I'm using a Canon 1000D DSLR, and have recently begun experimenting with downloading subs in native (raw, cr2) format, as opposed to fits. For some reason, the cr2-files seems to be different in size. How can this be? My cr2-files seems to be 3906 x 2602 pixels, while the fits-files are 3893 x 2595.
Now, the 3893 x 2595 are values I have entered in Ekos (based on information about the camera).
So what is happening here? Is it a matter of me having entered wrong values, or is there a geometrical difference in the formats?
Yes, seems related - but as you say, reverse order of size, as I understand it. Also: I found out because PixInsight complained that subs downloaded as cr1 and subs downloaded as ftis had different geometry.
These are the values reported by libraw and gphoto2, apparently sometimes they are different. Previously, there was a problem with the FITS files having different sizes but that problem was resolved quite a bit back.
Yes, but it seems that it happens here again. Or might there be something wrong with my settings?
Is there something I can try or experiment with? I have no experience with dcraw and manipulating gphoto2 from the command line, but I'd be happy to try something.
No, I haven't seen that yet, at least. Only that cr2-files are slightly bigger than the fits-files. I choose file format in the capture tab, currrently building a dark library with my 1000D. Choosing "native" gives me cr2-files slightly larger than if I choose fits. But so far all fits have the same geometry.
Hi Jasem, Magnus,
This looks like same libraw bug (see indilib.org/forum/ccds-dslrs/1895-dslr-i....html?start=12#16190) that I stumbled upon a year ago.
I can verify from this weeks session it is still present i.e. with Canon 60Da some FITs files (bias and flat) have different dimensions to lights 5202 vs 5184 and makes processing a mess.
I can understand a longstanding bug in libraw, but why does it impact say bias and not light exposures from INDI?
If we want to stay only with FITs workflow is the subframe workaround i.e. add 5202-5184 = 18 still the only option?
Andrew