Starlight Xpress 694C issues in post processing - HELP! :-)

[JayS_CT]

All,

I'm capturing some beautiful images with the 694C on my SkyWatcher Esprit 120 (wit an ES 3" FF/FR) yielding 588mm @ f/4.9. I'm running into a couple of issues that are starting to drive me a little crazy and seeking some advice an suggestions.  First, the Matrix on the 694C is GRBG. This is confirmed in SkyX and other packages reading the FITS header. I've been in touch with Starlight and they've been helpful is trying some things.

The first issue is the Master Bias does not seem to take care of some of the content in the lights, where as Maxim DL gives a clean image. Specifically there are verticals that disappear when calibrating with Maxim whereas in PI, they're either "light" or "dark" (but always present) depending on whether or not I have the calibrate box checked for Master Bias when calibrating with the debayered and aligned lights..  I just cannot make the vertical "lines" disappear (I can blacken the background enough to minimize but not eliminate).

Next up is color balance. I've tried a number of things to reduce the green content and bring out the reds, but again not meeting with much success. I recently did the western Veil and while I can bring out the Blues, the Reds are just not coming out.. best I can achieve is some shade of "brown"..  Starlight said to use a color balance of 130 for Blue, 110 for Red (presumably leaving Green at 100).  This is being done with Maxim DL though and I'm not finding a similar "scaling" within PI..

My goal would be to have a "modified" matrix pattern when I debayer to bring things much closer to reality. I've read some posts on CN, etc. to use SCNR to bring the Green levels down before going into an Automatic Background Extraction..  Helps, but still not there.  I've done Photometric Color Calibration and Color Calibration selecting a Preview section for the "dark" portion in the image.. and still just not getting there.  Hate to think Maxim is capable to reading the headers and doing this "broadly" easier when it is generally not a match for PI capabilities.  More than willing to share lights and Master Bias (no Darks per Starlight, and actually Darks do make noise pattern worse).. Since there is no Amp Glow, darks are as important it appears as the BIAS frames.

Appreciate any and all help!  Really not understanding with calibrating is not removing vertical lines that are clearly there in BIAS..  :-(

Jay S.

[Juan Conejero]

Hi Jay,

As always: we cannot help you unless you upload a data set where we can reproduce and understand these calibration problems. To help you with PCC, we need also a linear image where you are having problems.

[JayS_CT]

Hi Jay,

As always: we cannot help you unless you upload a data set where we can reproduce and understand these calibration problems. To help you with PCC, we need also a linear image where you are having problems.

Juan,

Thanks.  Specifically what do you want me to supply?  All the FITS lights and the Master Bias?  I assume if you have both of those, you can get to the issues I'm referring to?  Really appreciate the help..

Jay S.

[JayS_CT]

Juan,

Download cloud link for the lights, master bias and a master dark. I don't use the dark on recommendation from Starlight as it's really not needed it appears and does increase noise.. The master bias file does a good job except for the vertical lines after calibration and integration.  Color is somewhat driving me mad. :-)  Again Matrix is not RGGB but GRBG.

https://my.pcloud.com/publink/show?code=kZ7BDr7ZcbI9lrgnhGu7K15bNqsPk7d1fowV

Appreciate any help and suggestions..

Jay S.

[chris.bailey]

Is that the master Bias you have been using? Its a debayered RGB file so not suitable for calibration the non-delayered light.

All the calibration files need to be un-debayered.

[JayS_CT]

Is that the master Bias you have been using? Its a debayered RGB file so not suitable for calibration the non-delayered light.

All the calibration files need to be un-debayered.

Thanks.. Always thought Debayer was first step.. Helped on the verticals, but still struggling with color balance.  Actually was closer doing it the way I was doing it.  :-(

Jay S.

[Niall Saunders]

Hi Jay,

If your camera manufacturer suggests that you don't need, or need to use, any Darks, then there is no need to use Biases either.

Hoever, as all sub-frames (inlcuding Darks) already include a more-or-less statistically identical Bias element, then this will automatically be removed during the Calibration phase within PI.

Just remember that all sub-frames are basically the same format:-
(L + D) * (F + FD)
where
L is the Light data at temperature t_L snd exposure time e_L
D is the Dark data at temperature t_L snd exposure time e_L
and
F is the Flat data at temperature t_F snd exposure time e_F
FD is the FlatDark data at temperature t_F snd exposure time e_F
and also
Each sub-frame (L, D, F and FD) includes the bias data, B - statistically 'identical' no matter which frames are being calibrated. This means that, for example, when the D term is subtracted from the L data, the B data totally disappears, and the D data is 'calibrated out' of the L data.

When you then 'divide' Dark-calibrated L data by the Dark-calibrated F data, the Flat data is statistically removed from the L data (again, with no requirement for B data).

All of these Calibration stages must take place before any CFA deBayer stages which must then take place next, and before any star alignment/image registration process. Finally, the image set can be integrated - ready for post processing.

I hope I haven't made too many mistakes - hope it helps.

[dave_galera]

Silly question, but why would Starlight Xpress state that darks are not required, are they pre-processing the subs with the on-board controller (in-camera dark suppression)?

[Niall Saunders]

Hi Dave - yes, that was also where my questions were going !!

[JayS_CT]

All,

Again, thank you all.. I've been working it up and through background extraction. Have tried SCNR to remove some residual green. The manual has an error on Matrix.  It is GRBG (although I've tried several different ones ). I'm just have difficulty getting the blues and reds in this image.  I do really appreciate the help and on other objects I'm very comfortable and getting good results (color wise).  Just can't seem to nail this one. I truly love the camera and want to try to get the best "procedure" down before I start working on any other noise reduction, edge enhancements, etc. A lot of that works out well in Adobe Lightroom as well.  The PCloud download folder now has a non-debayered Master Bias and as I mentioned it does eliminate the vertical lines (which the manual points out in the BIAS section).  I should also mention I'm using a 2" STC Astro light pollution filter which I realize can be shifting things.

https://my.pcloud.com/publink/show?code=kZ7BDr7ZcbI9lrgnhGu7K15bNqsPk7d1fowV

Jay S.

From the SX-694C manual on Dark and BIAS.. https://www.sxccd.com/handbooks/Trius-H694C%20handbook.pdf

DARK
"Most competitive brands of CCD camera require a ‘dark frame’ to be subtracted from your images to achieve the best results. A dark frame is simply a picture which was taken with the same exposure as your ‘light frame’, but with the telescope objective covered, so that no light can enter. It records only the ‘hot pixels’ and thermal gradients of your CCD, so that these defects are largely removed when the dark frame is subtracted from the light frame. The TRIUS-H694C CCD is quite different from those used in other brands of camera and generates an extremely low level of dark noise. Indeed, it is so low that subtracting a dark frame can actually INCREASE the noise in your images! This is because the statistical noise of the dark frame can exceed the ‘pattern noise’ from warm pixels and hence add to that of the subtracted result. If your test pictures have an exposure time of less than about 10 minutes (as above), then don’t bother with a dark frame, just ‘kill’ any hot pixels with your processing software. In TRIUS-H694C, the ‘Median filter’ can do this, but other software (e.g. Maxim DL) will provide a ‘hot pixel killer’ that can be mapped to specific locations in the image, or methods such as ‘Sigma combine’ may be used.??.

BIAS
‘Bias frames’ are somewhat more useful than dark frames when using the TRIUSH694C. A bias frame is essentially a zero exposure dark frame and records any minor readout defects that the CCD may possess, so a ‘bias frame subtraction’ can clean up any ‘warm columns’ or shadings that are created during readout. To record a bias frame, cover the camera aperture and take a 1000 th of a second exposure. If you take at least 10 such frames and average them together, the resulting ‘master bias’ can be used to clean up readout defects for many months before CCD aging changes require another set to be recorded."

[dave_galera]

The sensor is a Sony ICX694ALG EXview

[JayS_CT]

The sensor is a Sony ICX694ALG EXview

Dave,

Yes.. There is a mono and OSC version of the camera.  Very little noise left after BIAS subtraction and Integration with Windsorized Sigma Clipping for pixel rejection.

Jay

[dave_galera]

And this is Atik's take on the same chip that they use in their 4 series of cameras:

Due  to  the  very  clean  nature  of  these  CCDs,  dark  frames  are  unnecessary  in  most  cases. This is due to the low noise which stands at an amazing 5 electrons RMS typical.

[JayS_CT]

And this is Atik's take on the same chip that they use in their 4 series of cameras:

Due  to  the  very  clean  nature  of  these  CCDs,  dark  frames  are  unnecessary  in  most  cases. This is due to the low noise which stands at an amazing 5 electrons RMS typical.

Dave,

Good find.. Didn't think to check ATIK..  What Matrix are they claiming?  Just curious..

Anyone wanting to take a whack at what I have in the cloud download, I'd greatly appreciate learning.  I realize stellar (clusters, etc.) are going to likely process a little differently..  The chip is supposed to be red sensitive, which is what is frustrating me in not being able to pull that out. :-(

Jay S.

[chris.bailey]

I have the Atik version of that camera which I tend to use at star parties for the simplicity it offers.

I don't use darks for calibration.
I find this chip needs a good soaking. My subs for galaxies are typically 10-15 minutes.Matrix for debayering is GRGB BUT you do need to be careful how you have the .fits setup in PI against how they are written by you capture software. I have PI setup as UP_BOTTOM
If you don't use Darks you will need to include Cosmetic Correction in the workflow to deal with warm pixels BEFORE debayering unless you are using a very large dither. Warm pixels on an OSC turn into little crosses when debayered. Interpolation during alignment will then 'smear' these into warm blobs that can be 5 or 6 pixel artefacts. A dither therefore needs to be at least 5 pixels to allow them to be rejected during stacking.
You will ALLWAYS need to colour calibrate. Light pollution and LP filters, atmospheric extinction etc will always throw the colour balance off kilter. PI has brilliant tools for CC.

My base workflow is Calibrate (Bias and Flats) > Cosmetic Correction > Debayer > Align > Stack > Crop > DBE > Background Neutralisation > Colour Calibration > Stretch