Saturday, May 11, 2019

#185 - Tuesday, April 30, 2019 - A Stellar All-Nigher: Texas Star Party Night #3

After getting to bed around 2:30 AM on Monday night, I slept in till 10 AM again, showered, and scarfed down a lunch of pasta, meatballs, and veggies.  We had an important date to make: a trip to the McDonald Observatory!  This would be my third visit, and even though it was a technical tour sure to go over my parent's heads, it's fun for anyone to stand underneath a big scientific instrument.

It was Science Fiction day, so my dad chose an appropriate T-shirt.
He is indeed 6'5", in case you were wondering.

My mom and aunt had gotten NASA t-shirts, so I wore mine too, wanting to match.  We took a group photo in front of the building housing the 82", the first scope built there.

The view from up top was awesome.


We got tours of the 107" Harlan J. Smith telescope and the 300" Hobby-Eberly telescope.  My family was quite impressed!


The sky was clear all day, and the forecast looked very promising.  After dinner, it was time to rock and roll.

I trotted down to the lower field after dinner and pulled the covers off my scopes, woke them up from hibernation, connected the computers, and began cooling the cameras.  While I waited, I ran up to the upper field to set up my Nikon D3100 on my other tripod to do what I have been wanting to do for a while now: an all-night, sunset-to-sunrise timelapse.  I set it up at the top of the upper field, set the exposure time to 15s, focused, and let it roll.  I had it connected to AC power, so I wouldn't have to come back and touch it all night.

As darkness fell, I checked focus on the Takahashi on my own Celestron AVX mount (with my uncle's SBIG STF-8300M camera), pointed it at M101, and let it roll on collecting LRGB frames.  Once that was rolling, I moved my attention over to my Borg 76ED on the borrowed Celestron AVX from Derek to try polar aligning it and aligning it again.  That went well this time.  Then I realized I had a problem: I wasn't going to be able to guide it.  The guide scope I normally use, my Orion 50mm, was over on the Tak happily guiding away, and my other Orion 50mm had a different kind of T-shaped adapter that I bought accidentally a while back.  So I was going to have to choose a target that I could do with short exposures.  I settled on M63, the Sunflower Galaxy, and after getting it centered, I set up the sequence for LRGB frames, and let it roll.

While those images were rolling, I walked my family through the "Rising Star" Texas Star Party observing list, which included identifying constellations such as Leo and Corvus, finding the North Star, and seeing globular clusters Omega Centauri and M13.  The whole list is designed to be done naked-eye, but I showed them some of the objects in the telescope as well when I set up my 8-inch SCT later in the night.  We got a good ways through the list.  While facing east looking at Hercules after midnight for M13, my aunt asked what the backwards-C-shaped grouping of stars above Hercules was.  I told her congrats, you've discovered Corona Borealis, Crown of the North!  I was really excited that they were taking such interest in the sky.

After the Milky Way was up, I decided to go ahead and change targets on the Tak to something in that vicinity.  I have attempted the Elephant Trunk Nebula, IC 1396, a few times before, but it's very large and I could never get quite enough signal on it.  I decided to try it in narrowband.  Now, if you punch IC 1396 into the Celestron hand controller, it won't quite have the elephant trunk part of that larger nebula centered, so instead I flipped over to Cartes du Ciel on my tablet and downloaded the DSS (Digital Sky Survey) image for the area, and used that image to choose the centering point.  Then I used CdC to slew the telescope to that spot.  Then in Sequence Generator Pro, I put it on the Frame & Focus mode taking 1-second exposures, lined up where the stars were with relation to the nebula according to the SkySafari app, and centered where I thought the nebula was as best I could.  I flipped over to the Ha filter, and saw that it was out of focus!  I thought that Astronomik filters were supposed to be parfocal with each other, but apparently this wasn't quite the case.  So I slewed over to a nearby bright star, threw on the Bahtinov mask, and got it focused.  To see how that compared to the L channel, I changed to that filter, and it still looked in focus!  So I think that they are close to each other, but a small offset in focus in the L channel can show up as a bigger offset in focus in the Ha channel or something.  I would have to remember to focus in Ha and not L if I was planning on doing narrowband.

Anyway, I got back onto the Elephant Trunk, took a 5-minute test exposure, and...nothing!

Guess I need a longer exposure?  Or maybe I wasn't quite as on target as I thought.  I checked using astrometry.net, and I was indeed pretty close:

On the Results page on astrometry.net, there's a link for seeing your image aligned in Worldwide Telescope.  I reduced the image opacity on my overlaid image to see the imagery underneath.  You can see the Elephant Trunk portion on the right half of the image.

Instead, I moved over to M16 Eagle Nebula to see how that would fare.  Quite well, as it turned out!  And with a 10-minute subframe, my stars were still round!  I couldn't believe it!



It looks like my fix of moving the gears closer together in both axes on my AVX was working beautifully to virtually eliminate backlash.  I wanted to try 15 minute subs, but still wanted to get enough subframes for good signal-to-noise ratio, so I decided to push that experiment off to another night when we have the Milky Way for longer.  I got about seven subframes that night.  After looking at the forecast, I decided I didn't want to take my changes on not getting some complete datasets, so I went ahead and went back over to M101 to get the blue frames that I didn't get previously.

Back over on Derek's AVX with my Borg and ZWO ASI1600MM Pro, the periodic tracking error was pretty bad.  Even at 20s, my stars looked pretty wonky.


But I decided to press anyway, and took a bunch of 30s frames in LRGB.  It's so nice having an electronic filter wheel now, I can just set the sequence in Sequence Generator Pro and it will chug along, changing filters for me.

Once the Milky Way was appreciably high, I switched over to the Dumbbell Nebula, and ran 30s LRGB on that.

I also set up my Nikon D5300 on my Sky-Watcher Star Adventurer.  I had to traverse quite a ways from where my rig was to find another power outlet -- the ranch set up a new power pedestal this year down in the lower field, but were still using the extra-long power cables with outlet boxes every so many feet, which were originally made to run from the RV area onto the field.  This meant that the first box was quite a ways down the cable from the power pedestal, so I had to travel about halfway down the field to find an open one!  Luckily, I've gotten pretty good at finding my way around in the dark out there.  It is so dark I think nearly all of the ambient light comes from the stars.  There is some light pollution in the north, however, from the oil fields.  Anyway, I put the 35mm f/1.8 lens I was borrowing from fellow club member Paul on my Nikon D5300 and took more Rho Ophiuchi region frames.

Single 3-minute subframe, ISO-1600.  The high background is a result of airglow.

Sometime during all of this, I set up my 8-inch Schmidt-Cassegrain on its NexStar SE mount that I brought for visual observing.  After my family went to bed around 2 AM, I needed something to do, so I worked on the "Smoke and Mirrors" Texas Star Party 2019 regular observing list.  It was made up of dark nebulae and galaxy pairs.  Some of the galaxies I couldn't quite snag, but one extremely cool thing I saw was Barnard 86, "Herschel's Hole in the Heavens."  It's a very dark blotch inside of dense open cluster NGC 6520 in Sagittarius, set against the thick background of Milky Way disk stars.  It really does look like a hole in the sky!  It was extremely cool.  I added it to my imaging list for next year's TSP.

After an overall pretty successful night, I shut everything down and covered the scopes as pre-dawn light began to fill the sky.  An extremely slender crescent moon rose above the break in the hills to the east.  As I made my way up to the upper field to grab my Nikon D3100 I had set up there for the all-night timelapse, I saw Venus crest the hills, chasing the moon.  I climbed into bed at 6:30 AM, before the sun rose, but with a nearly fully-lit sky.  An excellent night!

Here's the timelapse video:

I accidentally set the ISO to 1600 instead of something higher like 6400, since I normally image at 1600 on my DSLRs.  So the Milky Way doesn't really come through, unfortunately.  It's still a fun video though.  Some of the bright flashes you see in the video are the now-dimmer Iridium flares, airplanes, and other satellites.  I did catch the edges of two meteors, though!




[ Update May 11, 2019 ] 

Got the Dumbbell Nebula image processed today!
Date: 30 April 2019
Location: Texas Star Party, Fort Davis, TX
Object: M27 Dumbbell Nebula
Attempt: 12
Camera: ZWO ASI1600MM Pro
Telescope: Borg 76ED
Accessories: Starlight Xpress filter wheel, Astronomik LRGB Type 2c 2-inch filters
Mount: Celestron AVX (club member Derek's)
Guide scope: N/A
Guide camera: N/A
Subframes: L: 76x30s
   R: 78x30s
   G: 34x30s
   B: 34x30s
   Total: 1h51m30s
Gain/ISO: 139
Stacking program: PixInsight 1.8.6
Post-Processing program: PixInsight 1.8.6
Darks: 20
Biases: 0
Flats: 0
Temperature: -20C

Not bad considering the lack of ability to guide!  I really love my ZWO camera.  It's so sensitive.  I got some nice structure on the nebula, despite my messy stars.  Deconvolution helped bring that out a bit.

Here are the processing steps:
- Stacked biases and created superbias
- Calibrated darks with superbias
- Calibrated lights with darks and superbias
- So yeah there is definitely something wrong with the biases - calibrating with just darks
- Calibrated lights with just darks
- SubframeSelector
- 1.567 arcsec/px
- 0.059 e/ADU
- L frame 34-4 has the highest score (93.8)
- Registered to L frame 34-4 using StarAlignment
- Stacked with ImageIntegration, Linear Fit Clipping for all
- Applied DynamicBackgroundExtraction to each LRGB channel
- Applied LinearFit to RGB channels with L as reference
- Combined RGB channels with ChannelCombination
- Applied MultiscaleLinearTransform to RGB image for denoising, without mask
- Applied MultiscaleLinearTransform to L image for denoising, with mask (zero-clipped)
- Created model PSF (point spread function) using DynamicPSF for deconvolution process
- Created range mask from stretched L image
- Created star mask from unstretched L image
- Applied Deconvolution to L image
- Stretched L and RGB using HistogramTransformation
- Combined L and RGB channels using LRGBCombination
- Applied PhotometricColorCalibration
- Adjusted curves with CurvesTransformation
- Tried to kill greenish background with ColorSaturation
- Brought in Photoshop for blue halo removal using Noel Carboni tools
- Brought back into PI for one more DBE to flatten remaining gradients

[Update May 24, 2019]

All righty, here is the completed Rho Ophiuchi complex image!

Date: 28 April 2019 & 30 April 2019
Location: Texas Star Party, Fort Davis, TX
Object: Rho Ophiuchi region
Attempt: 6
Camera: Nikon D5300
Telescope: Nikon Nikkor AF-S 35mm f/1.8 lens at f/2 (borrowed)
Accessories: N/A
Mount: Sky-Watcher Star Adventurer
Guide scope: N/A
Guide camera: N/A
Subframes: 1: 86x180s (4h18m)
   2: 95x180s (4h45m)
Gain/ISO: ISO-1600
Stacking program: 1: PixInsight 1.8.6
2: DeepSkyStacker 3.3.2
Post-Processing program: PixInsight 1.8.6 and Photoshop CC 2019 
Darks: 56 @ 40F, 22 @ 52F
Biases: 29 @ 40F, 20 @ 50F
Flats: N/A
Temperature: 28 Apr:  52-53F
30 Apr: 38-46F

So I tried processing it just in PixInsight first, but it didn't come out quite the way I wanted.  Back when I was using DeepSkyStacker, if I stacked this kind of wide-field Milky Way image using settings like I would for a deep sky object, it always came out looking overdone or something like that.  Like too much was brought out.

Processed with PixInsight alone -- not super happy

So I brushed the dust off of DeepSkyStacker and stacked these images there using the Auto-Adaptive Weighted Average stacking algorithm that seems to work quite well on these widefields.  I saved the 32-bit TIFF out of DSS and brought it into PixInsight to process, and was much happier with the result.  I ended up having to use the method in Light Vortex Astronomy's tutorial on using the Hubble palette on narrowband images to remove the pink tinge from the stars, but it worked pretty well!

So here's the process:
- Stacked in DeepSkyStacker 3.3.2, Auto-adaptive weighted average
- Didn't apply changes, saved as 32-bit TIFF, brought into PI
- Crop
- Applied DynamicBackgroundExtraction
- Applied PhotometricColorCalibration for color calibration
- Denoised with MultiscaleLinearTransform, with stretched luminance mask
- Stretched with HistogramTransformation
- Created magenta star mask with ColorMask utility, dilated stars in mask with MorphologicalTransform, then desaturated magenta in ColorSaturation
- Adjusted curves with CurvesTransformation
- Created regular star mask and desaturated red - still not quite killing it
- Applied HDRMultiscaleTransform to increase contrast
- Another round of DynamicBackgroundExtraction
- DarkStructureEnhance script

It turned out quite nice!

If you're interested, I put this image on a bunch of products on Zazzle, from canvas prints to phone cases to throw pillows!  Click here for all of them.  Most items are customizable -- you can change the size (and thus cost) of prints, or change which phone the phone case is for, etc.  Also Zazzle has "faux canvas prints" now that are way cheaper than the real canvas prints!

[ Update June 9, 2019 ] 

While it didn't turn out quite as awesome as I'd hoped, I still got better detail than my first attempt with my DSLR back at the 2017 Texas Star Party.  

Date: 30 April 2019
Location: Texas Star Party, Fort Davis, TX
Object: M63 Sunflower Galaxy
Attempt: 2
Camera: ZWO ASI1600MM Pro
Telescope: Borg 76ED
Accessories: Starlight Xpress filter wheel, Astronomik LRGB Type 2c 2-inch filters
Mount: Celestron AVX (friend Derek's)
Guide scope: N/A
Guide camera: N/A
Subframes: L: 47x30s
   R: 40x30s
   G: 39x30s
   B: 42x30s 
   Total: 1h24m
Gain/ISO: 139
Stacking program: PixInsight 1.8.6
Post-Processing program: PixInsight 1.8.6, Photoshop CC 2019
Darks: 20
Biases: 0
Flats: 0
Temperature: -20C

Date: 22 May 2017
Location: Prude Ranch, TX - Texas Star Party
Object: M63 Sunflower Galaxy
Camera: Nikon D5300
Telescope: Celestron C11
Accessories: f/6.3 focal reducer
Mount: Celestron CGE Pro
Guide scope: Orion ST-80
Guide camera: QHY5
Subframes: 18x300s (1h30m)
Gain/ISO: ISO-1600
Stacking program: DeepSkyStacker 3.3.2
Darks: 24
Biases: 20
Flats: 20
Temperature: 47-59 F

Well, then again, maybe the detail isn't that much better! 😂 
This dataset had some difficulty, mainly in the fact that my stars weren't very round at all, which was a result of two factors.  One was the fact that I wasn't able to guide that night since I didn't have another copy of the right kind of attachment for my other 50mm guide scope.  The other was a result of my field flattener being a little off-kilter from my scope, which made the stars a little more seagull-shaped.


While I was processing, I got some really weird effects after I registered my frames.  I forgot to save a screenshot, but basically I was getting a bright cross-hatch pattern over top of each frame when I auto-stretched, but only the color data, not the luminance.  Now, the reference frame for registration was luminance, but StarAlignment  is just aligning the frames, not applying any effects.  At first, I thought maybe my histogram peak was so low that I was dipping well into the sensor noise, so I went ahead and stacked the frames and combined, but it was bad.  


So I went back a step and aligned the frames from before I applied the score from SubframeSelection in case that was doing something, but it was the same.  So I went back one more step and registered and stacked the light frames without calibrating them with darks and biases -- and that worked.  So this image ended up a little noisy in the color because I couldn't calibrate them for some reason.  Mind you, the luminance frames were just fine, and they used the same calibration!

After that, everything went pretty swimmingly.  I did have to bring it over into Photoshop to use Noel Carboni's Astronomy Tools to kill the little blue halos -- my Borg is supposed to be apochromatic, but it's not quite there.  (The blue halos don't show with my DSLR, but my ZWO is very sensitive and they do show a bit!)

So here's the whole process.

- Created master dark with ImageIntegration
- Skipped bias because still haven't taken new -20C biases (causing problems in previous images)
- Calibrated lights with dark using ImageCalibration
- SubframeSelector on all frames to produce weights
- Scale: 1.567 arcsec/px
- Gain: 0.059 e/ADU, 12-bit
- Noted highest-scoring frame
- Registered frames with StarAlignment to highest-scoring L frame
- Auto-stretch showed weird checkerboard pattern on RGB, not L though
- Stacked with ImageIntegration
- Cropped with DynamicCrop
- Applied DynamicBackgroundExtraction to each channel
- Applied LinearFit to RGB channels with L as reference
- Combined RGB channels with ChannelCombination
- Yeah the hatching is bad
- Tried registering calibrated (but not approved) frames - still bad
- Tried registering non-calibrated lights - hatching gone, so something wrong with darks (
(but only with RGB??)
- Deleting previously-registered frames and replacing with non-calibrated RGB frames
- Stacked non-calibrated frames - didn't want to re-run SubframeSelector, so used NoiseEvaluation instead as the Weight
- Cropped, DBE, LinearFit, combined RGB channels
- Applied MultiscaleLinearTransform to RGB and L, with extracted luminance mask
- Color corrected RGB with PhotometricColorCalibration
- Created sampled PSF (point spread function) from L frame for deconvolution
- Applied Deconvolution to L, 20 iterations, with range_mask-star_mask for mask
- Stretched L and RGB with HistogramTransformation
- Denoised RGB with MultiscaleLinearTransform, with extracted luminance mask
- Combined L and RGB images with LRGBCombination
- Applied another round of DynamicBackgroundExtraction
- Applied HDRMultiscaleTransform, with range mask
- Denoised with MultiscaleLinearTransform again, with extracted luminance mask
- Adjusted with CurvesTransformation
- Imported into Photoshop, ran Carboni's reduce small blue/violet halos routine
- Cropped with DynamicCrop
- Adjusted histogram to reduce background a bit with HistogramTransformation

And there you have it!






No comments:

Post a Comment