Saturday, February 14, 2015

First Comet Image - Lovejoy

I almost gave up on processing my Lovejoy images. Dealing with the light pollution from my backyard made it super hard to get a half-decent luminance image.

1. Capturing Data

Capturing the subs was fairly straight forward. Using Horizons from APCC, I could track the comet easily. On top of that, the comet's core was bright and small enough that I could guide with PHD2 on it.

Here is what I did:
  1. Create a sequence in SGP without centering the object (I did not even enter RA or DEC coordinates) with tiny dithering
  2. Use Closed Slew in TheSkyX to slew to the comet
  3. Adjust the scope position slightly such that the comet core isn't in the center
  4. Start Horizons
  5. Start tracking on Lovejoy
  6. Start PHD2 guiding on the comet core
With that I got a pretty good accuracy (less then 1 arcsec)*

I took 
  1. 5 LRGB subs (each 100 sec, Lum=1x1 binned, RGB=2x2 binned)
  2. 5 LRGB subs now guided on the stars
  3. 5 LTGB subs again guided on the comet
The first and third series for the comet, the middle series to get a good picture of the star field.

2. Calibration and Stacking

First, I had to figure out how to align the images. In CCDStack, I used StarSnap to align the images. First, I remove all reference stars and then double click on the comet core. This worked for all images except for the luminance image. In luminance, the comet does not have a well-defined core. I had to align those images manually. For that, I first zoomed in on the comet core:

Then I disable DDP and apply it to all subs:

Now, I could manually align the subs:

I had to set the strength to "medium" - otherwise to mouse movement was too strong.

Now, I could align all images.

The next problem was when I stacked the images, that the stars did not get completely eliminated (I timed the exposures such that the stars don't overlap, so when I stack the, they should be eliminated by data rejection):

After a lot of trial and error and analysis, I realized that this is because of the high background noise. There are always enough bright pixels so that the stars don't get completely eliminated. After more trial and error I found out that by doing an AutomaticBackgroundExtraction, I could lower the background enough to eliminate all stars.

But AutomaticBackground extraction was impossible with the strong vignetting that the subs have:


So, what I had to do was to crop all the subs first, then apply AutomaticBackgroundExtraction to all of them (using ImageContainers).

This is the resulting red stack:

Almost all star trails are gone. But the SNR is not that great :-(

3. Processing

First, I had to align the 4 stacks. I could load them into CCDStack to do a StarSnap alignment. But for some reasons, when I saved them and loaded them back into Pixinsight, they were just white. So, I used DynamicAlignment in Pixinsight to align the RGB subs on the Luminance image and then I cropped them to eliminated the borders.

Luminance Before DBE:

and after DBE:

From here on, I processed the images how I normally process my narrowband images (using this tutorial). With one exception: the first stretch, I do with DDP in Nebulosity. I could not stretch the image with HistogramTransformation myself. And DigitalDevelopment in Pixinsight did not work for me either.

At the end, I had these three images:

Comet Luminance:

Comet LRGB:

Stars: 

I exported these to TIFF format and then finished processing in Photoshop.

First I removed the comet from the star field (from this tutorial):
  1. Duplicate Layer
  2. Filter -> Noise -> Dust&Scratches
    • radius: 10
    • threshold: 6
  3. Use Clone Stamp tool, to eliminate remnants of bright stars
  4. Gaussian blur
    • radius: 10
  5. Make top layer invisible and select bottom layer
  6. Image -> Apply Image: Subtract blurred image from untouched image
Result:

Then I combined them using the last step in http://www.ewellobservatory.com/ccd/comet.cfm. This is how I developed my narrowband images in the past. I really like Clipping Masks in Photoshop as it allows me to keep tweaking all images in parallel (curves, levels, saturation...)

Finally end up with this image:
(click on image for full resolution)

Good, but not great - I wish I could get more detail out of the luminance image. I posted a question on the Pixinsight forum if somebody has an idea how to do that.

* Later on, I wasn't so sure anymore if this was the right approach. APCC Horizon seems to make corrections to the mount only ever few seconds or so. I.e. what might have happened is that PHD2 tried to keep the comet core in the center. Then Horizons made its correction that was now a complete overcorrection.

Other sources that I used:

Rogelio captured his awesome image with these subs: 
40x180 L = 120 min 1x1
20x60 R = 20 min 2x2
20x60 G = 20 min 2x2
20x60 B = 20 min 2x2