Page 8 of 8

Re: Another Offset Question

Posted: Fri Dec 28, 2018 12:43 am
by SteveJP
turfpit wrote: Thu Dec 27, 2018 10:59 pm

To progress this, as requested previously above, please make available somewhere a bias frame which has the histogram banging into the LHS in FITS Liberator and the corresponding SharpCap histogram CSV file (posted ~Dec 23).

Ok thanks. I'll do that soon.

Re: Another Offset Question

Posted: Mon Dec 31, 2018 7:51 am
by SteveJP
Hi Dave and Robin,

I've been running a few more experiments which may or may not be of interest. I appreciate the time you both have given my questions and I've certainly satisfied my curiosity about why you need to increase offset with gain, at least assuming you always want the least offset possible. I must admit to not being totally convinced about the 14/16 bit thing, since it's a totally reversible transformation. Obviously it matters to DSS and I totally accept that.

To summarise, here's what I've learnt about SharpCap etc in regard to the "Save 10/12/14 bit images in FITS file without stretching to 16 bit ...."

1 - If your using the ZWO native driver (ASI071 MCPro), the histogram output by SharpCap is unaffected by the above setting. This means when SharpCap is working in it's default 16 bits FITS mode, the histogram does not represent the actual FITs file. (Although it would be easy to manipulate it with Exccel so it would)

2 - If you use the ASCOM driver rather than the native driver, the above setting does not appear to make any difference. The file is created as if the tick box is not set. The resulting histogram file has 64K lines (ie 2^16) and the histogram does represent the contents of the FITS file.

3 - The graph below shows the histogram output for the ASI071MC Pro, 0.1ms exposure, Gain=90 (unity), Offset = 20 with the above setting not ticked. The solid lines shows the actual offset (median value = 17) before the 14 to 16 bit conversion, but the dotted lines come from an ASCOM histogram with the same settings showing the histogram after the 14 to 16 bit conversion accurately representing the FITs output.
(869.72 KiB) Downloaded 101 times
Also, although not directly an issue with SharpCap, but perhaps of interest

Some time back ZWO removed the ability to set offset in their ASCOM driver for the ASI071Pro and related cameras. The offset was fixed at about 50 regardless of gain. I've attached the result of this with a gain of 240 (optimum read noise), and the offset is clearly insufficient. I note however that ZWO have put back the offset setting - I dare say there were complaints. This histogram is certainly banging into the LHS.

Files attached (in zip file) are as follows (I don't expect a forensic investigation here, they are attached as requested)

1 - 0.1ms_G90_Off20_14Bit_00000.CameraSettings.txt (Ie 0.1ms Exp, Gain=90, Offset=20 & NOT stretched to 16 bits - ZWO driver)
2 - 0.1ms_G90_Off20_14Bit_00000.Histogram.csv
3 - 0.1ms_G90_Off20_14Bit_00001.png (FITS liberator view)

4 - 0.1ms_G90_Off20_16Bit_00000.CameraSettings.txt (Ie 0.1ms Exp, Gain=90, Offset=20 & stretched to 16 bits - ZWO driver)
5 - 0.1ms_G90_Off20_16Bit_00000.Histogram.csv
6 - 0.1ms_G90_Off20_16Bit_00001.png (FITS liberator view)

7 - 0.1ms_G90_Off20_16Bit_ASCOM_00000.CameraSettings.txt (Ie 0.1ms Exp, Gain=90, Offset=20 - ASCOM driver)
8 - 0.1ms_G90_Off20_16Bit_ASCOM_00000.Histogram.csv
9 - 0.1ms_G90_Off20_16Bit_00001-ASCOM.PNG (FITS liberator view)

10 - 0.1ms_G240_16Bit_ASCOM_00000.CameraSettings.txt (0.1ms Exp, Gain=90, FIXED ZWO ASCOM Offset ~50 - older ASCOM driver)
11 - 0.1ms_G240_16Bit_ASCOM_00000.Histogram.csv
12 - 0.1ms_G240_00001-ASCOM.PNG (FITS Liberator view - Did this cause complaints due to fixed offset ??)

Many thanks

Re: Another Offset Question

Posted: Tue Jan 15, 2019 1:51 pm
by turfpit
I found this interesting article, which I believe is relevant to this thread, on the QHY website ... 334403.pdf

Written by this gentleman I believe

Not an easy read but worth the effort for anyone who wants to get the most out of their imaging sessions.

I love the Horsehead image at the end - look at all the interstellar dust in the bottom third. Just shows what can be achieved
making use of F2.2 Celestron RASA and 10min exposure, with lowest GAIN value

Re: Another Offset Question

Posted: Wed Jan 16, 2019 8:37 am
by SteveJP
Hi Dave,
That'a a really good explanation by someone with impeccable credentials, and your right, that Horsehead is stunning.

Re: Another Offset Question

Posted: Wed Jan 16, 2019 12:40 pm
by MrAstroBen
Reading this makes me love my CCD even more 😉

Re: Another Offset Question

Posted: Wed Jan 16, 2019 3:37 pm
by oopfan
CCD, simplicity defined. It amazes me how aggressively CMOS is marketed to amateurs yet it is so much more complicated to use properly. There are simply too many levers to pull to get yourself into trouble. Thank you AAVSO for your online course on CCD Photometry which helped me make sense of this mess.


Re: Another Offset Question

Posted: Wed Jan 16, 2019 6:57 pm
by admin

there is good information in that article but it misses out on a fundamental point – when you stack images you increase the dynamic range proportional to the square root of the number of frames that you stack. This effectively reduces the disadvantage in dynamic range seen for shorter exposure/higher gain imaging because you can take more subframes in the same amount of imaging time and therefore regain some (in some cases most) of the dynamic range loss compared to lower gain settings.

In fact, if a camera had zero read noise the optimal way of using it would be to take the shortest possible exposures and stack zillions of them (you might limit this by the number of frames you're prepared to try to stack or the amount of data you want to store) but you would get the very highest possible dynamic range because none of the stars would saturate in those very short frames.

Anyway, the SharpCap Smart histogram/brain calculations take all of this into account – which was the hot idea of them in the first place.

Cheers, Robin