mp-e 65mm diffraction question

[Jon Ruyle]

Ever since I started doing macro photography three years ago, I've wanted the 65mm 1-5x macro lens. Recently, though, I started thinking about diffraction, and concluded that using this lens only gives a slight advantage over using an ordinary macro lens, then cropping.


I'll explain what I mean (and if I'm wrong, I'd be happy if someone would set me straight). Right now I use a 100mm macro, and with extension tubes I can get 2x. If I want to take a picture at 4x, I could crop to half size, but then of course my linear resolution would be cut in half.


Compare that with getting 4x directly with the mp-e. Effective f number is f times (1 + magnification) (I think), so that means that at 2x, my diffraction circles are 3 times as big as they would be at infinity (and after I crop, they are 6 times as big). At 4x without cropping, they are 5 times as big.


In other words, using the lens at 4x only gives a 20% linear improvement over using the lens I have at 2x and cropping. I'm assuming, of course, that diffraction is the limiting factor, which I think is reasonable beyond 2x.


Is that right? Even if it is, it doesn't mean the lens isn't worthwhile (plenty of people pay bigger bucks for even more marginal improvements). It just means it won't help as much as I thought.

[Daniel Browning]

In other words, using the lens at 4x only gives a 20% linear improvement over using the lens I have at 2x and cropping. I'm assuming, of course, that diffraction is the limiting factor, which I think is reasonable beyond 2x.





If you shoot wide open, then 5x is only effectively f/17 (2.8 * (1 + 5)). The image will be affected by diffraction at that f-number, sure, but not so much that cropping would be better. Keep in mind that even after diffraction gets really bad, there are still many benefits to increased pixels on the subject: first, you can restore a lot of resolution with sharpening (as long as you shoot low ISO to keep noise down); second, the frequency at which the OLPF (anti-alias filter) interacts with details will be higher (i.e. down in the diffraction softness where it can't do any harm to important content), so contrast/resolution will improve for that reason, and finally, because the bayer pattern causes color resolution to max out at a higher frequency than luma (green), you'll get higher color resolution (usually not important to the eye), but more importantly, fewer debayer/demosaic artifacts (which are sometimes visible).

[Jon Ruyle]

Thanks, Dan.


Okay, wide open, we're close. But I don't see myself using it wide open... I rarely even do that with the ordinary macro lens. My guess is I would rarely use it wider than f/8 (most photos I've seen taken with it are stopped all the way down to f/16) Even at f/8 and at 2x, diffraction are about the size of a bayer array (2x2). At this point, do you think the aa filter and demosaic artifacts are still relevant?


Maybe I should just stop whining and get the lens. [:)]

[Dann Thombs]

I don't know the technical details, but the resolution I get at 5x and f/16 is still excellent. Bring it down to 7.1 and it's even better. Plus a twist of the barrel and you go from one extreme to another, no extension tubes and whatnot to bother with.


Like you said, just get it already :)

[Dann Thombs]

A few shots and such:


http://www.flickr.com/photos/goshzilla/2456868532/ ("http://www.flickr.com/photos/goshzilla/2456868532/)


http://www.flickr.com/photos/goshzilla/2496132690/ ("http://www.flickr.com/photos/goshzilla/2496132690/)

[Daniel Browning]

At this point, do you think the aa filter and demosaic artifacts are still relevant?


I'm not sure. Have you considered focus stacking? Helicon Focus is practially magic: just snap a bunch of shots at f/2.8, throw them into the program, and it comes out with very deep depth of field. No artifacts, no touching up in photoshop necessary. CombineZ and Tufuse Pro are other options, too.

[Dann Thombs]

CS4 is very good too, and they improved on the latest update. I'd stack at 5.6 though. 2.8 will take forever.

[Jon Ruyle]

Dann:


Those are great pics! Exactly the type that makes me want the lens. Of course, it is a common trap to see someone else's great pictures and want the same equipment... Still, for me, this is a hobby. It's all about having fun. It strikes me that if I'm spending a lot of time sticking extension tubes on my 100 macro, it is time for the mp-e... I was just a bit worried that due to diffraction, my images with the mpe would be only modestly improved over those with the 100. There's one way to find out, I suppose :)


Dann and Daniel:


I didn't know about this focus stacking stuff. I spoke to a friend who knows a bit about image processing and he's thinks he'll try implimenting this on his own. He's asked me to take some pics for him to process.


Another thing he suggested was deconvolving diffraction, which only requires a single exposure and is a much simpler process. Have either of you tried this?

[Daniel Browning]

Another thing he suggested was deconvolving diffraction, which only requires a single exposure and is a much simpler process. Have either of you tried this?


I've only done it for astrophotography, where noise makes the deconvolution less effective. If you can shoot low ISO, I would imagine that the kernels for deconvolving diffraction are more effective than a typical "sharpening" process. I don't know of what software to recommend for it, the astro stuff is probably the wrong tool.


I highly suggest you checkout the forums at http://www.photomacrography.net/ ("http://www.photomacrography.net/) -- it's the single biggest concentration of macro expertise on the web.

[Jon Ruyle]

Thanks, Daniel. I just checked out photomacrography. Looks like another time sink... :)


Right. With astro you have noise, in which case (I'm told) something like maximum entropy should work better (I have no experience with this, though). And personally, my astro images have so many other problems that diffraction isn't even an issue.


Maybe for planetary photography deconvolution would help (f/ is high and light is plentiful.)

[Daniel Browning]

I didn't know about this focus stacking stuff. I spoke to a friend who knows a bit about image processing and he's thinks he'll try implimenting this on his own. He's asked me to take some pics for him to process.


There are some good guides for focus stacking on the web (GIYF). One pitfall to avoid is turning the focus ring on the lens: that is hard to move in the right increments and will also introduce breathing and other problems (apocryphal focus). It's better to use a focusing rail, bellows, or move the subject itself.

[Jon Ruyle]

Interesting.


The guy I spoke to said he wanted me to change focus, not move the camera.


Could you explain what you mean by "breathing" and "apocryphal focus"?

[Daniel Browning]

The guy I spoke to said he wanted me to change focus, not move the camera.


That should be fine.



Could you explain what you mean by "breathing" and "apocryphal focus"?


Breathing is when turning the focus ring causes the focal length of the lens to change. In this way, prime lenses (even macro) "zoom" as they focus. It's possible to create a lens that compensates for the focal length change so that changing focus does not change focal length, but this is not usually done except in expensive cinema lenses. As you can imagine, changing focal length slightly as you focus will cause a slight difference in magnification between layers in the stack. CombineZ can compensate for that without introducing artifacts as long as the changes are slight enough.

Apocryphal focus is a term that I learned from Joseph S. Wisniewski, a talented macrophotographer. He informed me that it causes exaggerated perspective (i.e. the same effect you would get from moving in closer and using a wider focal length) and that many find the resulting perspective to be unnatural. I don't know why using the focus ring causes it, except that perhaps it's the same thing as breathing. (Obviously, I'm not an expert in macro.)

If you use a bellows that can separate the movements between just the rear standard by itself (camera without lens) and everything (rear + front standard together), then it's possible to take layers that don't each have a different perspective.

The first possibility is by moving just the rear standard. The perspective will correspond to whatever the focal length of the lens is, but at least it wont change like it does when using the focus ring.

The second possibility is to move everything together (rear and front standard both). This results in "orthographic" perspective, which is as if the object were at infinity (i.e. flat or no perspective at all). Many macrophotographers desire orthographic perspective the most.

Hope that helps. :)

By the way, I can't stress enough how important it is to try a dedicated stacking program. Compared to shooting the layers in the first place, learning a new software program is a piece of cake.

[Dann Thombs]

When using the MP-E and stacking, just move closer to the object. The perspective doesn't change all that much, and the actual movement is so little, that it makes little difference. Here's a quick example using my handy dead wasp:


http://farm4.static.flickr.com/3223/2980798239_9a6dcdb41c.jpg





Which then ends up like this:


http://farm4.static.flickr.com/3210/2970786082_eaf21a5720.jpg

[Jon Ruyle]

The deail on that pic is pretty impressive... (unless, of course, that dead wasp is unusually large [:)])


I'll have to try this as soon as I have a chance.


Daniel: Thanks for the explanations. WRT breathing, you don't mean that the focal length really changes, do you? I'm guessing you're talking about how the framing changes rapidly because magnification increases asymptotically as you approach focal distance, which sort of looks like zooming. Or *do* you mean that the focal length changes? (It wouldn't for an ideal lens, but I don't understand what goes on with complicated real life things)


Thanks both of you for turning me on to this.

[Daniel Browning]

WRT breathing, you don't mean that the focal focal length really changes, do you?


Yes, I do. The numbers printed on lenses are really more like guidelines (usually corresponding to infinity focus) than actual reality. When you change the focus ring, it literally changes the focal length. The lens elements physically move. A 100mm focused on infinity could be 110mm focused at 1 meter.

[Jon Ruyle]

Wild. Actually, I've noticed that the 100mm macro doesn't obey the lens equation with f=100 wrt extension tubes. So maybe I shouldn't be surprised.