Precise positioning of apertures for telecentricity

[Lou Jost]

As Rik has explained elsewhere on this forum, lenses can be made telecentric if an aperture is placed in the image plane when the lens is focused at infinity. For reversed lenses this plane can be found by aiming the back of the lens at a subject infinitely far away and projecting the image on a sheet of paper (or tracing paper if the image is close to the rear of the lens). When the image is in focus, you've found where the aperture has to be. But how to support the aperture in exactly the right place? If the lens is going to be reversed on the camera or on anther lens, it's tricky to position the aperture with precision.

I find that this becomes easy if I use nested step-up and step-down rings on the lens. Here are two examples, shown first from above and then from below:





The one on the right consists of multiple rings with the inner ring free to screw in or out with great precision, thus raising or lowering it to the right place for the aperture. I then put the aperture on this ring. Then I put a male-male ring on top to connect to the next lens or the camera.

The one on the left doesn't have a shelf-ring added. I use that one to get objectives closer to the tube lens than they would be using an ordinary single ring.

[Lou Jost]

The left version is for a lens with a filter thread of 62mm. The front image plane is very close to the lens. I used a 62mm-55mm step-down ring, then a 55-52mm step down ring. Then I turned the pair of rings upside-down and attached a 52-49mm step-down ring to the inner threads. Then turned it over again and attached a 49-46mm step-down ring to the inner threads. That is the ring that gets adjusted to the right place for the aperture. The 52-49 ring had to be adjusted a bit also so as not to interfere with the 52mm male-male ring that goes on top of all of this.

[Lou Jost]

I've just tested a stack made with the reversed 70mm macro coupled to a Raynox 5320 tube lens. I used the above adjustable aperture-support to position the paper aperture accurately. The scale factors in Zerene are initially around 0.99998. After about 100 steps the scale factors are 0.9995. In practice, has anyone had any luck achieving values closer to 1.00000000 with home-made apertures? Is it possible to get much better accuracy?

[rjlittlefield]

The scale factors in Zerene are initially around 0.99998. After about 100 steps the scale factors are 0.9995. In practice, has anyone had any luck achieving values closer to 1.00000000 with home-made apertures? Is it possible to get much better accuracy?

"Possible" -- maybe. "Practical" -- I doubt it.

Assuming a sensor that is 10,000 pixels wide, a scale ratio of 0.99998 means that two images vary by only 0.2 pixels in total width, i.e., 0.1 pixels on either side of center.

I think it would be a very special subject and optics for which such a small difference would matter in any practical sense.

I suspect it would also require a very careful setup to even make the measurement accurately.

I tested Zerene Stacker using two images: a natural scene in its original size (3072x2048) plus a variant that was computationally scaled by 1.000010000 (1 part in 100,000).

When the source files were represented as 16-bit TIFF, Zerene Stacker was able to accurately recover the scale factor, computing it to be 1.000010265 .

However, when I simply reformatted the images to be 8-bit TIFF and then tried to recover the scale, the result was to make enlargement look like shrinkage: scale = 0.999999994 .

Given the vagaries of pixel noise, plus the well known difficulties introduced by "blooming" as subject features go out of focus, I think it would take very careful target selection, illumination, and selection of focus points to get accuracy much better than what you're seeing.

--Rik

[Lou Jost]

That's reassuring. This business of rings-within-rings gives me great precision, and I can't imagine how I could get more precision, so I am glad I don't need more.