Comet Processing with LRGB with RGB Binned - comet subtraction offset

Terri in Vermont

Hi Adam, I hope you're well.  I've captured some comet data in LRGB.   I usually capture my L at 1x binning but color at 2x binning just for data sizes as I use a big sensor.     I'm following your latest set of videos, but when I come to the comet processing it's complaining about different image sizes.. I understand.    

When using both binned and unbinned data in WBPP registration it upsamples all RGB data so I end up with aligned RGB frames that are upsampled.  I didn't initially think about this and didn't notice the implication until I went to subtract my binned RGB comet-only images which are created in a different path with unregistered and not upsampled data.     

Where I run into problems are in the cometalignment tool.    First when creating comet only images.    They are of the smaller size since they go through starxterminator and various steps without any resampling.      So I can not use the luminance bin1x data as a reference.   So I picked a color image close by and it ran and I integrated the comet only into a bin2X R G and B comet only image.    The problem is also there when I go to subtract these comet_only images from their registered images with stars and comet as they are a different scale.  I thought maybe (naively) that I could just upsample the comet only images to the bin 1X size and then subtract but the comet is offset so I get a subtraction not on the comet. 

But another anomaly that I didn't understand.    I thought maybe, because my L data is constant bin 1x and the reference image is an L image, that my comet subtraction would be good for this single set of data, but it's not, and the subtraction is offset here as well and it seems it's offset the same amount on L as it is in all the colors.    The comet path (comet path mask) is very linear and regular.   So I get a black hole in what seems an earlier point than where the comet is in each image.   

What adjustments should I make, or just process RGB and L entirely separately?   Can I do this and still end up with the comet in the same place so I can combine L with RGB?

Terri

BlockHead

Yeah, the consequence of having the binned data is going to give you a headache.

The only reason to use unregistered (comet-only) data is to avoid interpolating the images. 

But in your case- because your original data is binned you have already relinquished this possible benefit- but in your case you are generating an LRGB- so there isn't really a downside.

I think the simplest thing to do is use the star-aligned data (all of the L,R,G,B files) and do everything from this point. You didn't say if you were dong the traditional method of rejection to make the comet-only images of the SXT way. If SXT.. you will obviously need to run it on the star-aligned images. Everything else should proceed normally from there.

Is this close to a helpful answer?

-the Blockhead

Terri in Vermont

Thanks Adam, I did get it all working and using the flow the SXT way and while long, I think it is a good process.  SXT works really well.   I resampled as necessary to make the comet alignment process work properly, and then things aligned correctly with the correct offsets, and it subtracted the comet as it should.   And I did just get to your last video and notice that you mention the fact that someone asked a question about binned plus unbinned data together and you answered it.    For me, given how much information is here, it was helpful to not watch ahead and process in parallel with your videos..   but when I had the problem it caused me to pause and think about the logic of it (as you said in one or more of your videos) and I figured it out.   Like your result, my comet background is a challenge.   It’s very noisy due to the moon being out so my tails are washed out and I a lso see the star trail artifacts and as I attack the bright background it also causes my tails to fade.   I do see a blue ion trail but it’s dim.   I need to work on this a bit more to get the most out of my data.   I haven’t yet combined the luminance but don’t foresee any challenge and in fact it may help because it is at a better native resolution but still at 1.56”/pixel so not terribly undersampled but I usually like to be around 1”/pixel or even a bit better.    My RGB stars, like in your video, are blocks.