Diffraction versus Depth of Field

[chicagojohn]

Certain that this subject must have been discussed here countless times, I apologize in advance for bringing it up again.

But if someone would care to offer their experience anyway, what I would like to know is your thoughts/experience for a typical 1.5 CF sensor, at what f/number is the balance between DoF and diffraction optimized for a scenerio with about 60 sq microns per pixel (~6MP), and where one would want primarily to display images on a monitor at a 2MP resolution such as 1920 X 1080 versus making enlarged prints.

I realize there are a lot of mitigating variables, but I'm just looking for a rough rule of thumb, if it's possible to provide that on the basis I've stated, and also what sort of differences might be expected in comparing a 1:1 macro lens at around 100 mm with a typical 50 or 100 mm non-macro prime lens. That is, can a macro lens allow one to stop down further before adverse diffraction softening overcomes DoF?

[ravncat]

Originally Posted by chicagojohn

at what f/number is the balance between DoF and diffraction optimized for a scenerio with about 60 sq microns per pixel (~6MP), and where one would want primarily to display images on a monitor at a 2MP resolution such as 1920 X 1080 versus making enlarged prints.

Optomization is the tricky word, what do you mean by it? You also didn`t specify viewing distance, or monitor size, a 13 inch monitor and a 17 inch monitor that can both display at 1920 x1080 would have a different viewing size. That`s important. Let`s assume 50cm for a maximum dimension of 13 inches on the monitor. If we downsample to 2mp, you`ll probably see digital artifacts, but you shouldn`t really hit diffraction limit to beyond F16. If the image is smaller or viewing distance further - it will likely be beyond f22. So images for web are likely not going to have the diffraction issue. Now, if you aren`t downsampling, but take a 100% crop at that size, you`re probably going to be able to see diffraction start to creep in at f16 , and should be ok at f11, with that 6 megapixel 60sq micron pixel size.

At 1 meter for a 20 inch print from that same camera, you`ll be fine at f11, but you`ll start to see diffraction by f16. Move closer to 10cm away from your print and now you are diffraction limited at F2.8. Atop all of this there are fudge factors, based on your pixel structure and anti-aliasing algorithms and size of the circle of confusion that is acceptable to you. And what may be ok for someone with ok vision may not be for someone with 20/20 vision. As a result, optimal get`s pretty difficult to tell pretty quickly.

In general, with the camera you specified, for medium to large prints. i`d try to stay at f11 (F8 for very large prints maybe) or lower if possible, but wouldn`t hesitate to live with diffraction at f16 or 22 if there were some reason to go there - say you needed less light and didnt have a neutral density filter, or you wanted to try and get starbursts from point light sources, or were doing macro and needed more depth of field. After-all you can combat diffraction a little with judicious sharpening, and slight softness of the in focus area at the expense of more depth of field can be worth the trade-off.

for web use - I wouldn`t really care. for a 2 megapixel display, it would depend upon my monitor size and downsampling and sharpening skills... if I were really good, I wouldn`t care. If I had a really big monitor - I`d try to stay less than f16. (again, with your specified camera)

Originally Posted by chicagojohn

can a macro lens allow one to stop down further before adverse diffraction softening overcomes DoF?

No, diffraction isn`t dependant upon close focusing, Sometimes you just have to accept it (softness from diffraction) to get acceptable DoF.

To create a rule of thumb, Diffraction Limited Photography: Pixel Size, Aperture and Airy Disks is a pretty good site, go to the list of cameras, and the diagram with the square. Use the pixel size, and try to keep the airy disk radius to about 2x the pixel diameter (or 3 if you`re less demanding, 2 or 3 is usually ok) That should give you a good general rule of thumb. You can also do the calculations there. Don`t get to caught up in thinking about diffraction, just remember that it can soften your image and if you wanted a really sharp landscape you probably don`t need f16 or 22 to do it, often 8 and 11 will get you there too. I personally prefer to leave the background slightly softer than the foreground / subject, because the softness is a depth cue - things further away tend to be softer. I try not to shoot hyper focally unless I have a special reason to.

[chicagojohn]

Originally Posted by ravncat

Optomization is the tricky word.....

To create a rule of thumb, Diffraction Limited Photography: Pixel Size, Aperture and Airy Disks is a pretty good site, go to the list of cameras, and the diagram with the square. Use the pixel size, and try to keep the airy disk radius to about 2x the pixel diameter (or 3 if you`re less demanding, 2 or 3 is usually ok) That should give you a good general rule of thumb. You can also do the calculations there.

Don`t get to caught up in thinking about diffraction, just remember that it can soften your image and if you wanted a really sharp landscape you probably don`t need f16 or 22 to do it, often 8 and 11 will get you there too. I personally prefer to leave the background slightly softer than the foreground / subject, because the softness is a depth cue - things further away tend to be softer. I try not to shoot hyper focally unless I have a special reason to.

Thanks for that explanation, ravncat. My pixel diameter is about 7.8 microns (61 um^2), and the website suggests I should be good at f/22 or even f/32, maybe. So I'll investigate the 2X and 3X diffraction levels on my monitor (which happens to be 28 X 52 cm and which I typically view at 71 cm.) And I'll do a few experiments with enlarged prints too at small apertures just to see where the limits are for my equipment. I would have guessed one diameter instead of two or three, so you've given me a better starting point to work with.

I'm more interested in getting "optimum" DoF in macro focus, and it's also helpful to know that a 1:1 macro lens doesn't offer any inherent advantages in that connection with respect to reduced diffraction at any given stop.

That is a great website referral!

[ravncat]

It`s two or three because the peaks have to seperate and you have the bayer pattern to deal with, so individual pixels are interpolated from surrounding data - if you were using something like kodak`s dcs 760m, with it`s monochrome sensor and no AA filter and no bayer filter and no demosaicing, I think you`d be down to 1 pixel width.

[daft_biker]

Originally Posted by chicagojohn

I'm more interested in getting "optimum" DoF in macro focus, and it's also helpful to know that a 1:1 macro lens doesn't offer any inherent advantages in that connection with respect to reduced diffraction at any given stop.

Optimum DoF is a variable depending on the shot as sometimes you might be prepared to stop down more than others. Some subjects require more DoF than others, sometimes you'll wan't loads of detail in a specific part and some people have different ideas of what is good enough in terms of sharpness and detail. Working out the right aperture for the shot is part of the fun.

Whilst I agree that a macro lens doesn't reduce diffraction I would suggest that not all lenses are created equal and in practice you may well find you can live with stopping down further on higher quality lenses. It's not that better lenses suffer less diffraction, but even with a wee bit diffraction better lenses still seem to be better.

Generally speaking I can live with stopping my prime lenses down further than I can my zooms. For example f/22 on my 28-105mm is absolutely dreadful and I doubt I'd ever want to use it but f/22 is OK on my TS-E 90mm and I use it whenever I feel like it. The laws of physics don't change when I change lenses, the 90mm is just way better than the zoom.

[chicagojohn]

Originally Posted by daft_biker

Optimum DoF is a variable depending on the shot as sometimes you might be prepared to stop down more than others. Some subjects require more DoF than others, sometimes you'll wan't loads of detail in a specific part and some people have different ideas of what is good enough in terms of sharpness and detail. Working out the right aperture for the shot is part of the fun.

Whilst I agree that a macro lens doesn't reduce diffraction I would suggest that not all lenses are created equal and in practice you may well find you can live with stopping down further on higher quality lenses. It's not that better lenses suffer less diffraction, but even with a wee bit diffraction better lenses still seem to be better.

Generally speaking I can live with stopping my prime lenses down further than I can my zooms. For example f/22 on my 28-105mm is absolutely dreadful and I doubt I'd ever want to use it but f/22 is OK on my TS-E 90mm and I use it whenever I feel like it. The laws of physics don't change when I change lenses, the 90mm is just way better than the zoom.

Thanks again ravncat and daft biker. The lens in question is a macro prime lens and coincidentally happens to be 90mm focal length One thing I've read about since, and hadn't considered, is that some or all of the diffraction induced softness even at an f/32 can be removed by mild sharpening in PS. I've seen convincing examples of that.

So at ravncat's suggestion, I think that for now in macro I may decide to err on the side of DoF to minimize CoC, and deal with Airy discs later with PS when needed.

Thanks so much to you both, as you each had great points to make, and my understanding of the fundamental principles has been eniched as a result.