The Astigmatism You Didn't Order: How "Simple" Filters Add Aberrations

The 30-Second Version: Put glass in a beam and it will do something. Even "simple" filters add aberrations—and tilting them makes everything worse.

Glass Is Never Passive

People talk about filters like they're passive accessories. Pick the right wavelength, mount it in the optical path, done. The filter just sits there and does its spectral magic.

Optics disagrees. Optics is a mischievous gremlin: put glass in the beam and it will do something.

The technical reality is stated plainly in optical engineering references: all inserted light filters act like a plane-parallel glass plate, changing the optical path—and "not only change the optical path, but also cause artificial astigmatism (aberration)."

What Astigmatism Actually Does

Astigmatism means your optical system focuses differently in two perpendicular directions. Imagine a point of light that should focus to a crisp dot. With astigmatism, that point smears into a little line. Rotate your focus and the line rotates too—at one focus distance it's horizontal, at another it's vertical, and there's no setting where it's a perfect dot.

In practical terms, edge sharpness starts to feel weirdly directional. Vertical edges might look crisp while horizontal edges look soft (or vice versa). Fine details that your lens should resolve become ambiguous. Your MTF (modulation transfer function) drops in one direction more than the other.

You didn't ask for this. You just wanted to filter some wavelengths. But the filter is glass, glass bends light, and here you are.

Tilt Makes Everything Worse

If inserting a filter square to the optical axis introduces some astigmatism, tilting the filter gives the gremlin a second hat.

A filter tilted relative to the optical axis introduces two additional problems. First, more astigmatism—on top of whatever the flat filter was already contributing. Second, a parallel offset of the optical axis itself, because the two glass surfaces refract the beam like a weak prism.

The physics gives us a paraxial (small-angle) relationship: the lateral offset v equals [(n-1)/n] × d × ε, where n is the refractive index, d is the filter thickness, and ε is the tilt angle.

The practical implication: even tiny tilts can matter more than you'd expect, especially when your imaging is sensitive to alignment and your tolerances are measured in fractions of a pixel.

A 2° tilt on a 3 mm filter with n = 1.5 gives you a lateral offset of about 35 micrometers. That might not sound like much—but if your pixel size is 5 micrometers, you just walked the beam by 7 pixels. In applications requiring precise alignment, that's not nothing.

When Would You Intentionally Tilt?

Here's the twist: sometimes tilting a filter is actually a useful trick.

Flat glass surfaces reflect some light (we'll cover ghosting in detail in another post). Those reflections can bounce back through your optical system and create ghost images—faint displaced echoes of bright features in your scene. One way to kick those ghosts out of your field of view is to slightly tilt the filter, so the reflected light doesn't make it back to the sensor.

It works. But you're trading ghosting for astigmatism and lateral beam walk.

Optics never takes a bribe without charging interest.

If you intentionally tilt to suppress ghosts, keep the tilt minimal. Verify that you didn't create an astigmatism problem while solving your ghosting problem. And measure—don't assume.

Mitigation Strategies

The best defense is boring, careful practice.

Use optically flat, quality filters. Cheap filters with poor surface flatness contribute their own wavefront errors before you even consider the inherent effects of plane-parallel glass.

Keep filters perpendicular to the optical axis. No tilt means no tilt-induced aberration. This sounds obvious, but sloppy mounting happens more often than you'd think.

Check your retaining rings. A filter can arrive perfectly flat and end up tilted because the retaining ring wasn't seated evenly, or the mount has a slight angle, or someone over-tightened one side.

Minimize thickness. Less glass means less opportunity for trouble. If you can get the spectral performance you need from a thinner filter, that's less aberration.

Minimize stacking. Each additional filter is another plane-parallel glass element. If you're stacking three filters, you're stacking three sets of aberrations.

Test after adding filters. Don't assume your lens still performs to spec after you add glass in front of it. Verify.

The Bottom Line

Your lens is already working hard to deliver sharp images. The optical designer balanced aberrations, optimized coatings, and engineered that lens to meet a specific performance target. Every piece of glass you add is another opportunity for optical mischief.

A filter isn't just a spectral gate. It's an optical element with physical thickness, refractive index, surface flatness, and mounting geometry—all of which affect your image quality.

Respect the filter as an optical element, not just a wavelength selector. Choose quality glass, mount it carefully, and verify your system still performs the way you need it to.

This post is part of KUPO's technical education series on optical filters for machine vision. Questions about filter selection for your application? Contact our optical engineering team.