The DOF vs Diffraction challenge: FF or crop?

[erpman]

So, in my somewhat particular form of landscape photography I´m after maximum depth of field (1,5m-infinity) and maximum sharpness for making huge, wall-sized prints. Do note, these prints are printed at around 300ppi.

Since I can´t afford MF backs, but also because of the DOF consideration, I use a panoramic bracket and stitching tools to generate large image files for print. This also means that I use longer lenses (50-100mm), hence the critical DoF factor.

Having seen the reviews of the Pentax k5 I´m getting tempted to trade in my 5dmkII for it, in order to gain more DOF. I´m tired of soft images from having to stop down to f/22 and even further. But something tells me that the gain in DoF is at the expence of image quality resulting from the smaller sensor of the k5, and also, the lower aperture at which diffraction comes into the mix on a smaller sensor. Correct me if Im wrong. 

Stitching gives me all the resolution that I want, so that´s not an issue. The only issue is wether trading in a ff sensor for crop is gonna give the expected increase in dof/sharpness, or whether I´m just trying to overcome the laws of physics.

Could an alternative be to get the pentax 645d and just stopping it down more on somewhat shorter focal lengths? After all, the image quality and larger circle of confusion could make for a grand total with higher IQ overall.

Input is very appreciated!

[Sheldon N]

If you're already shooting pano and are that far into post production, why not consider focus stacking in addition to pano shooting?

[ErikKaffehr]

Hi,

DoF does not really exist anyway. If you make big print, viewed very close DoF will be essentially nil. That said a smaller format may have some advantage, about one stop. I absolutely agree on the recommendation to use focus stacking, although I never have done it on stitched images.

I have an article discussing DoF and diffrcation here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures

Some examples of extending DoF are given here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/29-handling-the-dof-trap

Best regards
Erik

If you're already shooting pano and are that far into post production, why not consider focus stacking in addition to pano shooting?

[stever]

there is no free lunch.  diffraction results in serious loss of sharpness beyond f16 in full frame, and f11 in crop frame - pretty much in direct porportion to the difference in DOF.  my personal experience is that some high quality simpler design lenses such as the Canon 90 TS hold up better at f19 or f22 than complex designs like the 100 L - but it's a marginal difference.  however, using a tilt-shift lens can seriously improve DOF in one plane and this can be used with stitching (search posts on tilt-shift panoramas)

focus stacking can be very effective (for motionless subjects) but there's a serious learning curve.  if you're doing panos, i think you're better off with a good MF lens as you need to duplicate the focus stacks for each slice of your pano - or shoot tethered with the Helicon software to duplicate the focus stack for each slice

going to a crop frame camera will reduce resolution without increasing DOF

[erpman]

As I feared, and kinda had calculated myself, there is indeed no free lunch with regard to the earlier onset of diffraction on crop sensors.

Focus stacking is not an option, as the process is slow enough as it is, but TS might be an area to look into.

But how about MF? Im wondering if going with an MF system like the pentax 645d, shorter focal lengths and smaller stitches (3 images instead of 12-15), and then allow for some more enlargement/resampling in PS facilitated by the IQ produced by the larger sensor/lens.

[hjulenissen]

Focus stacking is not an option, as the process is slow enough as it is...

Perhaps focus stacking would allow you to reduce the number of stitched images (using a wider lense), keeping the total time spent in check (but altering the balance between DOF and maximum sharpness)?

-h

[erpman]

Just did a side by side comparison between the 5dmkII and the pentax 645d. The 5dmkII image was upscaled to 200% in ps, the pentax to 400%. Pentax still looks better. There is something about the tonality and smoothness of mf that seems to give it the upper hand. Look at the feathers; they look so much cruder on the 5d. This is definitely interesting.

Both images received the same sharpening and I just tweaked the contrast of the 645 for comparison.

[deejjjaaaa]

there is no free lunch.  diffraction results in serious loss of sharpness beyond f16 in full frame, and f11 in crop frame - pretty much in direct porportion to the difference in DOF.  

that is 2x crop @ f11 by the way...

[erpman]

Not bad, what system is this?

[stever]

your comparison of the 645 and 5D2 is consistent with others i've read.  the 645 will have higher IQ.  whether it will get you to the combination of sharpness and DOF you're looking for may still be a problem - you can shoot the 645 at f22, but DOF will be about the same as f16 with FF, so the gain in DOF will come from being able to shoot wider lenses - you can do the math on how much wider and whether that will get the DOF you're looking for.

if you can go enough wider to do single row vertical panos with a tilt lens then there are potentially big gains in apparent DOF for suitable subjects

it might be possible to do multi-row with tilt on the bottom row only, but doesn't seem very practical

focus stacking is the only way to seriously increase depth of field but is very time consuming and can be quite frustrating

dee - i think you misunderstood me - diffraction starts to become seriously noticeable with crop frame beyond f11

[deejjjaaaa]

Not bad, what system is this?

m43, SLRGear is using 12mp camera to test m43 lenses ... so f11 on 2x crop is like actually f22 on FF (24x36) and still PL25/1.4 delivers good results @ f11...

[deejjjaaaa]

dee - i think you misunderstood me - diffraction starts to become seriously noticeable with crop frame beyond f11

which crop you are talking about ? if 1.5x then as 2x @ 12mp sensor is still good @ f11 -> w/ proper lenses 1.5x might be OK (and not seriously noticeable) @ f11 + 1/2 stop down... it is another story that most lenses are too bad to deliver on 1.5x and above @ those apertures...

[Wayne Fox]

there is no free lunch.  diffraction results in serious loss of sharpness beyond f16 in full frame, and f11 in crop frame - pretty much in direct porportion to the difference in DOF. 

While I do not disagree with your statement, one thing I've found is you can fix some of the softness from diffraction in post ... something you cannot do with out of focus from not enough "depth of field".  Shooting something at f/22 may leave the file soft, but a little work and prints look pretty good. You probably loose some of the micro detail, but it can still work pretty well.

[erpman]

While I do not disagree with your statement, one thing I've found is you can fix some of the softness from diffraction in post ... something you cannot do with out of focus from not enough "depth of field".  Shooting something at f/22 may leave the file soft, but a little work and prints look pretty good. You probably loose some of the micro detail, but it can still work pretty well.

I do agree on this, I use RawDeveloper which has a R/L Deconvolution algorithm and it does a fairly good job at restoring diffraction softness. Actually, when stitching, this means that you can, if done properly, slightly alter focus between images as you move through the scene since the diffraction softness/extended depth of field covers up small differences in focus from frame to frame.

Now, I did an interesting comparison with regard to the Pentax 645D. Since its sensor is not "true" 645 but slightly smaller (33.0 x 44.0) it means that 55mm is "standard" lens on this camera (crop factor of 0,79 i think). This translates to 43mm on FF, a very handy focal length for the work that I do. When you put this into the equation, along with the increased resolution and IQ and upscaling possibilities of the files, it gets closer to my goal. Have a look at the calculations from the Cambridge Dof calclulator at the bottom.

This means that I theoretically get 1m closer to my subject when using hyperfocal focusing. Or is there something that I´ve overlooked? Could there be an error with the calculator assuming full 6x45 sensor size and therefore a larger CoC? The pixel pitch on the 645D is 5.93, smaller than on the 5dmkII. Does this mean that comparing f16 on the 5d to f22 on the pentax would be unfair?

[stamper]

While I do not disagree with your statement, one thing I've found is you can fix some of the softness from diffraction in post ... something you cannot do with out of focus from not enough "depth of field".  Shooting something at f/22 may leave the file soft, but a little work and prints look pretty good. You probably loose some of the micro detail, but it can still work pretty well.

Brian Peterson in his latest book on Exposure brings attention to this "problem". Two images shot, one at f/8 and one at f/22. Blows both up to 200% and invites the reader to look at them. They are trees. He points out there is only a little softness in the one shot at f/22 and states it isn't worth bothering about it. I am paraphrasing him. It isn't a scientific appraisal but a practical one which I think matters the most. I have since rethought my beliefs on this subject which meant that I rarely went beyond f/13. Practical examples are more important to me than theoretical ones.

[ErikKaffehr]

Hi,

On the other hand it's like using a 6 MP camera instead of 24 MP camera. Half the resolution is lost, cutting usable pixels in one fourth. Now, I have made decent A2-prints from 6MP, but a significant amount of information is lost.

The solutions to infinite depth of field is "Scheimpflug" or focus stacking. On the other hand, selective focus is one of the expressions we have in photography. It may be better to have main subject in focus than having everything slightly unsharp.

Diffraction depends only on aperture, there are no lenses more tolerant of diffraction. It's a property of light. The form of the aperture may matter, but I presume that all lenses we discuss have circular or near circular apertures.

This page illustrates the effects of defocus and diffraction:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1

Sharpening is discussed here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=2

Best regards
Erik

Brian Peterson in his latest book on Exposure brings attention to this "problem". Two images shot, one at f/8 and one at f/22. Blows both up to 200% and invites the reader to look at them. They are trees. He points out there is only a little softness in the one shot at f/22 and states it isn't worth bothering about it. I am paraphrasing him. It isn't a scientific appraisal but a practical one which I think matters the most. I have since rethought my beliefs on this subject which meant that I rarely went beyond f/13. Practical examples are more important to me than theoretical ones.

[stamper]

Are you saying diffraction results in a loss of resolution? A new one on me but I am willing here to learn.

[Bart_van_der_Wolf]

Are you saying diffraction results in a loss of resolution? A new one on me but I am willing here to learn.

That's not only what Erik is saying, it's a commonly known fact called physics, unfortunately. Deconvolution sharpening can restore some of the resolution lost due to diffraction blur, but some loss remains and noise may increase. The fact that you act surprised suggests that you don't see a difference between an actual image taken at e.g. f/8 and f/22. I'm puzzled by that. Are you saying that you can fully remove the blur caused by diffraction, or do you see no difference to begin with?

Cheers,
Bart

[Walter Schulz]

Diffraction depends on "effective aperture" and therefore depends on which magnification is used in a particuliar picture.
If you're focussing objects far, far away (like leaves on a tree) effective aperture and the aperture number are very, very close together.
If you're doing macro work (details of bank notes), you will see diffraction effects at lower aperture numbers.

k(eff) = k (1 + magnification)

k = aperture number of the lens
k (eff) = effective aperture number
magnification = sensor or film size / image size

I've done some work at magnification 1:1 to 3:1. When doing pixel peeping (all of us are weak at times ...) you will see serious detail loss with 3:1 and f/11 (lens setting) because "effective aperture" is f/44.

For people able to understand german language (and not afraid of maths): http://www.traxel.de/foto/drf/schaerfentiefe.pdf
Hope this helps to see the whole picture (sorry for the lame pun).

Ciao, Walter

[stamper]

That's not only what Erik is saying, it's a commonly known fact called physics, unfortunately. Deconvolution sharpening can restore some of the resolution lost due to diffraction blur, but some loss remains and noise may increase. The fact that you act surprised suggests that you don't see a difference between an actual image taken at e.g. f/8 and f/22. I'm puzzled by that. Are you saying that you can fully remove the blur caused by diffraction, or do you see no difference to begin with?

As I stated in an earlier post I have very rarely shot smaller than f/13 because that was the advice that I read. I saw the examples in Brian Peterson's book but it didn't mention loss of resolution. Yes there was some softening at 200%. I have a lot of photography books and a lot of the photographers are happily shooting away at f/22 so it becomes confusing as to who are right and who are wrong. I have done a lot of shooting with ND filters and sometimes a slow shutter speed means going smaller than f/13 but mostly at sunset, so detail isn't critical. As I said still learning.

Cheers,
Bart