Adding long focal lebgth data into widefield images

[rdryfoos]

So--how to add long focal length data into widefield images.  I have combined long and short FL data--but it always results in a FOV corresponding to the long FL image.  I know this can be done in PS.  I also know there HAS to be a way in PI.  Anyone know how to do it?

Rodd

[rdryfoos]

Come on aficionados--or designers--this has to be able to be done.  How? 

[pfile]

the size of the output of StarAlignment corresponds with the reference frame. so choose a widefield reference frame and you'll get registered images from the longer FL instrument which are mostly black with a patch of signal in the middle.

rob

[RickS]

the size of the output of StarAlignment corresponds with the reference frame. so choose a widefield reference frame and you'll get registered images from the longer FL instrument which are mostly black with a patch of signal in the middle.

You can use StarAlignment like this to match a longer FL integration with a wide field integration and then combine them with GradientMergeMosaic.

[chris.bailey]

If the wide field data is very wide it can be worth turning on distortion correction and changing the registration model to 2-d surface splines.

[rdryfoos]

the size of the output of StarAlignment corresponds with the reference frame. so choose a widefield reference frame and you'll get registered images from the longer FL instrument which are mostly black with a patch of signal in the middle.

rob

  But then I have to combine the black bordered image into the widefield and the black areas will ruin the image.  Are you suggesting turning the whole thing into a mosaic?

[rdryfoos]

the size of the output of StarAlignment corresponds with the reference frame. so choose a widefield reference frame and you'll get registered images from the longer FL instrument which are mostly black with a patch of signal in the middle.

You can use StarAlignment like this to match a longer FL integration with a wide field integration and then combine them with GradientMergeMosaic.

  Is it the same process as constructing a mosaic--the same tutorial?  Essentially turning it into a mosaic with 2 panels--1-the long FL and 2-the short FL?

[pfile]

i think that is what rick is saying. however since the narrow-field image should have black pixels where there is no signal, those black areas should be rejectable during imageintegration even without using mosaic tools... it could however look a little rough around the edges of the narrow-field image due to aliasing.

rob

[rdryfoos]

i think that is what rick is saying. however since the narrow-field image should have black pixels where there is no signal, those black areas should be rejectable during imageintegration even without using mosaic tools... it could however look a little rough around the edges of the narrow-field image due to aliasing.

rob

  I know its a very popular technique that is done amazingly ell in other programs. PI must be able to do it as well....perhaps not as conveniently....but better.  I have done a few mosaics using the PI tutorial--I still need to use it and can't do it from memory.  So if there was a tutorial--or at least a listed work flow, it would be great.
Rodd

[RickS]

Is it the same process as constructing a mosaic--the same tutorial?  Essentially turning it into a mosaic with 2 panels--1-the long FL and 2-the short FL?

Yes, you're using GMM to try to get a good match between two panels just like a mosaic.

i think that is what rick is saying. however since the narrow-field image should have black pixels where there is no signal, those black areas should be rejectable during imageintegration even without using mosaic tools... it could however look a little rough around the edges of the narrow-field image due to aliasing.

Worth a try, Rob, but you may find that the subs from different systems are mismatched enough that rejection of good data becomes a problem.

I know its a very popular technique that is done amazingly ell in other programs. PI must be able to do it as well....perhaps not as conveniently....but better.

Here's an example where I did this in PI: https://pixinsight.com/forum/index.php?topic=12279.0

This image is a combination of two wide field data sets (TEC160/STXL16200: image scale 1.069 arcsec/pixel and Ceravolo 300/PL16803: 1.26 arcsec/pixel) and one longer FL (RCOS 14.5/STX16803: 0.55 arscsec/pixel):

I calibrated/registered/drizzle integrated (2x) the two wide field data sets separately and calibrated/registered/integrated the RCOS data set.  Then I registered the C300 and RCOS lum against the TEC lum.  Attempting to combine all three data sets with GMM I had some problems with the stars in the RCOS data causing issues at the edges.  I tried CloneStamping out the problematic stars but that didn't work, so I build a feathered elliptical mask using the GAME script and a bit of PixelMath and Convolution and clipped the RCOS data to it.  After that I was able to combine all three lum masters in GMM and the rest of the processing wasn't particularly noteworthy...

I'm working on another image using the same three scopes right now... will report back if I find a better way

[pfile]

actually every time ive tried to do this the FHWM differences were enough of a showstopper regardless of the method. at any rate that is probably a function of the junk telescopes i am using rather than anything purely intrinsic to the process

rob

[msmythers]

rob

I've had the same problem. I have thought about just removing the stars from one image completely and see where that gets me.


Mike

[RickS]

actually every time ive tried to do this the FHWM differences were enough of a showstopper regardless of the method. at any rate that is probably a function of the junk telescopes i am using rather than anything purely intrinsic to the process

It's all about compromise, Rob

[rdryfoos]

actually every time ive tried to do this the FHWM differences were enough of a showstopper regardless of the method. at any rate that is probably a function of the junk telescopes i am using rather than anything purely intrinsic to the process

rob

Not necessarily--I have combined data from a TOA 130 and a Televeue np101is (just strait registration and integration), 2 very nice scopes, and as mentioned above, the FWHM values were different (due to seeing, filters, long and short FL--what ever).  It worked--but many of the stars looked a bit strange at 2:1 or 3:1 because different size stars were combined.  essentially it ended up that the short focal length data was used in the long focal length image--kind of the opposite of what we are talking about because you lose the FOV of the widefield.