Hyper Focal Distance

[Mike Sellers]

Here is another focus technique from Merklinger:
If the smallest object in the scene is 2 mm wide, such as a blade of grass, divide the focal length by 2 to get the aperture then focus at infinity. So if you are using a 35mm lens you divide by 2mm = about f16 then focus at infinity and all will be in focus.

[Dick Roadnight]

Here is another focus technique from Merklinger:
If the smallest object in the scene is 2 mm wide, such as a blade of grass, divide the focal length by 2 to get the aperture then focus at infinity. So if you are using a 35mm lens you divide by 2mm = about f16 then focus at infinity and all will be in focus.

I believe in Merklinger's theory of focusing far rather than near, but this in a pretty useless theory for 300mm digital lenses, when diffraction kicks in at about f16!

[feppe]

I believe in Merklinger's theory of focusing far rather than near, but this in a pretty useless theory for 300mm digital lenses, when diffraction kicks in at about f16!

I've also had mixed success applying his methods to small sensors (APS-C and MFT). I should do some more testing, though. I've only tested at relatively short focus distances in the city, 5-50 meters. Results might be different with landscape photography.

[ErikKaffehr]

Hi,

In my view the only method which works really well with longer lenses is focus stacking. The reason is mainly that with digital the area of critical focus is very small. That is not really depending on digital, it's just that higher precision is possible.

I have a small article on different methods here:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/29-handling-the-dof-trap

Perhaps it may be useful.

Best regards
Erik

I've also had mixed success applying his methods to small sensors (APS-C and MFT). I should do some more testing, though. I've only tested at relatively short focus distances in the city, 5-50 meters. Results might be different with landscape photography.

[Mike Sellers]

In that case the program for the pocket pc would work fine. I used it with my 70-210 lens at the long end and I really liked the results.

[feppe]

Hi,

In my view the only method which works really well with longer lenses is focus stacking. The reason is mainly that with digital the area of critical focus is very small. That is not really depending on digital, it's just that higher precision is possible.

I have a small article on different methods here:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/29-handling-the-dof-trap

Thanks. I might need to check out focus stacking for some subjects. DOF is helped by the smaller sensors I generally use, but there are some (film) MF and LF scenes which require pretty involved approach to get acceptable DOF. I still haven't nailed it despite using Cambridge in Colour's excellent DOF calculator in the past, which takes print size into account. Finding the sweet spot between diffraction and acceptable focus requires quite a bit of planning, previsualization and calculations, some which are not feasible in the field (unless someone knows a full-featured DOF and diffraction calculator for Android phones). Focus stacking is probably not feasible due to cost with LF, unless there are clear planes of focus like in your distant mountain example.

Also, Rome had at least four cinematographers.

[ErikKaffehr]

Hi,

With large format you can probably stop down a lot. Film is less flat than sensors and large format lenses don't resolve as high as lenses for smaller formats, at least in general. In large format you also can use Scheimpflug with ease. Scheimpflug does not really increase DOF just changes the plane which is in focus.

Some LF cameras (like Sinar) have a Scheimpflug calculator built in, BTW.

Best regards
Erik

[Dick Roadnight]

Thanks. I might need to check out focus stacking for some subjects. DOF is helped by the smaller sensors I generally use, but there are some (film) MF and LF scenes which require pretty involved approach to get acceptable DOF. I still haven't nailed it despite using Cambridge in Colour's excellent DOF calculator in the past, which takes print size into account. Finding the sweet spot between diffraction and acceptable focus requires quite a bit of planning, previsualization and calculations, some which are not feasible in the field (unless someone knows a full-featured DOF and diffraction calculator for Android phones). Focus stacking is probably not feasible due to cost with LF, unless there are clear planes of focus like in your distant mountain example.

The Rags DOF Calculator tells you the size of the Airy disc (the COC that diffraction limits you to) so you can work it out... to focus from 30m to 400m @ f8 with COC = 0.01 you need to stack about 8 images... and this is the street outside my house, not a hypothetical situation.

All the formulae are available in Merklinger or on Wikipedia, and I can create an Excel spreadsheet on my laptop and download it into my Nokia phone - can you not do that with yours?

I ignore print size when taking photographs, as I might want to print it 6 foot tall, and my standard print size is 18 * 24" @360 ppi.

 

[feppe]

The Rags DOF Calculator tells you the size of the Airy disc (the COC that diffraction limits you to) so you can work it out... to focus from 30m to 400m @ f8 with COC = 0.03 you need to stack about 8 images... and this is the street outside my house, not a hypothetical situation.

All the formulae are available in Merklinger or on Wikipedia, and I can create an Excel spreadsheet on my laptop and download it into my Nokia phone - can you not do that with yours?

I ignore print size when taking photographs, as I might want to print it 6 foot tall,  an my standard print size is 18 * 24" @360 ppi.

Yeah, I've used the same DOF calculator, and while clunky it has some of the best feature sets available.

I have an Android phone which has a pretty good Photo Tools app, which allows for customized CoC, among other things. I've added a setting for 4x5" LF printed at 20x24", for example. It doesn't have a diffraction calculator, but the thresholds are pretty easy to memorize.

It would be nice if someone would program an all-inclusive DOF calculator, which would take everything from aperture, focal length, sensor size/film format, diffraction and print size/viewing distance into account, without having to use different calculators for them.

[Bart_van_der_Wolf]

I believe in Merklinger's theory of focusing far rather than near, but this in a pretty useless theory for 300mm digital lenses, when diffraction kicks in at about f16!

Hi Dick,

With all due respect but Merklinger is wrong, when we're talking about digital sensors.

It is impossible to distiguish a focused infinity from a defocused infinity as long as the blur diameter of the infinity defocus is smaller than the sensel. In fact one is unlikely to see any difference, especially with a sensor array that's fitted with an AA-filter, when the blur diameter is less than 1.5x the sensel pitch. All one needs to do is use that as the COC criterion for the hyperfocal distance calculation.

The discussion usually comes from using too large a COC. Of course with film this becomes a different issue, and Merklinger is correct.

Cheers,
Bart

[Bart_van_der_Wolf]

It would be nice if someone would program an all-inclusive DOF calculator, which would take everything from aperture, focal length, sensor size/film format, diffraction and print size/viewing distance into account, without having to use different calculators for them.

Hi Harri,

There is such a tool, unfortunately only available for Windows. It's customizable, can output tables, or compare different scenarios, and it displays the blur diameter at infinity, or for an object at a given distance from the focus plane.

The recommended COC for a given print viewing distance is given in the PDF that goes with the application, and is very simple to compute. One only needs to make a note of some viewing situations (but remember that people tend to walk up to the image for close inspection, if possible), and plug those values into the calculator. You can even use a couple of COCs side by side.

I wouldn't mind programming something myself (and I probably will), but it would also be Windows only (Mac is too much of a hassle, and would require an investment on my part).

Cheers,
Bart

[Dick Roadnight]

Hi Dick,

With all due respect but Merklinger is wrong, when we're talking about digital sensors.

It is impossible to distiguish a focused infinity from a defocused infinity as long as the blur diameter of the infinity defocus is smaller than the sensel. In fact one is unlikely to see any difference, especially with a sensor array that's fitted with an AA-filter, when the blur diameter is less than 1.5x the sensel pitch. All one needs to do is feed that as the COC criterion for the hyperfocal distance calculation.

The discussion usually comes from using too large a COC. Of course whith film this becomes a different issue, and Merklinger is correct.

Cheers,
Bart

What you say is correct, and I appreciate that Merklinger's examples refereed to film, but what does Merklinger say that you think is incorrect? I merely mentioned his books as a source where the formulae could be found.

For critical stacking work I would use COC of 10 (about 1.5 time sensel pitch).
For snapshots I would use 30, and I do not use anything with an AA filter.
At f22 the Airy Disc is 30ish... for hand-held work I think, on balance that f16 (or f22) is a good compromise, depending how well the camera performs at High ISO.

[feppe]

Hi Harri,

There is such a tool, unfortunately only available for Windows. It's customizable, can output tables, or compare different scenarios, and it displays the blur diameter at infinity, or for an object at a given distance from the focus plane.

The recommended COC for a given print viewing distance is given in the PDF that goes with the application, and is very simple to compute. One only needs to make a note of some viewing situations (but remember that people tend to walk up to the image for close inspection, if possible), and plug those values into the calculator. You can even use a couple of COCs side by side.

I wouldn't mind programming something myself (and I probably will), but it would also be Windows only (Mac is too much of a hassle, and would require an investment on my part).

Cheers,
Bart

That's a pretty good program, thanks. Doesn't have a diffraction warning, though, and an integrated CoC calculator would also be useful - I dno't trust any baked-in CoC figures as they often rely on decades-old standards (small prints).

[Bart_van_der_Wolf]

That's a pretty good program, thanks. Doesn't have a diffraction warning, though, and an integrated CoC calculator would also be useful - I dno't trust any baked-in CoC figures as they often rely on decades-old standards (small prints).

There is a simple calculation possible for diffraction (assuming green light of 555nm). When one takes 1.5x the sensel pitch as one's guideline, then 1.108x the sensel pitch in microns give the widest aperture value at which diffraction becomes visible (so just adding 10% and rounding up is easy enough to do by head). For a guideline of 1x the sensel pitch it would become a factor of 0.738x sensel pitch (subtract 1/4 of the sensel pitch in microns) for the onset of visible diffraction.

There is no explicit diffraction warning (because it's wavelength dependent), but once one takes the sensel pitch (or 1.5x the sensel pitch) criterion as calculated above as a guideline, then it's easy enough to see if the blur (either infinity or POI) exceeds that. The 'Options|COC criterion' menu choice (e.g. critical) for a given recording format, gives a very good starting point.

Besides, if an 8x10 in (~20x25 cm) print looks good at normal viewing distance (say 10-12 inches, 25-30 cm), then it will look good at any viewing distance provided that the viewing distance scales proportionally with the output size. Of course for close inspection, only pure resolution will survive scrutiny.

Cheers,
Bart

[feppe]

Besides, if an 8x10 in (~20x25 cm) print looks good at normal viewing distance (say 10-12 inches, 25-30 cm), then it will look good at any viewing distance provided that the viewing distance scales proportionally with the output size. Of course for close inspection, only pure resolution will survive scrutiny.

Good point - but many insist on sticking their nose to the print no matter how big is it

[Dick Roadnight]

...if an 8x10 in (~20x25 cm) print looks good at normal viewing distance (say 10-12 inches, 25-30 cm), then it will look good at any viewing distance provided that the viewing distance scales proportionally with the output size. Of course for close inspection, only pure resolution will survive scrutiny.

Cheers,
Bart

Some people buy big cameras to make big prints (not just for weight training), so if my sensor is N times bigger than 24 * 36mm, I want to be able to make prints 3N times bigger than you could with a 24Mpx 24 * 36mm sensor with an Anti-Aliasing filter.

I want people to be draw in to my pictures (and not just ignore them and walk past) and to move in close and look for detail... the old master paintings did this half a millennium ago, but some photographers seem to spend all their time trying to convince everyone that low-res is adequate.

[Bart_van_der_Wolf]

What you say is correct, and I appreciate that Merklinger's examples refereed to film, but what does Merklinger say that you think is incorrect?

Given the topic of hyperfocal distance, I assumed we are talking about this paper, where he say"s:
"So, the bottom line runs something like this. In order to ensure that distant objects are imaged sharply, we
must focus at the distance to the farthest object".

For objects at infinity that would mean one has to focus at infinity. That IMHO doesn't hold for digital sensors where one can focus closer (hyperfocal distance with a sensel pitch related COC). That will also give better foreground resolution than infinity focus will.

Cheers,
Bart

[Bart_van_der_Wolf]

Some people buy big cameras to make big prints (not just for weight training), so if my sensor is N times bigger than 24 * 36mm, I want to be able to make prints 3N times bigger than you could with a 24Mpx 24 * 36mm sensor with an Anti-Aliasing filter.

I want people to be draw in to my pictures (and not just ignore them and walk past) and to move in close and look for detail... the old master paintings did this half a millennium ago, but some photographers seem to spend all their time trying to convince everyone that low-res is adequate.

Hi Dick,

I fully agree, there is no substitute for real resolution, and it can be gotten by reducing the need for magnification, IOW plenty of high quality pixels.

As for the old master painters though, they were very good at suggesting detail where there is none... Try looking at some of the old portaits how they suggested the intricate fabric of lace by coarse dabs of paint. Here they also used the 3D texture of paint itself to add to the suggestion. Same with distant detail, sometimes just well applied blobs of paint.

Cheers,
Bart

[Dick Roadnight]

Hi Dick,

I fully agree, there is no substitute for real resolution, and it can be gotten by reducing the need for magnification, IOW plenty of high quality pixels.

Bart

Thanks, Bart, I am glad we are in agreement...

"Magnification" is image size divided by subject size... and it is still important, especially as the lens (rather than pixel count) is often the limiting factor, but I generally think in terms of "original camera pixels per print inch".

[Bart_van_der_Wolf]

Thanks, Bart, I am glad we are in agreement...

"Magnification" is image size divided by subject size... and it is still important, especially as the lens (rather than pixel count) is often the limiting factor, but I generally think in terms of "original camera pixels per print inch".

Yes, I meant to say output magnification.

Cheers,
Bart