WCS information in FITS headers

[perdrix]

Hi Robin,

In the last release (6.2.2) of DeepSkyStacker I added support for using the WCS information from the FITS headers for determining offsets and rotation for image stacking.

I recently had a report of images created by Sharpcap where the WCS information in the headers resulted in mis-registration of the stacked images.

Disabling the use of the WCS data in DSS resulted in the images being registered and stacked correctly.

SharpCap v4.1.14310.0

[admin]

Hi David,

oooh, this is a can of worms... I did a lot of work on this a while back as I had had reports of similar issues. My initial plan was to check the information created by other plate solving engines like Astap and Astrotortilla and then make sure SharpCap matched those. That didn't work out because they didn't all agree (part of which came down to not always agreeing about which way is positive in image pixel space, particularly for the Y direction).

So, one part of the problem is that it may depend on what plate solving engine has been selected by the user in SharpCap.

On another tack, you are mentioning that DSS is using this for registration purposes... In what sense are you using this data? For offsets/rotations between frames? That would definitely be a problem because SharpCap is not necessarily writing the plate solving info for a specific frame into the headers, it is writing the most recent plate solving info it has. That means that the data is likely to be the same from frame to frame during a capture, even if the view is drifting - unless the user plate solves manually at some point during the capture process, which would bring the 'latest' data right up to date.

cheers,

Robin

[perdrix]

Oh dear - DeepSkyStacker makes the (not unreasonable IMHO) assumption that the capture software is plate solving every frame. If even one light frame in an image set lacks WCS data, then DeepSkyStacker falls back to registering frames the "hard way".

If you aren't plate-solving every frame, but instead are inserting the WCS data for the first image in a set (for example) into all light frames, than that has to be a bug (doesn't it?).

Is there any way to force Sharpcap to plate-solve every light frame?

David

[perdrix]

If there's concern over time to plate solve - this topic on cloudynights is really interesting:

https://www.cloudynights.com/forums/top ... on-skyshot

Particularly if he's prepared to open source the code!

Update: It is open source! It's written in Rust. https://github.com/ashtree74/skyshot-solver

David

[admin]

Hi,

historically, SharpCap has always done plate solving on request rather than automatically. Somewhere along the way we had a request to add the WCS info when it was available (I think it helps support colour calibration in PixInsight among other things), and in those use cases the data being close but not perfect worked OK. Now we have a new use case for that data in stacking and that's no longer the case.

Going back in the time machine again, at one point the options for plate solving were Astrotortilla (and related similar solvers), Astap and PS2/3 - all of which are relatively slow (particularly for all sky solving). SharpCap's built in algorithm is significantly faster, but users still have the option of choosing their plate solving tool, so 'fast' isn't guaranteed. I have been talking to the author of SkyShot about the possibility of including it in SharpCap, but that doesn't change the fact that users may have configured Astrotortilla and downloaded all the indices and still want to use it...

Since I'm just coming up to the full release of 4.2, I'm not keen on making this sort of change right now, so it will have to stay as it is. Even when I do get around to this, any 'solve every frame' feature will have to be optional, so I don't think you can rely on the data being exact even in newer versions of SharpCap, let alone for those who are still using versions from the past... Sorry about that.

cheers,

Robin