So what is the route a new user should go in order to find out what dimensions he should enter in the Ekos requester that's a asking for his cam's image size?
I am still wondering why we don't take advantage of the existing code and have CR2s converted.
I think Herrhausen is asking for an additional command line tool to convert DSLR RAW files into FITS exactly the same way, the gphoto2 driver is doing it.
Either a command line tool or the same functionality incorporated in Ekos. This would enable users to convert older, already existing CR2 flats to FITS files of the same resolution as their older FITS lights.
As copello had suggested in another thread dealing with the same problem, a small change in the code did the job just fine.
In gphoto_ccd.cpp the line
if (read_libraw(tmpfile, &memptr, &memsize, &naxis, &w, &h, &bpp, bayer_pattern))
was changed to
if (read_libraw("/tmp/myfile.CR2", &memptr, &memsize, &naxis, &w, &h, &bpp, bayer_pattern))
I copied the CR2 file to /tmp/myfile.CR2 and Copello's hack converted it for me in the exact same way it is done when FITS photos are taken in Ekos. What I am asking for is a file requester that would ask for the CR2 file to be converted. Ekos would then treat this file like any other "tmpfile" input. This way the user would get a FITS file with the correct dimensions.
I've tried a bunch of other converters but without exception they ALL produce output files that are not of the same dimensions. Cropping the files did not do the job either as Regim did not accept such files.
OK, I've changed the geometry for my 1000D to the same values as given by PI on the CR2-file. And it seems to produce fits-files with the same geometry now. So at least here, I think I now have control over the geometry.
When we are at it: is there a way to disable the preview of CR2-files in Ekos? It's big and hijcacks the screen in a way the Fits-viewer does not.
While in the CCD tab in Ekos, click the Options button to the lower right of the screen. Select Ekos in the popup and the Capture tab. There you can dselect the DSLT preview.
I got headaches from the geometries caused by my specific workflow, just 1 example:
- I need to catch 30 rapid dusk/dawn flats with constant light so I use local download (Raspberry Pi 3) and CR2
- these later converted to FITs for Siril processing.
- python check all options 5202/5184 x 3405/3465 and outputs only 5184 x 3405
I'm surprised the 600D indeed all canons don't suffer from same issue since as I understand the root bug lies in libraw.