I tried grinding a custom lens out of acrylic glass (diameter 20mm, focal length 30mm). Not very surprisingly, the result is a quite poor quality lens. I think it mostly suffers from geometry and material defects, while the surface quality is relatively good.

However I'm confused about the difference I observed between two configurations:

Configuration 1

I got very poor image quality when I tried to use it as a single-element camera lens:

Single lens diagram

(Image by DrBob, GFDL, https://commons.wikimedia.org/w/index.php?curid=9690292)

I tried to add an aperture to increase the image quality, but it didn't help much, just made the image darker.

Configuration 2

However, I got relatively ok image quality when I placed it in front of my regular camera lens, as a sort-of close up filter (lens 1 in diagram below):

Close up filter diagram

(Image by Tamasflex, CC BY-SA 3.0, https://commons.wikimedia.org/wiki/File%3AClose-up.png)

Why is the image quality better in the second case? Is there a general way to estimate the relative effect on quality of each lens element?

  • $\begingroup$ In front of the camera your lens didn't have to do the imaging, it just modified the effective focal length of the camera lens. If you trace the image rays, you will find that the rays trough each point on the focal plane will go trough a relatively large area of your lens. This averages out the errors. The detailed calculation can be done the same way lens systems are calculated, either by raytracing or with one of the optical matrix methods: en.wikipedia.org/wiki/Ray_transfer_matrix_analysis, which can include the calculation of optical errors. $\endgroup$
    – CuriousOne
    Commented Apr 21, 2016 at 7:20
  • $\begingroup$ @CuriousOne So would I be correct in assuming that the lens closest to the sensor is usually the most important, as that is the one that has "to do the imaging"? $\endgroup$
    – jpa
    Commented Apr 21, 2016 at 7:28
  • 1
    $\begingroup$ I can't tell you which lenses contribute the most to the wavefront errors, but in general the rays converge closer to the focal plane, so there is less of the averaging going on (and that's where e.g. scratches and dust will show really badly). Also in designs with aspherical lenses and with lenses with large curvature, like the next to last in your picture, that's where I would suspect the most critical surfaces. From a production cost point of view a "good" design will probably distribute the total error budget as evenly over as many surfaces as possible. $\endgroup$
    – CuriousOne
    Commented Apr 21, 2016 at 7:33


Browse other questions tagged or ask your own question.