M8 Backfocusing with respect to aperture

[jonoslack]

Advertisement (gone after registration)

Jono,

and I thank you to share the Eoin's use of M8/Aperture (i did not see before)!

Great site indeed.

 

Flavio

 

HI Flavio

I'm going to start a thread on this on later on this evening - keep an eye open, I think it deserves a thread of it's own.

[nzav]

While I suppose that Nick and Steve are correct in assuming that this is caused by an asymmetrical increase in depth of field, there is also a real, if very subtle effect with some lenses. If you stop down, the focus may actually shift, as explained here: Spherical aberration.

 

I would agree with MJH wholeheartedly with only one reservation. Leica (and earlier Leitz) optics are extremely well corrected, and I believe any residual spherical aberration would most likely be negligible. Leica earned their reputation by designing and producing lenses that are better than any of their contemporaries. A less well-corrected lens may exhibit such noticeable defects, but when the lenses cost in the thousands, spherical aberration should not be one of the things you get for your money...

[jonoslack]

I would agree with MJH wholeheartedly with only one reservation. Leica (and earlier Leitz) optics are extremely well corrected, and I believe any residual spherical aberration would most likely be negligible. Leica earned their reputation by designing and producing lenses that are better than any of their contemporaries. A less well-corrected lens may exhibit such noticeable defects, but when the lenses cost in the thousands, spherical aberration should not be one of the things you get for your money...

HI there

Well- it's pretty easy to demonstrate - it's one of those things which would never have come up using film, but it becomes quite obvious doing testing with digital like this.

 

The theory is there, and the experience is there as well - mine today from 'fresh' and the experience of others who have done the same sort of testing.

[tom0511]

What a wonderful place this is.

 

Today I have learned:

 

1. that I should search past threads

2. that I wasn't going mad, and that there is a backwards focus shift when you use a smaller aperture (despite the increased depth of field).

3. that I can alter the focus distance myself without having to resort to Solms.

 

All I need now is a 2mm allen key!

 

Many thanks to everyone who has chipped in. Specially to Guy for the explanation of how to change the focusing, to John who linked to the old thread, and to Michael who explained about focus shift.

 

And yesterday it was how to use Aperture with the M8! (thanks to Eoin for that one)

 

Adjusting the camera helps only if really all lenses show the same tendancy (IMO).

Now in my experience some lenses show more focus shift , and others less.

Also focus shift IMO is only really a problem for fast lenses, because otherwise DOF helps to compensate for focus shift.

My Noctilux focuses perfect at f1.4 but a little of at f1.0.

Now if I get it calibrated to focus perfect at f1 it would probably be off a little at f1.4.

What to do?

If you know the tendancy of your lenses one could compensate a little when focusing.

 

Also frienkly I only use f2 and faster if the light forces me to do so.

Shallow DOF might be nice to separtae from the background, but its hard to handle, not only to focus correct , but also to have a subject which works for shallow DOF.

For example often (not allways) the "One-Eye not really sharp"-portraits dont look perfect to me.

 

I think focus shift is a real disadvantage of rangefinders (compared to SLRs), and I think its more pronounced by the harder sharp-unsharp-transition of digital sensors (vs film).

On the other side if you know your lenses I think one can handle it in a way that it is no real problem for real life.

 

If I wanted a percect in focus image with shallow DOF I think the AF ofmy d2x +85/1.4 makes it much easier to get a keeper compared to M8+75/1.4.

cheers, Tom

[lars_bergquist]

............

By the way, what is the circle of confusion (CoC) for the M8?

 

No camera 'has' a circle of confusion. This is just used to compute the d.o.f scale of the lenses, and of course d.o.f. tables. The 1/30 mm size is based on the rule of thumb that a circle of confusion of 1/10th of a millimeter *in the print* still looks sharp to the eye, at a normal wiewing distance (about 30 cm or 12 inches). So this 'standard' is really good only for a 3x enlargement.

 

The M8 needs more enlargement for a same-size print because of the crop factor. So my

advice is that while reading d.o.f at one stop larger aperture, as oftern advised, may be enough for 35 mm, with the M8 you really need two stops larger, i.e. read at 5.6 if you are using 11.

 

The old man from the Age of Contact Printing

[mjh]

No camera 'has' a circle of confusion. This is just used to compute the d.o.f scale of the lenses, and of course d.o.f. tables. The 1/30 mm size is based on the rule of thumb that a circle of confusion of 1/10th of a millimeter *in the print* still looks sharp to the eye, at a normal wiewing distance (about 30 cm or 12 inches). So this 'standard' is really good only for a 3x enlargement.

 

The M8 needs more enlargement for a same-size print because of the crop factor. So my

advice is that while reading d.o.f at one stop larger aperture, as oftern advised, may be enough for 35 mm, with the M8 you really need two stops larger, i.e. read at 5.6 if you are using 11.

As you said, the circle of confusion just defines what someone judges as “sharp enough”; there are no absolutes here. In general, there are two approaches: first, one could ask what detail could be resolved by a typical viewer from a normal viewing distance, just as you explained, and calculate the CoC from that. Alternatively, one could argue that “sharp enough” means “as sharp as it can possibly get with the given sensor”. The KAF-10500 has a pixel size of 6.8 x 6.8 microns, and since we need 2 x 2 pixels for capturing colour, we could set the CoC to 2 x 6.8 = 13.6 microns – or 19.2 microns if we base the CoC on the diagonal of the 2 x 2 pixel square rather than its sides. Or we could acknowledge that luminance is actually computed from the green-sensitive pixels mostly: approximately 1.4 x 1.4 pixels are necessary for resolving green (and thus luminance) detail, so we should divide the CoC by 1.4. All of these methods of calculating the CoC are valid and have their uses.

[lct]

Advertisement (gone after registration)

...my advice is that while reading d.o.f at one stop larger aperture, as oftern advised, may be enough for 35 mm, with the M8 you really need two stops larger, i.e. read at 5.6 if you are using 11/...

Don't know for the M8 but one stop larger is OK for the R-D1 IMHO.

[barjohn]

MJH,

 

What value for CoC do you use to apply the DoF calculator (0.0x mm ?) Two stops as suggested here for using the lenses calcualtions/display would indicate a very shallow DoF for the M8 compared to the R-D1 (shown as 0.02mm), is this correct?

[mjh]

What value for CoC do you use to apply the DoF calculator (0.0x mm ?) Two stops as suggested here for using the lenses calcualtions/display would indicate a very shallow DoF for the M8 compared to the R-D1 (shown as 0.02mm), is this correct?

Two stops is an exaggeration. The old rule of thumb would indicate that 1/1500 of the sensor diagonal is about right; that’s 22 microns (0.022 mm) compared to 19 microns (0.019 mm) for the R-D1s. That’s probably sharp enough for most purposes, but if you aren’t content until you’ve got all the detail the sensor could capture, you should use 14 microns (0.014 mm) or thereabout (16 microns or 0.016 mm for the R-D1s). Keep in mind that depth of field doesn’t have a well defined meaning; it depends on what you regard as sharp enough.

[jaapv]

Indeed there is, but given all the talk about 'focus shift' and 'backfocussing' it makes me wonder how people have ever managed to get anything in focus before the M8 came along :-)

 

 

 

Because of the different nature of film and sensors, Steve. Film has thickness which makes the point of light projected into it expand by refraction and coma. When the light hits the film at an angle it turns into an ellipse. Also the process is threedimensional. The thinner the film the sharper. Think of it as shining a torch into a murky plate of soup. A sensor acts more like an ideal plane where the lens produces optimum definition, limited only by the pixel size and Airy disks of the projection. This makes a digital image more defined, giving rise to the noticeability of things like DOF, spherical aberration and focus shift. And then the results a viewed at 100% crops from 20 cms distance on the monitor. When was the last time you viewed a 2x3 meter print that close up?

So it is perfectly normal that things which were irrelevant on film start to be noticed on a digital file. As soon as one makes a print of the digital image they turn out to be irrelevant there as well, though.

[sean_reid]

So it is perfectly normal that things which were irrelevant on film start to be noticed on a digital file. As soon as one makes a print of the digital image they turn out to be irrelevant there as well, though.

 

Well said! That's right.

 

Cheers,

 

Sean

[johnr]

There is a simple lens formula v=u*f/(u-f) which relates object distance, image distance and focal length.

 

So for a 50mm lens a 0.1 mm change of lens position from the infinity setting gives an object distance of 25 metres and for an 0.2mm change the object distance is 12.5 metres; for 0.3mm change the object distance is 8.4 metres.

 

We can imagine that the 0.1mm changes in lens position equate to equal acceptable sharpness changes or depth of field.

 

So the change is not linear and there is not a simple rule like one third in front and two thirds behind; as the focusing distance becomes shorter the depth of field becomes progressively nearer to equal amounts front and back

 

John

[lars_bergquist]

My recommendation of two stops (which of course makes d.o.f pretty hypothetical except with wide angle lenses) was based on the assumption of a maximum print size of about 18x24 cm or 8x10". Why no more? Because that is the largest print that you can take in as a whole at normal reading distance, i.e. about 30 cm or 12". A 24x30 you will hold with your arms outstretched, a 30x40 you hang on a wall. Thus, you adjust your viewing distance to an angle of view approximately equivalent to 18x24 at 30 cm.

 

Sticking your nose into a larger print, scrutinizing it with a magnifier, in a hunt for fuzz and grain, is a perversion and the only one I know of that should be punishable.

 

The old man from the Age of Contact Printing

[carstenw]

Lars, I just went gallery-hopping today, and the vast majority of the great prints I saw today, including some excellent ones by Bruno Barbey, were *not* really sharp when viewed from up-close. The few that were sharp even at close range got a little dollop of extra technical respect, but they weren't better images at proper viewing distances. It is important to keep things in perspective, as you say.