleachers is probably pretty limited.

 

You really should see Doug Herr's shots or Rob Stevenson's basketball shots from a DMR--manual focusing--even manual follow focusing, is not the main issue.

 

Though of course with longer lenses and higher magnification it's much easier in a lot of cases to rely on great AF from Canon and Nikon, but even there it doesn't mean that MF is not capable in the right hands, and your sarcasm (as usual, IMO) is misplaced.

[Jamie Roberts]

Go to the photoforums here. I won't contaminate this thread with images, but I can show you dozens if not hundreds of shots of flying birds, jumping dolphins, running dogs, playing kids, etc. All focussed confidently with the M8 and wide open and long lenses.. And so can many other users. It is really just a matter of technique.

Btw, how did your autofocus do on the rice paddy? I bet it is still hunting for focus

 

Amen Jaap

[markowich]

Go to the photoforums here. I won't contaminate this thread with images, but I can show you dozens if not hundreds of shots of flying birds, jumping dolphins, running dogs, playing kids, etc. All focussed confidently with the M8 and wide open and long lenses.. And so can many other users. It is really just a matter of technique.

Btw, how did your autofocus do on the rice paddy? I bet it is still hunting for focus

 

next time i go to see Arsenal play i will definitely watch out for those m8/7/9 shooters who verify your theory. and i shall search for M and birds and dolphins and running dogs meanwhile.

incidentally, the (nikon and hasselblad) AF worked very well in the rice paddies, since the sensors are small enough (so that they do not see too much of the periodic structure).

peter

[markowich]

Peter, really now... you won't find a rangefinder at a sporting event but that's not because it won't focus well.

 

I would argue that it's primarily a matter of reach: the market for 28mm sports shots from the press bleachers is probably pretty limited.

 

You really should see Doug Herr's shots or Rob Stevenson's basketball shots from a DMR--manual focusing--even manual follow focusing, is not the main issue.

 

Though of course with longer lenses and higher magnification it's much easier in a lot of cases to rely on great AF from Canon and Nikon, but even there it doesn't mean that MF is not capable in the right hands, and your sarcasm (as usual, IMO) is misplaced.

 

jamie,

i do not dispute that there are focusing geniuses around (and i mean it). but the commoner (like myself) has to follow different rules.

as far as sarcasm is concerned, you certainly know the movie 'The good, the bad and the ugly'. most of the forum is soooooo good with leica, it certainly needs some of the other stuff, no----)))

peter

[giordano]

Japp, the formula you're after if perhaps b' = (e * f^2) / (k * z) where b' is the base length, e the visual acuity (0.0003mm at approx. 1 arcmin), f the focal length, k the aperture and z the circle of confusion (CoC).

If you choose a CoC value of 0.023mm for the M8 by example, this would make a base length of about 60mm to focus accurately a 135 lens at f/4. Smaller CoC values would lead to higher figures on the grounds of the same formula. But again i may be wrong as i'm no techie at all.

 

If that's the formula, the aperture certainly is not insignificant. For f=135mm b' is:

f/2 115mm

f/2.8 82mm

f/4 58mm

f/5.6 41mm

 

 

(BTW, dimensional analysis suggests that the formula requires that e be an angle, presumably in radians, not a length in mm.)

[lxlim]

I'm using the 1D MkII N and have less high ISO capability and I stay below 1250, but I'll go down to 1/60s so my light conditions are not too far away from yours. I generally shoot on Ai Servo focus, but when I get down to these levels I'll switch to standard. I'll add a ttl flash at this level if at all possible as I'm doubtful if many pictures I'd shoot at these levels of illumination are actually worth having. If the face is grey & flat - why shoot it ?

 

Ai Servo does take a moment to lock-on in low light, but usually my subjects are not moving fast, otherwise a flash would be pretty essential. Not sure if it's 3/4 second to focus, but it maybe. That would compare with several seconds with the M I suspect.

 

I have an MP, M7 and an M8 plus six Leica lenses, so I'm well committed.

 

Where do you find that the M's can outpace the 1D ?

 

 

-- Where do you find that the M's can outpace the 1D ?

 

Its a stupid sort of business in a sense but it pays well. You could take a peek at a sample on my homepage http://www.syfphotography.com/

 

I photograph students performing on stage for our national judging event. Bands, Chinese Orchestras, Choirs, dance, etc. Then I create a montage of them in performance with graphics, event titles, etc and sell it to them. And because I continue to hold the rights to the created images, I can continue selling them pictures after 10 - 20 years.

 

In order to create the montage, I literally run in the gallery facing the stage to capture several angles. I have at most 8 - 9 minutes to do this as each group has 12 - 16 minutes on stage.

 

Where the RF wins over my Canons is that non AF lenses have less play or decentering than zooms do and the weight of the M system.

 

I can sprint 50m select the best spot, set down the tripod, FOCUS once and get the shot FIRST time. Canon's AF is good when the subject is relatively big but in a scattered formation like an orchestra it is only good for 90 - 95% of the time. I tried Nikons' AF which is better but I was not looking for a little better solution but a much better solution.

 

Emplaced pocket wizard triggered cameras are a partial solution but less useful than I anticipated because the variations were too many. Running lightweight with an M8 produced much better and more consistent imagery than my Canons. And because I have to do this every day for 6 weeks I would probably die of exhaustion using a Canon.

 

Fringe benefit is I am shooting less but better images, less chimping to see if the images are in focus, files are smaller, post processing is a lot easier with a CCD sensor.

 

But for most part I am happy with Canon's AF but I usually use the One Shot rather than AI servo. I do regret selling my 1DmkIIn, the AI servo is better I think. When I am shooting dance, its the 1D for me though

 

Don't get me wrong the M8 is good in some situations and doesn't cause me eyestrain, in others I rather have the Canons AF.

[Advertisement]

Hallo,

Look here: Leica M

[dcuthbert]

This is a pretty interesting subject, albeit it a bit too techy for non-maths loving people.

 

Yes RF is more accurate if you learn the technique, but there lies the issue. The Leica book is rather pants, sorry I've said it. So if all the rumours are expected, you will see an influx of people buying M9/10's etc, all coming from the Canon/Nikon game and all experiencing the same issue with understanding the technique.

 

Now I know this could be a sticky in the customer forum, but they have a Leica M8/9/10 and I'm sure many would appreciate a well-written guide to this technique.

 

Thing is, you need to search for it, go through many arguments about focusing and, well, it gets rather longwinded in the end.

 

So how does creating a sticky post that explains the techniques and keeps on track. That way newcomers can come in and really learn how to use this amazing tool, all of you don't need to keep on explaining the same stuff, or typing "use the search" and overall the community helps those in need, easily.

 

I've noticed an unanswered question in the customer forum, one I'm pretty keen on knowing the answer to, yet there isn't one..

 

http://www.l-camera-forum.com/leica-forum/customer-forum/94048-fast-focusing-portraiture.html