Dark frames and DSLR's

[sharpie78]

This is my first post as a relatively new user to PixInsight and I am hoping to get some guidance regarding dark frames please.

I am using a Canon eos450D (rebel xsi) on a 130mm triplet. The camera has had a filter removal mod (not replaced although apparently I should have had it replaced as I'm using a refractor) and it is cooled with my diy cooler box (max cooling currently -16C). I can replicate and regulate the temperature to sub zero conditions reasonably well using a diy PWM controller (as long it's no warmer than say 10C air temp.).

I dither my lights by the maximum distance allowed in APT (5 pixels) with the same setting in phd.

I have made my way through the "Inside PixInsight" book and have a basic grasp of how PI calibrates and processes data (difficult considering mathematics and algorithms etc are way over my head). What I would like to figure out (hopefully using some of you guys' existing knowledge and experience) is the easiest/most effective way of dealing with dark frames.

As far as I can tell from reading other posts......I need to Dither to help with outlying pixels and colour mottle (ref. Tony Hallas) and use dark frames to deal with thermal noise and amp glow. Read noise is dealt with by the bias frames which is subtracted from dark frames to avoid adding it back in during calibration later. (please correct me if I'm wrong)

What I don't quite get yet is dark frame optimization....I get how it works but I'm unsure as to what kind of frames to collect or whether I should use a temperature matched dark library.

Should I take some...lets say...30 minute darks at the ISO levels I use ? How many should I take ? What temperature should I take them at ? Warm/Ambient or cooled or several temperatures ? Are DSLR CMOS chips Linear enough for Dark Frame Optimization to work ? Do I need to use an offset pedestal for biases (which I still don't fully understand anyway) ? Do I have to take seperate darks for my Ha filter and CLS filter ? (my gut says no on that one)

Now you see my problem.....too many questions. I sort of feel like I'm looking too far into it but I've used my time and effort to build a cooler to help deal with noise so I might as well understand this whole calibration process properly.

I already have quite an extensive dark frame library but the thought of having to recreate it in 6 months (ish) chills me to the bone.

If DFO combined with Dithering works well for other people I will be a happy camper and if anyone has anything to add/clarify please......have at it.........

[bulrichl]

Hi sharpie78,

I am using a Canon eos450D (rebel xsi) on a 130mm triplet. The camera has had a filter removal mod (not replaced although apparently I should have had it replaced as I'm using a refractor)...

Yes, you should use an UV/IR cut filter in order to limit longitudinal (= axial) chromatic aberration. This will facilitate focussing and will yield sharper stars.

... and it is cooled with my diy cooler box (max cooling currently -16C). I can replicate and regulate the temperature to sub zero conditions reasonably well using a diy PWM controller (as long it's no warmer than say 10C air temp.).

A cooled DSLR camera is a very special equipment that not many people will use, and I have no personal experience with it.

I dither my lights by the maximum distance allowed in APT (5 pixels) with the same setting in phd.

I have made my way through the "Inside PixInsight" book and have a basic grasp of how PI calibrates and processes data (difficult considering mathematics and algorithms etc are way over my head). What I would like to figure out (hopefully using some of you guys' existing knowledge and experience) is the easiest/most effective way of dealing with dark frames.

As far as I can tell from reading other posts......I need to Dither to help with outlying pixels and colour mottle (ref. Tony Hallas) and use dark frames to deal with thermal noise and amp glow.

Dithering is good practice.

Read noise is dealt with by the bias frames which is subtracted from dark frames to avoid adding it back in during calibration later. (please correct me if I'm wrong)

Read noise cannot be removed by subtracting. By subtracting a MasterBias from a light frame you remove
1) the bias (a constant offset that is added in order to avoid negative values) and
2) the small variations of the constant offset between pixels. This variation is NOT read noise, it is a fixed pattern (an unwanted signal).
Read noise is random noise that can only be diminished by a greater number of frames.

What I don't quite get yet is dark frame optimization....I get how it works but I'm unsure as to what kind of frames to collect or whether I should use a temperature matched dark library.

Take a look at my guide: https://pixinsight.com/forum/index.php?topic=11968.0

Should I take some...lets say...30 minute darks at the ISO levels I use ? How many should I take ? What temperature should I take them at ? Warm/Ambient or cooled or several temperatures ? Are DSLR CMOS chips Linear enough for Dark Frame Optimization to work ? Do I need to use an offset pedestal for biases (which I still don't fully understand anyway) ? Do I have to take seperate darks for my Ha filter and CLS filter ? (my gut says no on that one)

Generally, dark frame optimization is expected to result in a better calibration result if you
1) cannot control the sensor temperature in a narrow range (i.e. low quality of sensor temperature control) and/or
2) cannot avoid sensor temperature differences between darks and lights (i.e. low reproducibility of sensor temperatures).
However, if your camera exhibits "amp glow" (I guess the 450D does), it might be that with dark frame optimization enabled this artifact is not removed completely. As I also don't know how well your diy cooler performs, I cannot give good advice here - you will have to give it try and compare which approach makes a better calibration result (i. e. a better SNR in the integrated image).

Bernd

[sharpie78]

Hi Bernd

Thanks for your advice, I'll have a read through your guide in a little while once I start this evenings Light frames plan off.

Thanks for clearing up the bias terminology, I knew it was removing something and had got my random and fixed pattern noise confused.

Generally, dark frame optimization is expected to result in a better calibration result if you
1) cannot control the sensor temperature in a narrow range (i.e. low quality of sensor temperature control) and/or
2) cannot avoid sensor temperature differences between darks and lights (i.e. low reproducibility of sensor temperatures).
However, if your camera exhibits "amp glow" (I guess the 450D does), it might be that with dark frame optimization enabled this artifact is not removed completely. As I also don't know how well your diy cooler performs, I cannot give good advice here - you will have to give it try and compare which approach makes a better calibration result (i. e. a better SNR in the integrated image).

I think I will do as you suggest and just give it a try. Time to go down the rabbit hole and start analysing my cameras sensor characteristics as well I guess. At least I'll have a better understanding of how calibration frames work best in my setup and better calibrated images will result....hopefully

Quote

    I am using a Canon eos450D (rebel xsi) on a 130mm triplet. The camera has had a filter removal mod (not replaced although apparently I should have had it replaced as I'm using a refractor)...

Yes, you should use an UV/IR cut filter in order to limit longitudinal (= axial) chromatic aberration. This will facilitate focussing and will yield sharper stars.

I was refering to replacing the filter in the camera with a better quality Baader UV/IR cut filter, apparently not replacing it during the modification creates a difference in spacing of something that changes the focal plane of the red wavelengths on the chip causing a bit of star bloat or I think more accuratley...fringeing. Typically I can't find the post I read it in to get clarification.
I have got clip-in Ha and CLS filters for the camera to deal with UV/IR bloat and they work well but I do see a little bit of what I think is this minor issue. Annoyingly I've also read somewhere in a post that clip-in filters can also cause a small amount of fringeing but can I find that one to clarify further ?? can I heck. 

To be fair most of my issues are minor but I like to get into the nitty gritty of it all and try to understand why I'm following the instructions in the many useful tutorials and guides.

Thank-you for the information and clarification, hopefully someone with a cooled DSLR will have a look at this post at some point too and provide their perspective or best practices.
 

[ChoJin]

could you elaborate on your DIY cooling system for the dslr? (even going into the details of electronic design if you designed it yourself)

[sharpie78]

Hi ChoJin. It's a variation on the design by Gary Honis (google "dslr cooler box" and you're sure to find a link).

I have gone with a dual peltier design. The Peltiers I used are TEC12710's. They're mounted on either side of the cooler with water blocks doing the cooling instead of a heatsink and fan.

The water blocks are connected to a standard PC cooling radiator mounted on my pier with 2x 120mm fans cooling the water flowing through the radiator.

I control the power to the peltiers with a DROK PWM controller and can control the flow of water through the system with a simple pc fan controller which changes the speed of the water pump.

I also have a 50mm fan inside the cooler box to circulate the cooled air around the DSLR.

I wanted to monitor the inside temperature and humidity so I have a TemperHum USB temperture monitor mounted in there also.

I'm able to cool to around 20C below ambient temperature and can regulate the temperature quite well.

I have found that my DSLR doesn't like going much below -13C (internal box temp) which gives me EXIF data of -11C (chip temperature).

Is there anything else you'd like to know ?

[ChoJin]

Thx!
this gives me a pretty good picture of the setup.