Problem with BatchPreprocessing

[adriano.valvasori]

I'm having problems with BatchPreprocessing as at the end of routin I get this error:

Error: RANSAC: Unable to find a valid set of star pair matches
Applying error policy: Continue on error

===== StarAlignment: 1 succeeded, 17 failed, 0 aborted, 0 skipped

I use a Canon 1100D.
Using the individual processes ImageRegistration (StarAlignment) and ImageIntegration, everything works fine.

Thanks.

Ciao

[pfile]

check if "use CFA images" is checked in BPP.

also, you may need to change the star detection thresholds in the BPP interface.

rob

[adriano.valvasori]

Rob, Thanks a lot, I'll do the tests.

Adriano

[jcinpv]

I get the same error message. This was supposedly fixed in BPP v1.31. I have v1.34.
Why wouldn't "Use CFA" be checked when I'm using RAW images from a Canon 60D camera? Isn't it supposed to be checked?

Here's my error message (tail end of the BPP process):

Registering target image 102 of 102
Loading target file:
/Volumes/1.5TB Media/Astronomy/Astrophotos/60D/2014/04/23/Registration/calibrated/light/debayered/MARKARIANS CHAIN_LIGHT_30s_1600iso_+35c_00236stdev_20140423-20h44m27s027ms_c_d.fit
Reading FITS: 32-bit floating point, 3 channel(s), 5202x3465 pixels: done
Noise reduction: done
Structure map: done
Detecting stars: done
137 stars found.
Matching stars: done
55 putative star pair matches.
Performing RANSAC: done
*** Error: RANSAC: Unable to find a valid set of star pair matches. (NOTE: This happened on *all* previous images.)

* Applying error policy: Continue on error.

===== StarAlignment: 96 succeeded, 6 failed, 0 aborted, 0 skipped =====
1446 s

*********************************************************************
* End registration of light frames
*********************************************************************

And at the end of all this, no image was left on the screen. Why not?
And what's this about changing the star detection threshold? Isn't 55 stars enough?

There's nothing about that to tell me, a nubie to PI, what the problem is. Maybe the programmer of this script will have a clue.

So, if you're the programmer, share the clue.

I've attached one of the raw images. You tell me if it isn't good enough to be processed (this is a screen shot of the original RAW image, so lots of detail is lost).

Thanks.

John C.

[pfile]

i was not clear - it should be checked. BPP does not try to divine whether or not the images are CFA, so you have to tell it. of course a DSLR image (or any other OSC) should be treated as a CFA image.

rob

[pfile]

in your log there star alignment claims to have succeeded on 96 images...

do you have images named …_c_d_r.fit somewhere in the directory that you specified for the BPP work area? those are the files that star alignment writes out and they are the files that will be integrated in the next step.

i can't remember if BPP opens the integrated result and displays it. if not it's somewhere in that directory with a name like MASTER_light-binning1.fit or similar.

rob

[jcinpv]

Thanks for the reply, Rob.

I directed all output to a subfolder of the Lights, Flats, and Darks.

I did notice that there are additional subfolders in that subfolder named calibrated, master, and registered.
Calibrated has additional subfolders named flat and light. Flat has 52 fits files, and light has 102 fits files and a subfolder named debayered which has 102 fits files.
Master has the dark and flat masters.
Registered has 96 fits files.

Opening just one of the registered files in Nebulosity and stretching it, the image looks good but it contains all the hot pixels. Why bother including my Darks it the hot pixels aren't being extracted from the Lights? Or is this done in a later step?

I supposed that with what I have, I can proceed with the workflow. But it bothers me that the errors show up and that the hot pixels are still in my registered images.

John C.

[pfile]

under normal circumstances, the hot pixels should be removed during calibration. however, if you look around here you'll see a lot of threads about DSLR darks and what canon does to them behind the scenes. this can sometimes interfere with the normal order of things (with respect to cooled, dedicated astro CCDs)

probably the best thing you can do to avoid problems is cool your DSLR to a fixed setpoint; if that's low enough you might get away without any darks at all. if you can not do that, then probably the next best thing to do (if you really don't like these hot pixels) is not to scale the darks ("optimize" them in PI parlance).when PI scales a dark, it scales it until the noise in the final frame is minimized. however, this can and does leave uncorrected hot pixels around. of course this implies that your dark subexposure length and temperature matches that of your lights. without active cooling the temperature matching is pretty hard to accomplish with a DSLR.

or, if you have dithered during aquisition, you can simply choose one of the sigma clipping methods to reject outliers and get rid of the hot pixels that way.

or, another option is to use CosmeticCorrection - here you load a master dark as a guide and then mess around with the sliders while looking at a preview to see when you have removed an acceptable number of hot pixels. if you save the process icon for the CosmeticCorrection parameters you like, you can point BPP at that icon and it will then do CosmeticCorrection after calibration but before registration.

rob

[jcinpv]

Thanks again, Rob. That all makes perfect sense, really.
I do intend to get set up to do dithering, but I'm going to have to find a decent autoguide camera that works with PHD in Windows 7. My SBIG SG-4 doesn't seem to want to be recognized by PHD.

As for cooling the 60D, I haven't looked around yet, but that will put a lot of extra weight on my SV80ED and the entire system, and I'm near max payload now.

John C.

[pfile]

it's a pain. i tried to build a cooler box but could just not get the temperature down enough. so i took another piece of copper plating and made an L bracket that fit under and across the back of my 50D. i found some 3m insulating material and shimmed between the live view screen and the plate. the bracket attached using the 1/4"-20 tripod mount on the bottom of the camera. against the back of the plate, i installed a peltier and a heatsink. i could get my exif temperatures down to around ambient this way, which ended up being about 10C below what it normally would have been. this was a totally lousy solution but those 10C really did make a difference in the dark signal…

rob

[jcinpv]

That's pretty close to the solution offered by http://www.centralds.net/cam/?p=2728

John C.

[pfile]

well the difference being that they open up the camera and install the cold finger basically right behind the sensor. here i was going thru several layers of plastic and air. i think they can get the sensors down to -20C or something like that.

rob

[jcinpv]

And all that for only $1,200! Let's see. First, do I trust them? Second, what's a comparable camera cost new?

Oh, and by the way, I managed to get through a BPP without any errors. I turned off "Optimize dark frames."

The integrated and auto-stretched image is attached.

John C.

[pfile]

good news. looks like it could use a little bit of DBE followed by CanonBandingReduction…

rob