Upgrading narrowband filters: is it worth switching from 7 mm to 3nm?

How much light pollution are you dealing with? If you’re not at a dark site, then generally the narrower the better. And, even at a dark site, narrower will allow better imaging with the moon out.

I have the Antlia Edge 4.5nm filters as well as Chroma 3nm. They are both solid with no halos. I used to use the Zwo 7nm and, yes, they have halos on Oiii, but they were inexpensive. To me, the Antlias are the best bang for the buck, with high transmission and no halos. Chromas are simply the best, but cost as such.

Dave Stirling · May 28, 2026, 03:03 PM

Nikhil · May 28, 2026 at 05:02 AM

Hello all.

With galaxy season coming to a close, I've started a narrowband project for my monochrome camera. With only preliminary data, I noticed a severe problem with my ZWO OIII 7 mm filter. It seems to produce terrible halos around bright stars (I’ve included the images below) and this is apparently a very common problem with this filter. I've upgraded to the Antlia OIII 3 nm filter and I'm waiting for it to arrive.

I wanted to ask experienced people in the community, do you think it's worth it to upgrade my other Ha and SII 7 nm narrowband filters while I'm at it? Those filters don't seem to produce artifacts but I'm mainly asking with regards to the benefit of going to a smaller bandpass.

Thanks very much for your input!

📷 1000012068.jpgfirst image description: HOO image of zeta oph. Very noticable halo around the bright central star.

📷 1000012062.jpgsecond image description: same as first image but only Ha data. No noticeable artifacts.

I think halos are a function of the quality of the filter rather than the bandpass. You would likely see improvement in halos around OIII specifically with the higher quality filter, so it would be interesting to see how much halos diminish by swapping out the OIII with the new filter you’ve ordered first. Highly recommend 3nm for high light pollution (I’m in Bortle 6).

That would make sense that halos depend on filter quality.

Judging from your’s and the previous commenter’s inputs it seems that a 3 nm bandpass won’t make too much of a difference since I am at a dark site.

Thanks for sharing your experience!

Tighter filter bandpass = better contrast = longer exposures = you’ll need better guiding

Most all NB targets are comprised of Ha therefore you can get a quicker, faster final stacked result just shooting/stacking a lot of Ha subs regardless of LP or the Moon (as long as it’s >60 degrees from target).

As an AP rule of thumb fewer targets have [OIII] and even fewer have [SII].

Shooting through LP and the narrower bandpass means your data will be very faint unless the target is rich in that gas. Since the data will be fainter from the tighter bandpass you’ll need to gather more or longer subs.

Depending on your camera you’ll want to adjust your cameras gain and offset given the fainter data - increase gain to increase your SNR and swamp the noise and increase your offset to bump your signal off the black (left) side of your histogram.

I shoot a QHY268M Pro at Gain 56 and Offset 100 when shooting narrowband. My Antlia filters are all 2” mounted - Ha (3nm), [OIII] (3.5nm), [SII] (3nm). Sub duration is either 180s or 300s depending on target, LP and moon pollution.

Clear Skies and God Bless!

Kip

Nikhil · May 28, 2026 at 08:03 PM

Dave Stirling · May 28, 2026, 03:03 PM

Nikhil · May 28, 2026 at 05:02 AM

Hello all.

With galaxy season coming to a close, I've started a narrowband project for my monochrome camera. With only preliminary data, I noticed a severe problem with my ZWO OIII 7 mm filter. It seems to produce terrible halos around bright stars (I’ve included the images below) and this is apparently a very common problem with this filter. I've upgraded to the Antlia OIII 3 nm filter and I'm waiting for it to arrive.

I wanted to ask experienced people in the community, do you think it's worth it to upgrade my other Ha and SII 7 nm narrowband filters while I'm at it? Those filters don't seem to produce artifacts but I'm mainly asking with regards to the benefit of going to a smaller bandpass.

Thanks very much for your input!

📷 1000012068.jpgfirst image description: HOO image of zeta oph. Very noticable halo around the bright central star.

📷 1000012062.jpgsecond image description: same as first image but only Ha data. No noticeable artifacts.

I think halos are a function of the quality of the filter rather than the bandpass. You would likely see improvement in halos around OIII specifically with the higher quality filter, so it would be interesting to see how much halos diminish by swapping out the OIII with the new filter you’ve ordered first. Highly recommend 3nm for high light pollution (I’m in Bortle 6).

That would make sense that halos depend on filter quality.

Judging from your’s and the previous commenter’s inputs it seems that a 3 nm bandpass won’t make too much of a difference since I am at a dark site.

Thanks for sharing your experience!

I’m envious of your dark site! Looking forward to some posts with the new OIII filter.

SoDakAstronomyNut · May 28, 2026, 08:23 PM

Tighter filter bandpass = better contrast = longer exposures = you’ll need better guiding

Most all NB targets are comprised of Ha therefore you can get a quicker, faster final stacked result just shooting/stacking a lot of Ha subs regardless of LP or the Moon (as long as it’s >60 degrees from target).

As an AP rule of thumb fewer targets have [OIII] and even fewer have [SII].

Shooting through LP and the narrower bandpass means your data will be very faint unless the target is rich in that gas. Since the data will be fainter from the tighter bandpass you’ll need to gather more or longer subs.

Depending on your camera you’ll want to adjust your cameras gain and offset given the fainter data - increase gain to increase your SNR and swamp the noise and increase your offset to bump your signal off the black (left) side of your histogram.

I shoot a QHY268M Pro at Gain 56 and Offset 100 when shooting narrowband. My Antlia filters are all 2” mounted - Ha (3nm), [OIII] (3.5nm), [SII] (3nm). Sub duration is either 180s or 300s depending on target, LP and moon pollution.

Clear Skies and God Bless!

Kip

I totally forgot about gain and offset, thanks for bringing this up! I wonder what difference there will be with this in mind. I’ll tinker with it when these clouds clear up, thanks!

Clear skies!

Nikhil · May 28, 2026, 08:44 PM

SoDakAstronomyNut · May 28, 2026, 08:23 PM

Tighter filter bandpass = better contrast = longer exposures = you’ll need better guiding

Most all NB targets are comprised of Ha therefore you can get a quicker, faster final stacked result just shooting/stacking a lot of Ha subs regardless of LP or the Moon (as long as it’s >60 degrees from target).

As an AP rule of thumb fewer targets have [OIII] and even fewer have [SII].

Shooting through LP and the narrower bandpass means your data will be very faint unless the target is rich in that gas. Since the data will be fainter from the tighter bandpass you’ll need to gather more or longer subs.

Depending on your camera you’ll want to adjust your cameras gain and offset given the fainter data - increase gain to increase your SNR and swamp the noise and increase your offset to bump your signal off the black (left) side of your histogram.

I shoot a QHY268M Pro at Gain 56 and Offset 100 when shooting narrowband. My Antlia filters are all 2” mounted - Ha (3nm), [OIII] (3.5nm), [SII] (3nm). Sub duration is either 180s or 300s depending on target, LP and moon pollution.

Clear Skies and God Bless!

Kip

I totally forgot about gain and offset, thanks for bringing this up! I wonder what difference there will be with this in mind. I’ll tinker with it when these clouds clear up, thanks!

Clear skies!

I use the same gain for everything. For QHY 56 gain at High Gain 2CMS mode, for Zwo and player one, set = to high gain mode point (Zwo=100 P1=125). All about equivalent.

I keep luminance and RGB shorter (2 minutes or so) and SHO longer (5-10 minutes) for scopes ranging from f2.7 to f7. Some people set the gain to 0 for broadband and take longer subs, but I’ve done the comparisons, and found it’s just not worth it. Need to have multiple sets of darks/biases, remember to switch…and then having fewer subs which makes satellite track and other outlier rejection not as effective.

Can it be every so slightly more optimal to switch gains around? Sure. Could avoid saturating stars in theory, but if you’re at a dark site and need to expose a long time to get background over read noise, it nullifies most of that advantage.

And if you only use one gain, you only need one offset.

Rick Veregin · May 28, 2026, 11:59 PM

Regarding the question is a 3 nm filter better for narrowband?

• If >= f6, assuming both filters have equal efficiency at the NB wavelength, generally a good assumption.

  • Contrast (NB signal/background) will always improve, inversely proportional to the bandwidth ratio. For example 7 to 3 nm, improves contrast about 7/3 = 2.3X

  • Iff also noise is background signal dominated, that is sqrt(background in e) » read noise, then S/N also improves by sqrt(7/3)=1.5X. This is true at 3 nm or greater bandwidth typically for 1 to 5 min subs, depending on the Bortle. For dark sites this requires a very low noise cameras and long subs typically. So for sure, 3 nm is more important for S/N improvement with more light pollution. If your noise is limited by read noise, there is no S/N improvement with 3 nm.

  • Note though, for NB targets in our galaxy all this is true, as well as galaxies in our local neighborhood, but as you go to 20 to 30 million light years for galaxies, the red shift will move NB wavelengths out of the optimum range at 3 nm, you will loose signal. For distant galaxies, 7 nm or more will be better for sure.

• If < f6, at 3 nm the efficiency drops, especially at the edge of the field, so you will start to loose signal and any advantage of the 3 nm filter. This is due to the shift in wavelengths due to the higher incident angle.

• For <= f4, you need to purchase special filters that are wavelength shifted to compensate, or just stay at 7 nm or higher for your NB filters, to accommodate the big wavelength shift.

Hope this helps

Rick

It does, thank you. This brand advertises that it is suitable to f >= 3, which is suitable for my setup.

On your note about contrast, I was figuring that all else being equal, S/N would scale as 1/sqrt(bandpass), since I’m at a dark site and I’m either photon-noise or background-noise limited, and I get a similar figure as yours. But I don’t know how this would look qualitatively, since I’ve only ever worked with 7 nm bandpass filters. I guess that would equivalently mean I would have to spend less time on a target if I got more advanced filters.

As far as cosmological implications for galaxies at larger distances, I think I would be limited to about 10 Mpc in Ha if I got the 3 nm filter, which is worth considering for some targets.

So maybe I should only upgrade the SII filter since it’s emission is so faint and I’m only going to use it for nearby nebulae.

Nikhil · May 29, 2026, 12:22 AM

Rick Veregin · May 28, 2026, 11:59 PM

Regarding the question is a 3 nm filter better for narrowband?

• If >= f6, assuming both filters have equal efficiency at the NB wavelength, generally a good assumption.

  • Contrast (NB signal/background) will always improve, inversely proportional to the bandwidth ratio. For example 7 to 3 nm, improves contrast about 7/3 = 2.3X

  • Iff also noise is background signal dominated, that is sqrt(background in e) » read noise, then S/N also improves by sqrt(7/3)=1.5X. This is true at 3 nm or greater bandwidth typically for 1 to 5 min subs, depending on the Bortle. For dark sites this requires a very low noise cameras and long subs typically. So for sure, 3 nm is more important for S/N improvement with more light pollution. If your noise is limited by read noise, there is no S/N improvement with 3 nm.

  • Note though, for NB targets in our galaxy all this is true, as well as galaxies in our local neighborhood, but as you go to 20 to 30 million light years for galaxies, the red shift will move NB wavelengths out of the optimum range at 3 nm, you will loose signal. For distant galaxies, 7 nm or more will be better for sure.

• If < f6, at 3 nm the efficiency drops, especially at the edge of the field, so you will start to loose signal and any advantage of the 3 nm filter. This is due to the shift in wavelengths due to the higher incident angle.

• For <= f4, you need to purchase special filters that are wavelength shifted to compensate, or just stay at 7 nm or higher for your NB filters, to accommodate the big wavelength shift.

Hope this helps

Rick

It does, thank you. This brand advertises that it is suitable to f >= 3, which is suitable for my setup.

On your note about contrast, I was figuring that all else being equal, S/N would scale as 1/sqrt(bandpass), since I’m at a dark site and I’m either photon-noise or background-noise limited, and I get a similar figure as yours. But I don’t know how this would look qualitatively, since I’ve only ever worked with 7 nm bandpass filters. I guess that would equivalently mean I would have to spend less time on a target if I got more advanced filters.

As far as cosmological implications for galaxies at larger distances, I think I would be limited to about 10 Mpc in Ha if I got the 3 nm filter, which is worth considering for some targets.

So maybe I should only upgrade the SII filter since it’s emission is so faint and I’m only going to use it for nearby nebulae.

Great that your filter is suitable for your setup.

Please note that contrast and S/N are not the same thing.

Contrast is just the the signal to background ratio. Since a NB filter reduces background, this scales as inverse bandwidth.

Signal to noise is either Signal/(Read noise+dark thermal noise) or Signal/sqrt(Background signal, which is the Poisson noise). Or if you are in between, the combination of the two. Not sure what you mean by photon noise, if you mean Poisson noise, then that is the signal/sqrt(background signal), where you are background limited for noise.

So again, either you are background limited or read noise limited for noise. If you are read noise limited, a narrow band filter does not affect that camera noise, you are limited by your camera. A filter only affects light coming in, not what your camera does to it. Mostly at dark sites you are likely read noise limited unless you are using exceptionally long exposures. So yes, you get a contrast benefit, but read noise limited means the filter will not help you at all for S/N, you are limited by your camera.

The SII channel is no different, if you are read noise limited in dark skies, 3 nm will not help S/N. You will get a contrast benefit though, which is important.

The basic message again is you will always get a contrast improvement as long as you filter efficiency at the NB wavelength is okay. But in dark skies, you likely will not get a S/N improvement.

Note, easy to determine if you are read noise or background sky noise limited. You know your read noise from your camera specs for your gain setting. Measure your background signal (or take the median signal over the whole frame, that is usually pretty close to the background signal), and convert to electrons.

Your read noise contribution in % is then: 100*Rn/[sqrt(Background +Rn²).

Rick