Combining Subs from multiple nights

[Notty]

Hello experts!  I have a couple of questions, not I am starting to lose the desire to just photograph different things each of the few nights I get to do AP, and want to start accumulating more and more data on objects to improve detail. 

1. When combining data, do you process each nights subs and calibrations then combine the finishing result from each i.e. from 5 nights you would ultimately just throw 5 finished tiffs at DSS or whatever?  I know it sounds obvious and I can't think of a different way of doing it but wanted to check!

 

2. Darks.  (I'm DSLR only) They are a pain in the neck!  I usually set them to run after my imaging is complete and I've fired off the bias and flats, but I usually come back in the morning to an error on APT saying it lost contact with the camera and I'm left with as far as it got.  (I have the camera on mains power by then and all my power saving settings on lap are disabled).  How many of you build up a library and if so, how close do the temps have to be?  I'm struggling to find the sensor temp in the metadata saved down to the lappy in Canon DPP.  I thought of shooting them light/dark alternately, but then surely the camera sensor temp is changing constantly, so the darks would too.  Must admit I have no idea how sensitive the Dark processing is to them being exactly the same temp as the Lights.  Would I be right in thinking then I can afford to go to cooled CCD I won't need to to them any more?  

 

3. Dithering.  As a mitigator to the above, I've read a lot about how it can mitigate against Darks (though not replace) and I'm trying to find instructions on how to enable it between APT and PHD2 (I'm confused about which package is in charge of the dithering) if anyone knows of any resources on in perhaps they could post a link?

 

This is what I got from 10 x 600s the other night, with a mere 6 darks, 50 or so bias and 20 of the Neumann flats with very rough processing,  but I'm greedy now for more detail hence more data!

  

[Jessun]

Hi Notty,

The experts will chime in here no doubt but here's a first answer.

1) I run all accumulated subs into the final stack. That gives each image the same statistical weight. Let's say one night only rendered one hour worth of useable subs and the other four nights much more. If you stack them separately night by night and then combine, that first night will end up responsible for one fifth of the end result when in pure hours it may represent be way less. You could in theory play with this effect allowing your best subs or best night have greater influence of the final stack but that is way to complicated for me.

Personally as I build night by night I simply throw more and more subs away from each session - ie I scrutinize star shapes and haze etc and just throw the garbage away. Then I do one single stack of all. Using a sigma reject routine most artefacts go away in one swift stroke. That could be all hot pixels or an entire bad column that a bias frame couldn't quite kill. This hinges on that my pointing is a little bit off night to night - which is always is....

2) Skip darks with a CCD? Yes quite possibly. I don't do them with my sony CCDs and many imagers running Kodaks don't either. Darks can be replaces by a hot pixel map in many cases if need be. A library can otherwise easily be kept with say 5-10 degree difference. You will see instantly when you're off either with temperature or as the sensor ages over time (months to years). Trial and error is good enough here.

3) Dithering is great. With the right stacking routine dithering on it's own effectively deals with CCD oddities and I think it also produces a better image as it almost does a drizzle job without software.

My approach lately has been to return to the same target over several nights so I don't dither. Each night is like a little dither. I don't know APT well and how it interacts with PHD. The trick is to pause PHD while the dithering is going on. I use Nebulosity with my cameras and no surprise that works well if I do select the function. DSLRs probably have most to gain from dithering.

/Jessun

[ollypenrice]

I agree with Jessun, though I think DSLR users should use a large dither in order to reduce the effects of colour mottle in the background sky. (Like Jessun I do multi night imaging and with our dual CCD rig it is impractical to dither.) Tony Hallas is a good DSLR imager and recommends a 12 pixel dither.

I don't use darks even with noisy Kodaks. I subtract a master bias as a dark, apply a defect map and set the hot pixel filter to an aggressive value. This works far better than darks for me, and since, whatever you do with your camera, you cannot know the real sensor temperature, your darks risk doing more harm than good. I would subtract a master bias as a dark and do a large dither. I'm afraid I can't help on how to set up PHD to do this but it is the capture programme which instructs PHD, not the reverse. Edit: this makes it much easier to stack the whole lot in one final go, as Jessun suggests. You don't need to worry about which darks go with which lights. Alternatively you can always calibrate your individual subs without combining them and then stack the lot at the end.

Olly

[ampleamp]

Caveat, I am no expert.

what Olly suggests in his last sentence is what I try to do and I use PI for all this. I always have dither enabled but only by usually 3 pixels.

I use the blink tool to note any duff subs to delete as well as to choose a suitably framed sub to act as a reference to which all the others are registered. I have become progressively more aggressive with what I delete on the basis of rubbish in, rubbish out.

Then I calibrate each night's session of subs with a master bias, flats and an aggressive cosmetic correction using the batch pre-processing script.

Image integration next, I add all the calibrated subs from all the sessions into the tool as they are all now calibrated and registered, and stack. 

Of course, all this is dependent on how much data you have, and patience .... ?

 

[wimvb]

Hmm, about dithering... I aim for 15 pixels dithering on my Pentax imaging camera, close to what Tony Hallas suggests. This works fine for me.

But when setting dithering in PHD, I wonder how that works. Does PHD know or care which imaging camera you use? So if you set dithering in PHD, doesn't that mean dithering on the guiding sensor? In that case you would still have to calculate how much dithering you get on the imaging sensor?

Just a question from an interested non-guider.

I also use hot pixel removal (cosmetic correction) in PixInsight rather than doing darks. I rather spend my cold nights collecting data than hot pixels.