APT - Wrong sensor size in FITS header

[jif001]

A strange thing happened last night. Over the past week I've been collecting LRGB data on M51, and finished last night having accumulated 7 hours altogether. Today I tried to stack them and got an error from DSS saying basically the files were not compatible with each other. When I looked closer I found that an hour of blue data I got last night has an image size of 4144 x 2822, with all other files at 4656 x 3520, including R and G collected earlier in the night. I've since discovered that every picture I took after that (other targets) had the same issue. I've just tested it again with some dark frames and it's back to normal. 

I'm using an ASI1600MM Pro with a 4656 x 3520 sensor, so I can't figure out why some of my files are the wrong size. The FITS header in these blue files correctly identifies the camera and other parameters such as pixel size and sensor temperature, but the image size is 4144 x 2822. Any idea what's going on here? I've wasted a precious hour of imaging because I can't now stack these pictures with the rest, and I'm keen to understand why.

Another possibly useful bit of information is that I was simultaneously running two instances of APT and the other was using an ASI294MC with a sensor size of 4144 x 2822. Given the numbers that seems too much of a coincidence. Could there have been some 'cross pollination' going on here? If so then why only the sensor size and not every other parameter? 🤔

Edited April 21, 2020 by jif001
Typo

[david_taurus83]

Could you have selected a region of interest by mistake with the 1600?

Have you opened a blue frame and seen how many pixels width and height it actually is?

 

[david_taurus83]

The gain is substantially different in those 2 as well. Looks like your correct dark frame is set at unity but the incorrect blue is much lower, I'm guessing 0 gain?

[jif001]

I don't think I could have selected RoI by mistake. It's not an easy mistake to make, and I've never used it before so it couldn't have been somehow left in that state either.

I hadn't spotted the gain, and every one of the blue pics has a low gain value like that.  But why would that influence the image size? I certainly didn't alter the gain between filters so I've no idea how that happened. Individually the pictures are OK and can be stretched as normal. The gain on the others is 1.

The images are all 4144 x 2822 (image editor file info dialog) and are 23Mb rather than the 32Mb normal file size.

I'm out again tonight trying to get the data I'm missing from last night. Keeping an eye on file sizes as a clue to what's going on!

Edited April 21, 2020 by jif001

[david_taurus83]

You can still use the data. Can stack the subs of the individual channels and them align them in post processing.

[jif001]

Images are a different size so not straightforward to merge with the rest. I'm more interested to understand why this happened so I don't fall into this trap again