Apparent image distortion vs critical focus plane

[ray_parkhurst]

So I am curious about a phenomenon that we probably all see, and I've had questions about since I started with photography. As focus is changed, there is what looks like a distortion of the subject. I have generally just ignored the phenomenon, since it is not a factor for single images, and when focus stacking only the in-focus elements are composited. But indeed I would think this is due to small changes in perspective, and the resulting change in specular highlighting.

I made a small animation showing the effect of small focus shifts, see below. The subject is Lincoln's jaw on a 1954-S uncirculated Cent. It's made from 4 images taken at slightly different focal planes, sort of like a stack, but within the depth of field. Mag is ~0.7x. The crops were taken from center of frame, and the lens was a 105PN at f2.8, so no expectation of CA issues. There is also no real change in perspective between shots, or angle of lighting, since the working distance is ~10in, and I'm only moving focus ~20um per step.



I'm sure there is a good explanation for this phenomenon, I just don't know what it is.

Edited to add:

Here's a 200% crop of the area that seems to distort the most:

[rjlittlefield]

This is one of the "utilized aperture" effects.

What happens is that due to a combination of specular illumination and surface texture, the aperture of the lens is not evenly illuminated. That is, the light forming the image of any particular subject feature comes predominantly through one part of the aperture, and not other parts.

If that utilized part of the aperture happens to be off-center, then as you change focus whatever subject features are being imaged by that part will move sideways, even while those features stay within the depth of sharp focus.

The utilized portion of the aperture varies for different parts of the subject, being mostly driven by the slope of the surface with respect to the locations of the light and the lens.

As a result, it's very common for features on one side of a ridge to move one way as focus changes, while features on the other side of the ridge move in some different direction.

That different movement on opposite sides of the ridge ends up giving the perception of tilting, very much as if the center of perspective were moving.

The wider the entrance cone, the more opportunities there are for this mischief to occur. Arrays of crystals seen through 10X and higher microscope objectives are prone to wiggle around like crazy as focus is changed.

--Rik

[Chris S.]

Ray,

I'm pretty sure it's the utilized aperture effect as discussed here.

Not that you are moving the light, as is the case in Steven's demo. But with your preferred subjects, it's challenging to illuminate them such that they produce a completely spherical wavefront, and as you change focus, different portions of the lens' entrance cone are filled with light. This makes it appear that features of the subject are moving as you focus in and out.

Edit to add: I see Rik typed faster than I did.

--Chris S.

[rjlittlefield]

Edit to add: I see Rik typed faster than I did.

Nah, I just happened to get started sooner...

--Rik

[ray_parkhurst]

Hmm, going two levels deep in the links, I see a thread started by Rik with similar animations that I actually replied to back in 2013!! I guess it's true that memory has a 5-year half-life if not reinforced. Thanks guys for the reminder about the earlier posts and explanations.

[Lou Jost]

Ray, are you thinking that this might be connected to the artifacts you saw during pixel-shifting on these same kinds of shiny coins? Seems like there might be some relation.

[ray_parkhurst]

Ray, are you thinking that this might be connected to the artifacts you saw during pixel-shifting on these same kinds of shiny coins? Seems like there might be some relation.

Yeah, that was my original thought. The areas where I saw the odd artifacts were areas of specular highlights.

In the pixel shifting action, no focus changes are occurring, and I can't imagine any lighting changes either. The change in utilized aperture also seems very small, but perhaps big enough to make a difference? I shot at widest possible aperture since it does little good to pixel shift if the lens is not resolving the sub-pixel details. Maybe if I shot with a smaller aperture, or with more diffuse lighting, the effect would mitigate, but how useful would the final image be?

[Lou Jost]

Since the lens doesn't move, I agree, it is hard to see how this would matter. I also always shoot at the widest aperture possible, for the reason you mention.