Creating Flats - Help! (any way, but with a light panel in my case)

[DKNicholson]

For some years now I have been very lazy when it comes to flats and have rarely done them. Recently I created a library of flats, but when added the files to the lights in DSS I lost not only the dust bunnies and other artefacts but also some of the details of the DSO and the 'sky' became very 'flat'. Plainly I was doing something wrong.

I had made a Flat Panel with variable output and put some opaque white sheets in front to ensure the light was even. The camera at the time was an Atik 314L+ with full-well of 19,500. I've now changed the camera and wish to make the best of it. Now I have a Starlight Xpress Trius SX694 Mono and the Full-well is stated as greater than 17,000 and system gain of 0.27 electrons per ADU.

How do I establish the correct exposure time for flats? I shall be needing LRGB and Ha, SII & OIII 1x1 binning and RGB 2x2 binning. I use SGPro and it has a Flats Wizard but it needs to know the exposure range to use - and thereby hangs the problem. How do I establish that range? For those who use SGPro, is it necessary to create flats during the imaging sequence, if so how? Besides manually placing the light panel in front of the telescope after each filter, I cannot see how to do it - particularly as I work remotely and a flip-flat is out of the question (£££)!

Any 'simple' help with this knotty problem would be greatly appreciated - not only be myself but by many others I suspect! 😁

Edited November 13, 2019 by DKNicholson

[discardedastro]

Not sure about SGP but Ekos/KStars lets you provide a target average ADU. The key thing is to avoid nonlinearity in pixels, but otherwise as high as you can manage. Depending on the bit depth of the camera that will vary and may be tricky to calculate. In general I've seen recommendations of around 30-40% of max ADU being a good ballpark; personally on my ASI183MM setup I target 10k ADU and that seems to work well.

As to the rest - if you can maintain your setup's state (including filter positions, etc) between sessions flats are reusable. I've gotten away once or twice reusing flats between nights even when taking the camera out to do collimation just by making sure I got the camera back in the same position by eye, but it won't be perfect. If you're running in an obsy then placing the flat panel on the wall somewhere and having the scope slew to point at it can achieve a pretty good result (just make sure the panel is "flat" in attitude with respect to the scope, as far as is possible). You need some way to turn the panel on/off, of course.

It's been a while since I played with DSS but are you making master flats and calibrating them with dark/bias frames? I think DSS can do this automagically, but worth checking. Using uncalibrated flats could lead to problems.

[DKNicholson]

22 hours ago, discardedastro said:

Not sure about SGP but Ekos/KStars lets you provide a target average ADU. The key thing is to avoid nonlinearity in pixels, but otherwise as high as you can manage. Depending on the bit depth of the camera that will vary and may be tricky to calculate. In general I've seen recommendations of around 30-40% of max ADU being a good ballpark; personally on my ASI183MM setup I target 10k ADU and that seems to work well.

As to the rest - if you can maintain your setup's state (including filter positions, etc) between sessions flats are reusable. I've gotten away once or twice reusing flats between nights even when taking the camera out to do collimation just by making sure I got the camera back in the same position by eye, but it won't be perfect. If you're running in an obsy then placing the flat panel on the wall somewhere and having the scope slew to point at it can achieve a pretty good result (just make sure the panel is "flat" in attitude with respect to the scope, as far as is possible). You need some way to turn the panel on/off, of course.

It's been a while since I played with DSS but are you making master flats and calibrating them with dark/bias frames? I think DSS can do this automagically, but worth checking. Using uncalibrated flats could lead to problems.

Thank you kindly for trying to explain. Sadly I feel none the wiser as, for some reason, I am not comprehending the terminology or how to measure/establish 'target average ADU' and 'non-linearity in pixels'. I concur with your comment 'may be tricky to calculate' as I have absolutely no idea how to do that.

I like your idea of positioning the flat panel in the obsy - I hadn't considered that and it would not be difficult to sequence that into SGPro so that all the flats were done at the end of the sequence.

The camera I have is renowned for not requiring dark frames - it's a high QE CCD. As for Bias frames I believe a library can be created which is periodically updated, so that is something I should do.

[discardedastro]

2 minutes ago, DKNicholson said:

Thank you kindly for trying to explain. Sadly I feel none the wiser as, for some reason, I am not comprehending the terminology or how to measure/establish 'target average ADU' and 'non-linearity in pixels'. I concur with your comment 'may be tricky to calculate' as I have absolutely no idea how to do that.

Sorry - I'll try harder, I'm not great at explaining things and this isn't my absolute forte!

Target average ADU as in the average value of a pixel within an image - so if every pixel were set to 10,000 you'd have an average 10,000 ADU image. If your maximum full well is 17k ADU then beyond that value your pixels will potentially not behave linearly, so you need to stay well clear of that.

Atop of this there's scaling to consider - your camera's analog-to-digital converter will have a bit depth for each pixel's reading - 8 bit, 10 bit, 12 bit, 14, bit, or 16 bit. The SX-694 is 16-bit so there's no scaling involved. If you had a 12-bit ADC then your 17k full well might actually be a fair bit below 17k in the readout in the software.

Basically, if you aim for the average intensity of your image to be around 10,000 you'd be fine. If SGP can draw a histogram you should find a fairly narrow peak forming (with even illumination) somewhere between 0 and 65,355 (maximum 16-bit readout). Adjust your exposure time till it's sitting around 10k.

Dark frames will always help, even with low-noise cameras, and their inclusion is easy - plus they're portable between shoots so you can use some cloudy nights to record a large number of frames, make a master, and that's your dark frame for the next 6 months at least! Same goes for bias frames. PixInsight has a "superbias" feature which will take a master bias frame formed from a goodly number of bias shots (50 or so) and effectively extrapolate the overall bias of the sensor - might be worth an experiment if you have PI or can post a master bias for someone who does to process.

[DKNicholson]

4 minutes ago, discardedastro said:

Sorry - I'll try harder, I'm not great at explaining things and this isn't my absolute forte!

Many thanks indeed for clarifying the process and associating it with the camera that I have - very helpful indeed.  SGPro does have a histogram function so I can certainly use that. As our weather is currently cr*p then I can do something about darks and bias frames, something else I hadn't thought of! As it happens I don't have PI but DSS will create master files so I should be able to use that functionality.

Thanks again, much appreciated.