Image Integration bug under Windows RC7

[mcgillca]

Hi - I have been having odd flats recently, but have been making do with old versions until new dust motes caused me to seek a solution.

When I use the Batch PreProcessing script, I was getting flats with an overall illumination of 0.02.

This is odd, since each flat frame has an average illumination of about 0.32.

I traced the problem to an issue with the Image Integration module.

I can integrate 10 callibrated flat frames, each of which has an average illumination close to 0.3, but the integrated field has an illumination of 0.02.

I tried with several different normalisation techniques, but found all but "No Normalisation" gave a problem.

I tried replacing the master flat frames with these, but the batch processing script didn't recognise them.

I am happy to post my frames somewhere if this would help.

Colin

[pfile]

what type of input files are you using?

[mcgillca]

Fits files, created by CCDsoft, taken from an Atik 460ex.

Colin

[Carlos Milovic]

Are the images you are integrating bias-subtracted? If so, then you should use multiplicative normalization only (if used).

As long as the only transformation applied to the flat frames is a multiplication, it doesn't matter the average value, since it is normalized when you perform a image calibration.

[mcgillca]

Yes, the flats are bias subtracted (at any rate, they have been through the  Batch Pre-Processing Script and are in the "calibrated" folder.

Using multiplicative normalisation gives an odd result, as per my original post. I also tried out other methods but found that only by using no-normalisation did I get an answer that looked anywhere near reasonable. The others were about 16 times too low.

Colin

[mcgillca]

I have also found that I can get correct masters by restricting the number of flats. I have found some filters can cope with 10, but others require a maximum of 6. Beyond that, I get the 0.02 value again.

Any ideas?

Colin

[pfile]

this is quite strange - if we can get juan's attention then perhaps you can upload your .fits files somewhere and he can take a look. or maybe carlos can look at it, i dunno.

[Carlos Milovic]

Sorry, this one seems a lot "lower level" for my skills. Let's wait for Juan.

[Juan Conejero]

I cannot reproduce this. As always, I really need some data to try to reproduce/understand/diagnose the problem.

Fits files, created by CCDsoft, taken from an Atik 460ex.

Just to rule out a common cause of errors. Are these files in 32-bit floating point format?

(...back to weekend standby mode )

[mcgillca]

Thanks, Juan.

I will check, but I'm pretty sure they are 32 bit (they've been calibrated via the batch pre-processing script).

I can upload the files - where is the best place to load them, or can you let me have your email so I can add you to my dropbox account?

Colin