Mono Newbie Question

[Spaced Out]

Hi all

I’m trying to get to grips with my new ZWO ASI 1600 its very different from my previous DSLRs and I want to try and get things right if I can !

So….. here’s a newbie question. Do people take different length exposures for different RGB filters ? I’ve been looking at the median ADU values for my RGB filters tonight and they are all slightly different.

Red = 1216

Green = 1360

Blue = 1328

Is this important, do I need to adjust exposure times for each filter to try and get closer ADU values ?

Thanks

[geordie85]

I've always taken the same length exposures for my rgb 

[rickwayne]

I doubt you need to worry too much about it. Close enough will be good enough in this case. More of an issue once you dive into narrowband, there's just a lot more light from hydrogen out there, for example.

It also simplifies your workflow and processing if the exposures are the same. One set of darks suffices. You might even get away with one set of flats if you have REALLY CLEAN filt--no, forget I said that. Do it right and take All Teh Flatzes, is my advice.

Edited March 20, 2020 by rickwayne

[Craney]

Another factor you may wish to consider is an additional Luminance exposure (L)  to eventually create an LRGB image.

I have read that the wisdom is then to make the L longer in total duration and then use the 2x2 'bin' feature on the RGB captures.  

There is a lot written on this  on the interweb (  )  but I gather it is essentially that the  image detail comes from  the L  and that should have precedence in the exposure run in terms of total time, whereas the colour  (RGB)  contribution to the final image  is not as critical in terms of data.   The 'bin-ing'   increases the sensitivity of the chip matrix, ( 4 pixel well depths instead of one)   to bring out  individual colour intensities but this obviously will result  in less resolution.... but hey! you have  bagged that already with the L channel.

..... I think this is  <sort of>  correct, but will be happy to be shot down in flames if not.

[wimvb]

If you have the zwo rgb filters for the as1600, you should be able to use the same exposure time. ZWO have developed these to allow one single rgb exposure time.

https://astronomy-imaging-camera.com/product/zwo-new-rgbl-filters-optimised-asi1600

Personally, even without optimised filters, I wouldn't bother with different exposure times. You would also need different darks, because these have to match the lights.

With my ASI174MM I use 240 s for rgb and 120 s for L. That works for me. Also, no need to bin the rgb, there is no significant advantage to bin cmos rgb frames, as opposed to ccd.

[ollypenrice]

The very dedicated image a star of the same spectral class of the sun (G2V) and experiment with exposure times until they get equivalent signal in each channel. They then keep the same proportion in exposure time for the main imaging run. Only a tiny minority do this and most, including myself, find that shooting at parity works OK because we have colour balancing routines available. Pixinsight's DBE or ABE work for me.  I agree with Wim: just use one exposure time.

8 hours ago, Craney said:

Another factor you may wish to consider is an additional Luminance exposure (L)  to eventually create an LRGB image.

I have read that the wisdom is then to make the L longer in total duration and then use the 2x2 'bin' feature on the RGB captures.  

There is a lot written on this  on the interweb (  )  but I gather it is essentially that the  image detail comes from  the L  and that should have precedence in the exposure run in terms of total time, whereas the colour  (RGB)  contribution to the final image  is not as critical in terms of data.   The 'bin-ing'   increases the sensitivity of the chip matrix, ( 4 pixel well depths instead of one)   to bring out  individual colour intensities but this obviously will result  in less resolution.... but hey! you have  bagged that already with the L channel.

..... I think this is  <sort of>  correct, but will be happy to be shot down in flames if not.

The question of binning colour is a bit more complicated than that. Firstly the signal gain for CMOS cameras is nothing like the gain it offers with CCD. It may not be worth bothering. And then we have to consider the image scale of the system in arcsecs per pixel. Anyone imaging with a 'middle of the road' scale of about 2"PP might consider binning colour. If your scale unbinned is coarse, though, say 3.5"PP, binning may prove just too coarse and might degrade the final image. At the other end of the scale anyone imaging at 0.5"PP is probably chasing the impossible and would do well to bin everything including luminance. Finally some more advanced processing techniques use RGB-only stars, without the luminance over them. For this you need top quality stars so probably unbinned.

Olly

[Laurin Dave]

I’d keep it “simple” to start within and use the same exposures for RG and B ..  I use 120 seconds for them and 60s for lum with my ASI1600 all at Unity gain and -10 deg C..  most importantly with this camera ( if you didn’t already know) is to use matched darks and flat darks to calibrate your lights and flats and not to use dark optimisation in the calibration routine ..  using the same exposures means you can build a darks library and keep using it for months..  best way to take darks is to take the camera off the scope put its cap on and point it face down onto a desk.  This avoids issues with any potential light leaks..  for my flats I have a dark library in steps of 0.1 secs from 0.1 to 1.5 secs and use whatever is closest.  
 

Dave

[wimvb]

CMOS cameras thrive on large number of subs, and most have sensors that have their peak sensitivity in the green part of the spectrum. If you find that the red and blue subs are a bit "thin", you are probably better off shooting more of these to bring the noise down. During post processing, the R and B channels can withstand the increased stretching that is needed to achieve good colours. Again, with the dedicated ZWO filters, this is probably not necessary. The median ADU values that you posted were already close enough.

The sky is seldom neutral; natural and man made light pollution will give a varying median level to your subs. Yesterday night at 1 am, the sky looked definitely blue to me, despite the moon not being around. Aurora, even if not visible, will affect your green subs. Man made light sources will affect all channels, depending on the type of light source.

[vlaiv]

Completely irrelevant in all aspects and most people use single exposure because it is the easiest approach - single master dark needed.

- color balance does not depend on exposure length - if you think it does - you can simply scale channels by multiplication to match what signal strength would be in equivalent exposure.

- SNR does depend on number of exposures vs exposure duration based on read noise and LP noise, and in principle you could find certain exposure length for each channel that is equivalent in some metric to other channels based on this, but that is not going to guarantee you same SNR per channel - in the end, most of SNR depends on target signal and that is different for each target. You can control how much SNR per channel you have by using different total imaging time (different number of subs of same duration) instead of sub duration.

- I've seen someone talk about color binning - it is almost the same thing with CMOS cameras and CCD cameras, or software and hardware binning. In fact if you are clever and do the math properly - you can make them be the same, but it's not really needed as difference will be small to start with. It is worth binning color and then resampling that to higher resolution of Luminance if you suffer color noise. You don't have to do it - you can do much stronger noise reduction on color channels prior to color transfer - effect will be pretty much the same. This is because we tend to perceive detail in luminance much more than in color.

[Spaced Out]

Thanks everyone, this is all good information for a mono newbie like me !

I saw that most folk use the same sub lengths for their RGB so I sort of assumed that was normal practice. I was trying to understand the processes involved and work out my own sweet spot for camera settings/exposure times, I felt I had to ask this question when I found differing ADU values (I was expecting them to be closer), I wasn’t sure how important the difference in values were.

Anyway, I’m just feeling my way around and trying to get my gear working. I am at the point now where things are not bad (after a headache with a wobbly focuser tube), I think I have some reasonable settings, I can start imaging.

I took my first ever RGB frames on Thursday night and I combined them today, this is the result. It was purely a test to see if I could learn the process, not really a target for me as such it was just available at the time.

This is 10 frames each of RGB with no calibration frames. OK, it’s not going to win any awards but I am just chuffed I’ve got a colour image out of a mono camera.      

Time to do some well planned imaging session now !

Thanks all for your help.

 

Edited March 21, 2020 by Spaced Out

[ollypenrice]

Very good. You have a lot to work with there. This is not a particularly colourful part of the sky but you have the blues showing in the star-forming regions of the 'Eyes' pair galaxies. Also some blues in the edge on spirals.

Remember that all digital cameras are mono. An 'OSC' camera is a mono camera with filters and that's exactly what you're using here.

Olly

[tooth_dr]

Just to jump in here - everyone is saying they are using 1/2 the exposure duration for luminance compared to RGB?  Instinctively I assumed the opposite.  This may have been where I was going wrong. Unless binned 2x2 and then use the same time?  

[vlaiv]

9 hours ago, tooth_dr said:

Just to jump in here - everyone is saying they are using 1/2 the exposure duration for luminance compared to RGB?  Instinctively I assumed the opposite.  This may have been where I was going wrong. Unless binned 2x2 and then use the same time?  

That sort of makes sense. I'll explain briefly why.

Duration of exposure (if you worry about that sort of thing - getting last bit of SNR) is related mostly to sky glow related noise. Cooled cameras nowadays have rather low dark current so that is not biggie, we often image faint stuff so shot noise of target is not dominant (and if it is - signal is already strong so your SNR is fine) that leaves sky glow noise as important bit - you want that to be larger than read noise. At that point you enter region of diminishing returns - yes, longer subs will get you better result but difference becomes marginal real fast.

RGB filters divide 400-700nm range in roughly 3 distinct equal "zones", each about 100nm wide. While L covers whole range. Sky glow can be thought of as rather uniform over that range. From that it stems that L filter will capture about x3 more signal in same amount of time as each of RGB filters. This means that it will capture Sky glow at same increased rate and consequently related noise will be larger by factor of about square root of x3 = x1.732.

To get same ratio of sky noise to read noise you should roughly expose L x1.732  = ~x2 shorter - or about half of RGB exposure.

[wimvb]

My motivation is much simpler than Vlaiv's. I don't want to overexpose the stars, and at the same time I want to get good signal. I tested several exposure times, and settled for 120 s Lum. This also works well with guiding results, and keeps the number of satellites/sub down. At the same time, rgb filters have only 100 nm passband, as Vlaiv noted, vs 300 nm for L. My camera is most sensitive for green, so doubling the rgb exposure time makes most sense to me. It also seems to me that if you use the same exposure time for all four filters, either your L subs are over exposed, or your RGB subs are under exposed.

Edited March 22, 2020 by wimvb

[tooth_dr]

1 hour ago, wimvb said:

My motivation is much simpler than Vlaiv's. I don't want to overexpose the stars, and at the same time I want to get good signal. I tested several exposure times, and settled for 120 s Lum. This also works well with guiding results, and keeps the number of satellites/sub down. At the same time, rgb filters have only 100 nm passband, as Vlaiv noted, vs 300 nm for L. My camera is most sensitive for green, so doubling the rgb exposure time makes most sense to me. It also seems to me that if you use the same exposure time for all four filters, either your L subs are over exposed, or your RGB subs are under exposed.

Thanks Wim. I now see my RGB are underexposed and Lum overexposed! 

[vlaiv]

Over exposure of luminance is not important - you end up with very strong stars after stretch anyway - as if they were clipped in recording.

Color on the other hand should not be over exposed, but luckily there is simple technique that one can use if they have over exposure (even if it is only star cores), shoot only a handful of short subs and replace over exposed pixels in final stack with suitable values from short stack.

[Freddie]

Over exposure with L of the actual target would be important to avoid, less so with just the surrounding stars.