PI Flat Calibration, Star Registration issues.

[ftherrmann]

I used PI flat/bias/dark calibration with data that I've processed before using CCDStack.  When I did the flat/bias/dark calibration PI removed all data with the exception of the brightest stars, which were also very dimished in brightness.  NO amount of stretching using either Histogram or Transform would bring the data back.  Seems like a simple process.  What is wrong?

I then used the Star registration and viewed the registered files (all ending in the postfix "r" in CCDStack.  When I blinked the PI registered "r" images in CCDStack they were NOT registered.   Image integration seemed to work, however the raw images were fairly well registered so I'm not sure if PI integrated actual registered images or not.

When using the ADCNR wavelet function to preview noise the preview wavelet windows are banded with alternating "blurred" and non-blurred horizontal stripes.  Repeated previews on different images and different areas all resulted in this same type of banding.  The previews are worthless with this banding. 

I'm using the 1.8 version and have been told by more experienced users that 1.8 has some issues and I may want to try to go back to 1.7. 

Please advise.

Thanks,

Fred

[sreilly]

I also had issues with 1.8 and reverted back to 1.7. Seemed that registration wasn't one of the issues though. Maybe the process sequence has some affect. I realize that using 1.8 and reporting issues is the only way to work out the bugs so I've decided to install 1.7 on one computer and 1.8 on another. At least this way I have both to try and can maybe report some issues as they happen.

-Steve

[ftherrmann]

RC2 fixed the blurred wavelet noise banding problem that happens when you executed the "Extract Wavelet Layers" script.  I haven't tried any flat calibrations with RC2 yet.

[pfile]

OSC data or mono? when you say you've processed it before with CCDstack, do you mean that you've successfully calibrated the files in the past with CCDstack?

where did the master dark/bias/flat come from? made with PI?

[ftherrmann]

Yes all calibration files were made with CCDStack.  Sounds like your telling me not all flats/bias/dark files are created equal?

Are the _r  registered files that are written supposed to be registered?  ie. load them and blink them in CCDStack and they are registered or is there something propietary/exclusive to the way PI does it's registration?   Seems logical to write out a registered file as opposed to writing out the same file with added registration info built into a file header.  (I don't know how PI does registration so I'm only guessing)

[pfile]

yeah, although it seems like creating a master frame in any program should yield the same results as in another, in practice trying to use master calibration frames from one program in another program rarely works right.

i think you should start from scratch with PI, using all of your flat,bias,dark subs and using PI to make the masters, if you want to do the calibration in PI.

the _r registered frames are actually rotated and translated (and perhaps scaled). if you open one you'll see some black areas that are due to the rotations and translations. but if the calibration has destroyed the data, then what's in the _r files is just junk.

[ftherrmann]

Calibration isn't really a biggie since I like the scripting in CCDStack.  Also not really in the mood to generate new dark files.  I can calibrate, register, normalize and stack subs in less than 5 minutes using the scripting in CCDStack.  About 3 clicks of the mouse and a sub is ready for PS and/or PI.

[pfile]

to be clear, if you want to integrate the CCDStack-calibrated/registered subs in PI, you should be able to do that with no problem. it's the use of master frames generated in one program in another program's calibration that causes problems.

[ftherrmann]

I understand.  I've been using the calibrated subs produced by CCDStack in PI.  I did have issues regarding PS CS5 Tiffs, but this has apparently been cleared up with the RC2 release.

[Juan Conejero]

Hi Rob,

to be clear, if you want to integrate the CCDStack-calibrated/registered subs in PI, you should be able to do that with no problem.

Unfortunately, this is not the case, since the calibrated data are (must be!) in floating point format. Since the FITS standard does not provide any way to know the numeric range to which floating point image data are referred, and we don't know the numeric range used by the application in question (because, as far as I know, it has not been published), then there is no way to import the calibrated data consistently in PixInsight.

In general, floating point images are incompatible among different applications. In PixInsight you can define a custom floating point input range with the FITS and TIFF Preferences dialogs (or, much better, using input format hints on the ImageIntegration tool). However, in order to do that you need to know the bounds of the range used by the alien data. The only way to achieve correct results is carrying out the whole process in PixInsight: image calibration, registration and integration.

[pfile]

[post deleted - i must be tripping, hard]

juan is right of course, i had forgotten about the fits floating point issues.