Good. Auto is only intended to find the image scale. It will just go trough all image scales. Once you know the image height, adapt focal length accordingly.
There was a problem with image scale of the guide scope focal length which was fixed two weeks ago. I don't know which version this is:
>> knro wrote:
>>I believe I found the issue. INDI::CCD was always using the primary focal length when calculating CDELT1,2. It's now fixed in GIT.
>>It's not set in the driver, it's calculated. So they need to use INDI Nightly or update from GIT. Since this is a major bug, I will also publish it to the stable channel.
Normally you don't have to worry about the values once they are set correctly. If you load an image in ASTAP, you can read the image dimensions in degrees in the viewer statusbar or in the alignment tab the image height as calculated from the header info. So below as indicated 1.32 degrees. You can overwrite this value (fro solving) by entering a manual value there. This is also handy if you try to solve a jpeg or raw file. This value is stored (save settings or by a solve) until new info is read from the header or command line overrides.
If you select in ASTAP temporary the external local Astrometry.net solver, there is a small calculator included intended for Astrometry.net commandline but could be used to check the values. In the example below, I typed a focal length of 580 mm and pixel size of 5 micrometer resulting in an pixel scale of 1.78" and image dimensions 1.15 x0.869 degrees. The correct pixel size should have been 7.8 micrometer and then the image height will match with the header value above of 1.32 degrees. The calculator values or only used for Astrometry.net command line not for the internal solver.
The image scale should be correct within 5% for best performance but offsets of 30 or 40% could still produce a valid solution.
I'm trying by clicking on load and solve from ekos, but it always fails. I see that when I do this, the image vertical dimension is always 1 degree, even if my FOV is different. It seems all my settings, like focal lenght and CCD data, are not taken into account when using load and solve. Is this the expected behaviour?
Go to the mount tab (tripod icon) on ekos and make sure your focal length is correctly set up there. Try saving the config , move back to align tab switch between the camera's and confirm the FOV is correct in Ekos before you hit solve.
If the suggestion of tkottary doesn't work, then you should install the latest nightly build INDI. What I understand is that there was a bug in the pixel scale reporting (keywords CDELT1, 2). Unfortunately there is no way to add an additional command line command in EKOS to override the FOV
There is a simple way to override the FOV setting (my previous remark was incorrect) . You can just enter it in the command line as show below. Just add something like :
This will override and other FOV (focal length, sensor size) setting. So once you found out the image height in degrees just add this value in the command line. Note that -fov auto doesn't work. This only works in the viewer. If the solver is more then 2 degrees from the start position the
will show the FOV used.
It doesn't work for me. It seems like any options specified in the field are not actually passed to the solver, which still shows 1 degree in its little window.
Mind, I'm using the "Load & Slew" button, unlike your screencap which shows you're using "Capture & Solve". I don't think Ekos/Kstars is doing the same things when using the two buttons.
Something strange is happening. The -fov command works only for the Ekos "Capture & solve " button, not for the Ekos "Load & slew" button. That's why I initially reported it doesn't work. This requires further investigation.
The command line option "-fov auto" works for Windows but not for Linux. I have to investigate why that doesn't happen. Note that this option is not ideal. I created it to because I was fed up with finding the image scale for some images without any data. ASTAP goes through all image scales from 9.5 degrees height down to 0.5 degrees with a stepsize of 33% less height each time. Due to the step size this works only for images of reasonable good quality. The option was intended to analyse images from an unknown source, not as a permanent setting.
The "-fov auto" work also in Linux. I didn't see that because in my test virtual machine it was using an old ASTAP version. Note that "-fov auto" disappears as soon you go to an other screen. Then you have to type it again. In the popup window you will see the word "auto". If you see height is 0°, the ASTAP version is too old. It doesn't work for "load and slew" only "capture & solve". It's highly inefficient since it has to search maybe ten times before it finds a match. It helps if you reduce the search area to maybe 10 degrees
The fixed 1 degrees is related to an INDI problem. Maybe Jasem/ @knro can clarify from which INDI version this is fixed.
Start the ASTAP program. It should be in the menu education or execute /opt/astap/astap Click on the stack menu button Σ (ctrl+A), go to tab alignment. See screenshot below . All settings in the blue rectangle are relevant. There is not much you can change. The "field height image" is forced by the FITS header so changing does not help when using the command line. The only thing you could try is to check mark "slow". And if your final image height is close to 1000 pixel you could set "Tetrahedron tolerance" to 0.007." To save these setting go to the viewer, select File, Settings, Save setting. Or exit viewer via File, Exit (and save settings)
If your having a noisy camera with lot of hot pixels you could experiment with option "Calibrate prior solving" This will apply a dark prior to solving . But up to now this option is very rarely required if ever.
Best way to fix solving problems
Almost all solving problems are due to the wrong field height or a too low image resolution. ASTAP solving is very robust if the correct image and data is provided. You better check the image which was used for solving. Ekos writes the "image to solve" at location /tmp Filename will be something like fitsgd5919.fits. Open this image in the viewer and note the image dimensions in pixels and degrees in the status bar of the viewer. The image height should be 1000 pixels or more. If close to 1000 pixels reduce the binning in Ekos to increase the image height in pixels. Then solve this image using the viewer solve button (or the button in the alignment tab) and see if the image dimensions in degrees change significantly after a successful solve. If so the FOV or sensor size is set wrongly in INDI or INDI has still the Cdelt1,2 bug.
- A solve log is written in the bottom of the stack menu. This could indicate why the solver is having problems to solve.
- You can also have a look in the text file /tmp/solution.ini after solving. This will contain warnings and errors. The same you see in the bottom of the stack menu.
- An other indication is try the button "Test button to show tetrahedrons" or use the CCD inspector from the tools menu viewer. If 30 stars or more are detected solving should be easy.
Finally you can also send me the "image to solve" from /tmp and I will have a look.