PixInsight ImageCalibration oddness

[joecoyle]

Hiya

So I've just returned from a week in France with some marvellous M63 data in tow. Taken with a TEC140 and Atik 460 on a Mesu. All in all very good images. Courtesy of the fantastic Olly Penrice.

I thought to myself I'll have a go at processing the Sunflower only to find PI behaving very oddly indeed.

I don't know if I'm totally missing something or if I've messed up PI... FWIW I'm following Kayron's tutorial at LightVortexAstronomy for stacking etc... 

I've got Olly's master calibration frames for the Atik, and when i plug them into ImageCalibration, this is what I get (autostretched light sub)...

Whereas, if I run the BatchPreProcessing script set to calibrate only, my images calibrate as expected...

I cant understand it - I've used IC and followed the exact same tutorial a hundred times.

Any ideas?

Thanks

Joe

 

 

 

[wimvb]

Your image doesn't seem stretched at all. Note that STF doesn't stretch the image, it only shows on screen what a stretch would look like.

Did you apply the STF as a permanent stretch? Drag the small triangle of the STF dialog box to the histogram dialog box. Then apply the histogram stretch to the image. Then save image as jpeg/png to post here.

If the data is generally good (and I expect it to be with the used setup and conditions), you can always let the BPP script go all the way. Look at the rejection maps that are created.

 

[joecoyle]

This is what i don't understand - its had a permanent stretch applied to it... It's very odd... for reference here is an uncalibrated light frame with a permanent stretch. So you can see the data is good.

 

 

This is the same frame calibrated with the BPP script:

Maybe I'll just calibrate and stack with the BPP script - although i prefer the manual method. 

Joe

[wimvb]

Without seeing and trying actual camera files, I can do little more than guess. Camera files from a ccd have 16 bit depth. PixInsight settings can be at 32 bit (or floating point). A 16 bit image that occupies the lower half of 32 bit, seems black untill you stretch it. But calibration itself should never remove data from an image file.

If I were you, I would use bpp script for calibration, and do the registration and integration with StarAlignment and ImageIntegration, where you have more control over rejection schemes and settings.

[andrewluck]

When I've had oddities like this I've found that leaving the bias frame out of IC can help.

Andrew