In the first part I made a little introduction on gain setting in CMOS cameras and also presented simple comparison made with luminance filter. In this entry I will show another comparison – but this time made with narrowband Ha 7nm filter. The difference between narrowband and LRGB filters is fundamental. Narrowband filters transmit much less light that comes both from target and from sky background. Ha 7nm filter for example passes about 2.3% of the light passed by luminance filter (assuming equal spectral distribution). It is now easy to calculate, that if one wants to achieve the same level of lightness in the frame as for luminance, you need to extend exposure time over 40 times!
Since narrowband filters pass so little of light, intrinsic noise of camera (read noise, quantization noise, thermal noise) becomes more significant.
I have chosen Sh2-112 emission nebula as a sample target. I made three series of subframes – total exposure time of each serie was 20 minutes:
- 20×1 minute, gain=100, offset=60, calibrated with 40×1 darks and flats
- 4×5 minutes, gain=100, offset=60, calibrated with 20×5 darks and flats
- 4×5 minutes, gain=0, offset=40, calibrated with 20×5 darks and flats
All images were made with Meade ACF10″ telescope, AP CCDT67 telecompressor and QHY163M camera on EQ6 mount. Sensor temperature was -18*C. Conditions: suburban sky (NELM 5mag), transparency good, seeing moderate.
Subframes were calibrated and stacked using Sum algorithm to avoid advanced stacking algorithms influence. Levels of both stacks were equalized with Linear Fit function in Pixinsight software.
First comparison shows all three stacks stretched with STF function in PixInsight:
And the same frames, but enlarged:
The differences between images are quite subtle. When you take a closer look into nebula details, especially low contrast features, then you possibly can notice some of them. The worst outcome is from 4×5 minutes stack made at gain=0. Intrinsic camera noise (read and quantization noise) are largest at gain=0 setting, and narrowband filter transmits low amount of light. That is why these noise becomes visible in the final picture.
The best effect was achieved with 4×5 minutes stack made at gain=100. When we increase gain we get rid of quantization noise, and read noise is reduced by half. The final picture is of better quality.
20×1 minute stack made with gain=100 is somewhere in the middle. We again reduce quantization and read noises, but also we add these noises 20 times to the final picture (because we have 20 subframes). That’s why the result is worse when compared to 4×5 minutes gain=100 stack. But still better, than 4×5 minutes stack at gain=0.
And here is screenshot made in MaxIm DL software:
|STACK||Background noise||Star #1 SNR||Star #2 SNR|
What conclusions can be made? Similar to the ones already made in the first part. Additionally you can easily assume that for CMOS narrowband imaging almost always it is worth to increase gain. There is no simple rule for single subframe exposure time and gain setting. There are too many factors that affects final outcome: sky brightness, telescope focal ratio, AD converter resolution, camera read noise, total exposure time, target dynamic range and probably some more. But for my current setup (QHY163M, f/7.5 telescope, suburban sky) I will probably work at the following settings:
- with LRGB filters I will use gain=100 and subframe exposure time 60 seconds (120 seconds for faint targets)
- with narrowband filters gain=100 and subframe exposure time 300s
But one way or another the differences between different settings final stacks are not massive. Unless you will mess up with settings really badly.