I shoot from a fourth-floor balcony in Nicosia. Bortle 7 on a clear night, Bortle 8 when the football pitch lights are on. Every spring, someone in the astrophotography forums asks: “Which filter should I buy for light-polluted skies?” The answers range from “get an L-Pro” to “move to the countryside.” Neither is particularly useful if you’re standing on a balcony with a one-shot colour camera and two hours before the neighbour’s security light comes on.
I’ve spent the past year testing three categories of filter from this exact spot. Here’s what I’ve learned about what each one can and can’t do — and which targets are genuinely worth chasing from a bright city.
The three filter families, briefly
Filters for light pollution fall into three groups. The differences aren’t subtle.
Broadband filters (Optolong L-Pro, IDAS LPS-D2, SVBONY CLS) block specific wavelengths emitted by sodium and mercury streetlights while passing most of the visible spectrum. They’re designed to darken the background without distorting star colours. Typical bandpass: 400–700 nm with notches cut at the worst pollution lines.
Duoband filters (Optolong L-eXtreme, SVBONY SV220, Antlia ALP-T) pass only two narrow slices of the spectrum: hydrogen-alpha at 656.3 nm and oxygen-III at 500.7 nm. Bandwidths range from 3 nm to 10 nm per channel. Everything else, including light pollution, gets blocked.
Monochrome narrowband filters (Astronomik Ha 6nm, Baader OIII 6.5nm, Chroma SII 3nm) isolate a single emission line each. They’re meant for monochrome cameras shooting separate channels that get combined into a false-colour image (the SHO / Hubble palette). Not relevant to one-shot colour cameras, so I won’t cover them here.
Broadband from Bortle 7
I started with a broadband filter because every beginner guide recommends one. From a Bortle 4–5 site, broadband filters do exactly what they promise — they cut the orange sodium glow, flatten gradients, and let you stretch the histogram further before the background blows out.
From Bortle 7, the improvement is marginal. On a 120-second unguided sub through my 61EDPH II, the broadband filter darkened the background by about 0.3 stops compared to no filter at all. That’s measurable, but it’s not the night-and-day difference you see in side-by-side comparisons shot from darker skies. The reason is straightforward: modern LED streetlights emit across the full visible spectrum, not in discrete lines. A broadband filter was designed for sodium-vapour pollution. Nicosia switched most of its streetlights to LED in 2023.
Broadband filters still have a place. They’re the only option for broadband targets (galaxies, reflection nebulae, star clusters, dark nebulae) because those objects emit light across the full spectrum. A duoband filter would block most of the photons you’re trying to collect. But from a city, broadband targets are hard no matter what filter you put in front of the sensor. The best strategy for galaxies from Bortle 7 is integration time, not filtration. A lot of it. Twenty hours on M31 from my balcony produced a usable result. Three hours didn’t.
Duoband from Bortle 7
The first time I dropped an Optolong L-eXtreme into the filter drawer and pointed at the Lagoon Nebula, I thought the stacking software had glitched. The nebula structure was just there, popping out of a nearly black background, after 45 minutes of 60-second subs. Without the filter, the same target from the same balcony required 4+ hours of integration and aggressive gradient removal to get a comparable signal-to-noise ratio.
A 7 nm duoband filter passes Ha and OIII and blocks everything else. From Bortle 7, “everything else” is almost entirely light pollution. The filter doesn’t make the nebula brighter — it makes the sky dramatically darker. Your signal-to-noise ratio improves by a factor I’ve measured at roughly 3–4× per sub compared to unfiltered, depending on the target’s emission line brightness.
The catch: duoband filters only work on emission nebulae. That means HII regions (Orion, Lagoon, Eagle, Carina), planetary nebulae (Ring, Dumbbell, Cat’s Eye), and supernova remnants (Veil, Crab). These are objects that glow in hydrogen and oxygen emission lines. Galaxies, reflection nebulae, and star clusters emit broadband light. A duoband filter blocks most of it.
From Bortle 7, this trade-off is easy to accept. Emission nebulae are the only deep-sky targets I can image well from the balcony anyway. Galaxies need darker skies or enormous integration times. A duoband filter turns the balcony into a surprisingly capable narrowband station for half the sky.
7 nm vs 3 nm: does bandwidth matter?
It does, but less than you’d think.
The SVBONY SV220 comes in both 7 nm and 3 nm versions. The 3 nm version blocks more sky glow and produces slightly higher contrast on faint nebula structure — I measured about a 15–20% improvement in signal-to-noise on the Veil Nebula compared to the 7 nm version. But the 3 nm filter also requires longer exposures to accumulate the same signal, because it passes a narrower slice of each emission line. From Bortle 7, the background is still bright enough that the 3 nm version is the better bet for most emission targets. From Bortle 4–5, the 7 nm version would be more versatile.
There’s also a practical issue: very narrow bandpass filters are more sensitive to the angle of incoming light. In a fast optical system (f/4 or below), the edges of the field see a shifted bandpass that can clip the emission line. At f/5 and slower, this isn’t a concern.
What about the Seestar’s built-in filter?
The ZWO Seestar S50 ships with a built-in dual-band filter that passes OIII at 30 nm and Ha at 20 nm, considerably wider than the 7 nm of an L-eXtreme. Those wider bandpasses are a deliberate engineering choice: the Seestar’s sensor is small and its aperture is modest (50 mm), so it needs every photon it can get. A 7 nm filter on a Seestar would require impractically long integration times.
From my balcony, the Seestar’s LP filter mode produces visibly better results on emission nebulae than its no-filter mode. But it doesn’t suppress light pollution as aggressively as a 7 nm duoband on a larger scope. If you’re shooting exclusively from the city, the Seestar’s filter is good enough for bright emission targets (Orion, Lagoon, North America) but struggles on fainter ones (Veil, Rosette) where the wider bandpass lets too much sky glow through.
Three filters at three price points
If you’re shooting from a bright city with a one-shot colour camera, here’s how I’d spend the money:
SVBONY SV220 7 nm (~$85 for the 2" version). The entry point for duoband imaging. Transmission is measured at 94% on Ha and OIII. Build quality is decent — the threading is clean, the glass has good anti-reflection coatings, and the bandpass is consistent across independent tests. For the price, it’s hard to argue against starting here. The 7 nm bandwidth is forgiving on slower optical systems (f/5–f/7).
Optolong L-eXtreme 7 nm (~$200 for the 2" version). The filter that established the duoband category. Marginally better coating consistency than the SV220 in my experience, with tighter bandpass tolerances. If you’re shooting at f/4 or faster, the tighter specs reduce halos around bright stars. Whether that’s worth the extra $115 depends on your optical system and how bothered you are by halo artefacts.
SVBONY SV220 3 nm (~$130 for the 2" version). For dedicated city shooters on slower scopes (f/5 and above). The narrower bandpass cuts sky glow more aggressively, and the price is well below the Optolong and Antlia 3 nm alternatives. My go-to filter from the balcony for the Veil Nebula and similarly faint extended emission.
Targets that work from Bortle 7 with a duoband filter
Not all emission nebulae are equal from a city. Here’s what I’ve found worthwhile from Nicosia with 2–4 hours of integration through a 7 nm duoband:
Orion Nebula (M42): bright enough that it works even without a filter, but the duoband tames the sky glow and brings out the faint outer nebulosity. Best from November through February when Orion is high.
Lagoon Nebula (M8): strong Ha emitter, low in the south from 35°N but still produces clean results through the duoband. June through August.
North America Nebula (NGC 7000): large angular size, strong Ha. Needs a wide-field setup (300 mm focal length or shorter) to frame the whole complex. July through October.
Veil Nebula (NGC 6992/6960): the target where a duoband filter earns its keep. The Veil is almost invisible from Bortle 7 without narrowband filtration. With the SV220 3 nm, the filamentary structure emerges clearly after 3 hours. June through September.
Dumbbell Nebula (M27): compact planetary nebula, bright in both Ha and OIII. Easy target, forgiving of short integration times. June through October.
Galaxies, the Pleiades, open clusters, and globular clusters: save these for trips to Troodos or wherever your nearest Bortle 3–4 sky lives.
The filter that doesn’t exist yet
What I actually want is a filter that passes Ha, OIII, and the broadband continuum — switching between them electronically, no filter wheel, no mechanical parts. Something like a tunable Fabry-Pérot etalon scaled down to a 2" cell. Solar astronomers have had narrowband etalons for decades (DayStar Quark, Lunt), but those are locked to a single line and cost as much as a decent mount.
ZWO’s built-in dual-band in the Seestar points in the right direction: a filter integrated into the optical path that the software switches on or off. The next generation of smart telescopes will probably offer user-selectable narrowband presets. Until then, swapping glass filters is the price of admission for city astrophotography.
Bottom line
From a Bortle 7 balcony, a duoband filter is what makes emission-nebula imaging possible at all. Broadband filters help at the margins but can’t overcome full-spectrum LED pollution. Start with the SVBONY SV220 7 nm if you’re budget-conscious, move to the 3 nm version if you find yourself chasing faint targets exclusively from the city.
And accept the trade-off: duoband gives you nebulae, not galaxies. For everything else, you’ll need darker skies.