Interlaced PHD image quality for SX-Lodestar compared with SX-ASCOM or Nebulosity-3

[Alan Hewat]

I use PHD and now PHD2 to image an SX-Lodestar camera (no guiding).

In both PHD and PHD2, images are degraded with the lodestar camera choice, due I think to interlacing corrections. Noisy pixels are fewer but bigger, making it difficult to remove them with applications like ImageJ. This does not happen for the SX-ASCOM camera choice with no interlacing correction, only double exposure. Almost identically good images are obtained with Nebulosity-3 with either the SX-ASCOM or native Lodestar camera choice.

Looking into PHD's cam_SXV.cpp  I find a comment that I think points to the problem

// Interlaced cams will be run in "high speed" mode (vertically binned) to avoid all sorts of crap

Frankly, I prefer to see the crap and decide myself what to do about it :-)  But I'm not sure how to remove the binning.

I enclose a pair of PHD2 dark field exposures from a (rather old) Lodestar (FITS converted to JPEG). I can filter the noise from the SX-ASCOM image, but not from the native Lodestar image because it is not single pixel noise.

What I would really like in PHD is the cam_SXV.cpp module that is used by Nebulosity, which works perfectly in Nebulosity for the native lodestar camera choice.

Alan.

[Bret McKee]

Alan:

I did some research a long time ago, and based on the Lodestar star shapes that Nebulosity displays, I suspect that it does even fancier noise reduction, and that is why the sxASCOM driver has the Gaussian blur option. 

Please try enabling that option and see if you can get the similar results to the Nebulosity results.

Bret 

[Alan Hewat]

 The raw PHD sxASCOM image looks just like the raw Nebulosity image with individual bright pixels that can be filtered easily with 3x3 median averaging. That's not a problem. The problem is that the Starlight Xpress camera choice in PHD bins pixels making noise filtering difficult. It doesn't help to blur further.

Alan

[kenm.h...@gmail.com]

Yeah...

I noticed that the V2.2 goes for 2 x2 binning on the Lodestar......

[Alan Hewat]

I noticed that the V2.2 goes for 2 x2 binning on the Lodestar......

Actually I think the "Starlight Xpress SXV" camera choice for the Loadstar only bins x2 vertically in either PHD1 or PHD2  i.e. it averages the even and odd interlaced lines.

In any case, binning before noise reduction is NOT a good idea. You can bin the Loadstar 2x2 (or more) in camera if you use the ASCOM chooser and click "Properties", but then the 3x3 median noise reduction is less effective. 

Binning is useful to increase sensitivity, so it would be good to have a PHD binning option (not in-camera) that could be applied after any noise reduction. (OK, it would make it slightly slower for very short acquisitions :-)

At present, clean Loadstar images are only obtained with 3x3 median noise reduction with either the "Starlight Xpress Loadstar Guider (ASCOM)" or the "ASCOM Camera Chooser" with no binning (or by subtracting a dark frame, which is not always appropriate for a camera with no temperature control, or eventually by mapping bad pixels).

Alan

[Andy Galasso]

On Monday, March 24, 2014 7:19:06 AM UTC-4, Alan Hewat wrote:

You can bin the Loadstar 2x2 (or more) in camera if you use the ASCOM chooser and click "Properties"

Actually, there's an easier way: in the Connect Equipment dialog select "Starlight Xpress Lodestar Guider (ASCOM)" for Camera, then click the Camera Setup button (just to the left of the connect button).  I find this preferable to using the ASCOM chooser since your camera selection is remembered across sessions and you do not have to keep choosing it when you connect.

Andy

[Alan Hewat]

On Monday, 24 March 2014 16:24:56 UTC+1, Andy Galasso wrote:

click the Camera Setup button (just to the left of the connect button)...

Thanks Andy. That is indeed more convenient. And thanks for pointing to the updated instructions for compiling PHD2. I'm downloading MSVS 2013 now. 

Alan

[kenm.h...@gmail.com]

Alan,

I'm pretty sure it's doing a 2 x 2 binning as the native 1 x 1 shows an "elongated" horizontal image....this this squared up in the PHD view.

[Christopher Madson]

Swapping to 2x2 binning on my lodestar dramatically improved the stars I was getting and gave me better guiding. I don't understand the math behind it, but the results are pretty spectacular.  I'm using a OAG and the binning was necessary for finding guide stars.