First Pass at the Veil Mosaic

I finally had enough data to make a very flawed Veil mosaic. Here it is at 1/5 scale:

Click for 1/5 scale image

And here is the link to download the full-scale image. You can use the cloud site's .jpg to look at it, but I suggest you use your own image viewer. Minor advisory, this is a large image, about 10K by 10K pixels and the file size is 12.5Mb.

You should immediately see some interesting things in the full image, some of which I'll detail here. I'll start with the major flaws (all the images below are at full scale if you click on them).

1) What's the big circular thing?

Bad Flat!

This is in the upper left mosaic panel, and results from improper calibration (using an outdated flat, rookie mistake.)

2) What's all that noise?

So Noisy!

Again in the upper left mosaic panel and comes from not only the poor calibration but also insufficient data.

3) What's with all the black pixels?

Pepper Spray

This is found in the upper and middle left panel; it results from pushing the panel a little too aggressively in one post-processing stage.

4) That green thing can't possibly be right!

Streaks to Left of Center: Not Really Green!

So far as I know there aren't any green nebulae. For reasons I don't understand this feature shows up rather strongly in the green channel compared to the red and blue channels; it's quite strong in luminance. I've seen one other LRGB image of this area, and it didn't have these green bits, so I must conclude something in my processing is at fault here.

The Dastardly 52 Cygni

There was one other flaw I had to skate around. I used the script Star Reduction from Blanshan and Cranfield to perform star reduction. It requires a starless image; this can be made either using StarXTerminator (aka SXT) or StarNet2. I used SXT, and it had a lot of difficulty with Magnitude 4.2 52 Cygni (see above image). It essentially suppressed the northern end of NGC 6960! So I switched over to StarNet2 and it handled things fine. 

Notice that StarNet2 didn't handle it 100% cleanly either, giving it what look like (but aren't) diffraction spikes. Looks kind of cool, I think, but then I grew up as a Newt user.

Now for some positives:

1) No seams. I used Photometric Mosaic, and not only didn't it produce seams, but it avoided the pinched stars that GradientMergeMosaic plagues me with. 

2) Color calibration by SpectrophotometricColorCalibration (SPCC). I can't say enough positive about SPCC. My old workflow made color correction very fiddly, with the result depending on how I felt the day of processing more than any sort of objective measure.

3) Not narrowband. For good reason the Veil is usually imaged in narrowband, but that results in false colors. It's nice to see things more as they "really" are (aside from a Green nebula).

4) Not oversaturated. In my opinion oversaturation is very common now. I've said before that a guiding principle of processing should be that less is more; this idea applies to sharpening, stretching, and also color enhancement. I did bump this a half step using ColorSaturation, but pushing it further seemed like too much of a good thing.

5) Reasonably sharp. Here's a neat feature at the south end of of the East Veil. This looks good thanks mainly to BlurXTerminator (BXT). In fact, aside from the one issue seen in SXT, I think the triad of BXT, SXT, and NoiseXTerminator (NXT) is difficult to beat.

An Optical Spiral That's Probably Not a Real Spiral

   

I suspect this curly-thing at center is nothing more than an undulation in a plane seen edgeways, but it does look fun.

So, with those issues listed above this thing obviously isn't done yet. The other night I shot additional light frames for the two panels that suffered from noise (along with same-night flat frames), and I think I can make sure that pepper spray of black pixels can be avoided.

This means doing much the processing over again, but that's life.

Mosaic Detours and a small surprise

 The mosaic is coming along, but there have been several detours along the way.

That difficulty I had with my guide camera resulted in too many bad frames in two of the panels' luminance and red frames. Why those two channels?  I think it's mainly because of how they fall in the filter sequence, but it could just be chance. These will need to be reimaged, meaning no finished mosaic until later this year.

Something was wrong with my luminance flat frame, too. It was leaving a large light circle in the calibrated images:

Lacking a time machine that could let me reshoot the flats as they were at the time the light frames were collected, I opted to create a synthetic flat of sorts by using PixInsight's ABE. This worked well enough, leaving only a few dust motes to be cleaned up by CloneStamp.

One last issue was a sort of swiss-cheese texture produced by the script StarReduction and by StarXTerminator. This was minimized using CurveTransformation twice: first to reduce the brightness difference between the "holes" and the "cheese," followed by a mild stretch to de-emphasize the background.  You could probably use a masked application of MLT to deal with it, too.

Here's a comparison between the starry original and the final reduced star version

Before

After

Vastly better, I think. Here is the portion of the workflow that is used to take luminance from star-filled linear integrated to nonlinear with fewer and smaller stars:

1. Open the original calibrated, aligned and integrated image (it's still linear at this point)
2. Delinearize the original using STF and HT, save as "NL"
3. Open StarReduction script, set target to NL and click the "Generate starless view" button. If you have both StarNet2 and StarXTerminator installed you'll be asked which to use and what options there are for it.  (I used StarNet2 with a 2x upsample.) When that's completed, close StarReduction and save the new starless image as NL_Starless
4. Enhance NL_Starless as you see fit. Certainly make cosmetic repairs. I sharpened it with MLT using layer biases (layer 1 = -0.2, layer 2 = -0.1, layer 3 = +0.15) Save the result as NL_Starless_Enhanced.
5. Reopen StarReduction, set target to NL, starless view to NL_Starless_Enhanced. Choose the reduction method and any associated parameters, and write them down so that they can be used for the other panels. (I used the Transfer method with a scale factor of 0.1) Check "Create new star reduced image" and if you want to use PixelMath or some other means of combining the stars and starless data check "Create 'reduced stars only' image".
6. Click the green checkmark to apply. Save the resulting image as NL_ReducedStars
7. If your image suffers from "Swiss cheese", deal with it now. Save the result as NL_Done. 
8. This isn't actually "done done." It will need cropping and normalizing before it becomes part of the luminance mosaic.

The settings you choose for MLT sharpening, StarReduction, and possible 'cheese' removal will depend on many factors, so play with them to see what what works best for you. It's probably a good idea to create and save process icons once you've found settings you like.

Lessons learned: Shoot flats ASAP after imaging. Inspect light frames ASAP after imaging to see what you collected.

-----------------------

Here's the surprise in panel 6 containing the southern portion of NGC 6960:

Panel 6, luminance, starless version

Look in the lower right corner, that thing that looks alike a ball on a stalk. At first I thought it was an artifact, so I looked at other images on AstroBin. I couldn't find it in any of the images there. Astrometry.net didn't ID it in a plate solve, either, so I went to NINA's Framing Assistant where I could quickly see the area in several surveys. This is what it looks like in the downloadable image files:

Panel area from NINA

And there it was. It shows up in the Nasa Sky Survey and HIPS 2, so it's real and not an artifact. But what exactly is it?  I processed my color frames and got this:

Panel 6 RGB composite

It's got a bluish tinge to it, so my guess is that it's a very faint reflection nebula. So far as I can find it doesn't have a designation. Is there anyone out there who can ID it?

Mosaic Workflow

I've been working on my Veil Mosaic project and here is the first tentative result, the luminance mosaic:

Original Luminance Mosaic

The full scale version of this is 10257x9687 pixels in size! This has a number of issues, but it really was just an exercise in stitching together six panels. That part worked flawlessly. The main issue I have with this is the stars. There are just so many of them that they obscure the nebulosity. The other issue is how to extend my workflow to incorporate the chrominance channels and deliver a full LRGB mosaic.

Most people suggest building an LRGB mosaic from channel mosaics, so that's what I will do. As for the mosaic-building tools, advice is mixed with most people indicating a preference for GradientMergeMosaic. My experience with GMM has been disappointing; many of my images include dense star fields, and GMM has had problematic issues with stars at the edge of panels. Instead, I'll use PhotometricMosaic.

The workflow might go something like this for each panel/channel combination, although the last two steps operate on channel or panel groups. It's assumed you've already created master frames for dark, bias, and flat frames.

1. Cull bad images from light frames (Blink)
2. Calibrate light frames (ImageCalibration)
3. Clean up residual hot pixels (CosmeticCorrection) 
4. Assess calibrated frames for quality and select reference frame (SubframeSelector)
5. Align light frames (StarAlignment)
6. Integrate light frames (ImageIntegration)
7. Sort all the resulting frames by panel; for each panel group use DynamicCrop to insure all the channel images for a given panel cover the same sky and have no edge artifacts from dithering. This insures the channel mosaics have identical dimensions and won't require aligning.
8. Background correction (ABE, DBE, or both)
9. Reduce noise (NoiseXTerminator)
10. When all this has been done, sort the panels by channel. If you're archiving images, this is a good time to send all the intermediate products off to storage, they're no longer needed. Only the images from step 9 will be needed.

Because the luminance images will become pseudo-masks for chrominance they need extra attention. Do these steps for each luminance panel:

1. Create a starless version (StarXTerminator or StarNet2, both have strengths and weaknesses)
2. Enhance the starless image (MultiscaleLinearTransform, UnsharpMask, NoiseXTerminator, etc.)
3. Reduce star bloat (StarReduction), apply the same reduction to all luminance panels.

Care should be taken to insure all the enhancements and applications of StarReduction are identical. This is an opportunity to learn how to use PI Containers.

Within each channel, normalize the images using LocalNormalization. The hope is that LocalNormalization will deal with background disparities and that the splining of PhotometricMosaic will make any remaining issues imperceptible. 

Next, create the channel mosaics by repeating these steps for each channel. 

1. Plate solve each panel (ImageSolver)
2. Register each solved panel (MosaicByCoordinates)
3. Merge the panels (PhotometricMosaic)
4. Reduce noise again (NoiseXterminator)

After you've done all four channels you're ready to combine them all as you would any single LRGB image. 

Taking the channel mosaics nonlinear requires you to try to stretch them in roughly the same manner, perhaps starting with the luminance mosaic and applying that same stretch to each of the chrominance channels. PI lets you do this using the STF process. Having done that you're ready to combine the channels and get on with color balancing, etc.

Notice I'm not using the usual PixInsight noise reduction and deconvolution processes. I think NoiseXTerminator provides superior noise reduction and the PixInsight Deconvolution process? I have never had any real luck with that thing. If your stars are round you're better off using StarReduction, which works exceedingly well and is free, too. Here is a too-quick application of StarReduction:

One pass of StarReduction

This image shows the effect of a single pass of StarReduction. There are a lot of blockish artifacts in this resulting from StarXTerminator being applied to the mosaic rather than individual panels.

With this workflow now defined I can get on with the processing!

The Start of a Mosaic

It's been kind of wild since the last post. We've had many days of air quality alerts, most of which have been for excessive surface ozone, a byproduct of smoke and sunlight and "normal" air pollution. Smoke at times thickened to concentrations similar to what was seen earlier on the East Coast. It wasn't healthy at all; hospitals reported a surge of people with breathing difficulty.

The air quality did improve for a bit and I was able to get out and do a little imaging. In fact, I managed to start one of my learning projects!

One item on my to-image list is a mosaic of the Veil Nebula that spans both the east and west sides. The Veil isn't immense like Barnard's Loop, but it's large enough to require something like a 250mm lens to fit it all in a single frame. My FSQ-106 has a focal length of 530mm and it really needs something like a 2x3 mosaic to encompass the Veil. That's 6 frames, and at about 2 hours exposure time for each it will make a good summertime project that could last into September.

Despite the ever-present smoke I was able to collect the data for subframe 1 which includes most of the East Veil (NGC 6992) and the Network Nebula (NGC 6995): left click the image below, then right click the enlarged image and choose "Open Image in New Tab" to see the image at 1/2 scale:

 

 

For fun, here's a try at a starless version using StarNet2 in PixInsight:

 

This is LRGB with all exposures 120s, L = 20 lights, R = 11, G = 12, and B = 12.

I think I dark-clipped this a little in my processing haste, but it will get another processing eventually.  Here it is tucked into its place in the eventual mosaic:

 

6995 is in the overlap area between subframes 1 and 3. The next target will be subframe 3 to complete the Eastern Veil and give me some practice using PixInsight to create a mosaic.

Some other tidbits from this too-rare night of imaging:

• The QHY-5II guide scope was flawless with over two hours of guiding without a single disconnect. It really does need USB3, it seems. 
• Not only that, but tracking errors were limited to 2 frames in 58. A rate of 1 bad frame in 29 is a lot better than the 1 in 6 that I had experienced earlier this year.
  
• NINA's Advanced Sequencer finished subframe 1 and started subframe 3 imaging without any attention on my part. This was the first time I had tried this. I wasn't willing to do another two to 3 hours of imaging so I reluctantly shut it down at that point.
• More NINA: Its mosaic feature is nicely integrated into the Framing Assistant and sets up the Advanced Sequencer for all the subframes with simplicity.
• Even More NINA: If you want to use the Framing Assistant with images while you're someplace without Internet, go to the NINA download page and grab the Offline Sky Map Cache file (2 GB) It replaces the existing cache folder AppData >  Local > NINA > Framing Assistant Cache. Don't forget to change the Framing Assistant Screen's Image Source setting to Offline Sky Map! Incidentally, installing this allows you to zoom out and use Framing Assistant like a (rather strange) planetarium.
• I seem to have gotten the hang of PI Deconvolution. I don't know why it was so temperamental before, but the key seems to be in the Deringing settings. A Global dark of 0.03 to 0.02 seems to work well, with Global bright typically between zero and 1/2 of Global dark.
  

 

Virgo Cluster Mosaic Progress

Last year I talked about creating a Virgo Cluster mosaic that would be suitable for poster-sized printing. Progress has been slow thanks to the abysmal weather we've "enjoyed" this spring, but it's hardly hopeless to think this won't be finished by the time of ALCON 2018. (The convention is being held here in the Twin Cities, and the notion was that a mosaic poster might be sold as a fund-raiser.)

Here's a graphic of my current progress in terms of the 4x4 grid:

Green = completed sub-image

That's six of the required sixteen sub-images done.

Here's a look at the top row stitched together using Microsoft ICE:

Not much to see in this row other than M98 at far right and M100 at middle right.

The middle two sub-frames need to be reprocessed to minimize the substantial moonlight in them. They were imaged during a first quarter moon which was nearby in the sky. This first row suggests that the finished mosaic will be the equivalent of a single 150 megapixel image taken at a focal length of about 135mm. The sub-images are taken using a Canon T2i (18 megapixels).

I wonder how long it will be before consumer-grade DSLRs are sold with 150 megapixel resolution. The new full-frame Canon 5DS has a 50.6 megapixel sensor!

There's a chance I may be able to get a couple more images done this week to put me at the halfway mark, but the Cluster's availability for imaging is fading fast as time goes on.  The next new moon will be the last opportunity to image it this year.

----------------------

Next time: First attempts at meteor detection using National Weather Service weather radio stations, a home-built antenna, and an SDR dongle. And possibly more venting about the President and my pseudo-conservative, pseudo-intellectual, Congressional Representative.

Virgo Cluster Mosaic Ib: A Less Aggressive Approach

Last time I settled on using a 200mm lens to make a mosaic of the entire--fifteen degree square--Virgo Cluster. This would require something on the order of 15 to 24 subimages.

A club member pointed out that the galaxies probably wouldn't look very good at that scale, and he was right. Here's a simulation of 200mm vs. 700mm.

700mm image

200mm simulation

The galaxies are starting to lose their distinction from stars, pretty much ruining the point of a galaxy cluster mosaic.

His solution is to image only the core of the cluster, an area of about 8° by 5°. That's roughly 1/5 of the entire cluster's expanse, but it does contain the densest concentration of galaxies of interest to visual observers.

I'm going to repeat the calculations from Part I for the AT65 telescope. N = 3.5 (round up to 4) and M = 3.25 (round up to 4). So this can be done with about 16 images using much better optics.

I had some time to put together the layout of the 16 frames:

Virgo Cluster mosaic layout. Blue box, suggested area to be imaged; green markers, centers of subframes; red marker, center of mosaic; red box, area of mosaic.

The final mosaic will have a pixel counts that are about three times the frame dimensions. Since I'll be using a Canon T2i, that will give a 162 megapixel mosaic!

Virgo Cluster Mosaic I: Choosing a focal length

The galaxy imaging I did this spring was so much fun that I'm considering a big project: A mosaic of the Virgo Cluster. One of  the club's members suggested that such a project could result in a poster that would be suitable for fund raising at the 2018 Astronomical League Convention (which the club is hosting). I don't know anything about poster printing and marketing, so I'll leave that to others.

The mosaic, on the other hand, interests me.

My images were all taken using my TV-102; with the 0.8X FF/FR it has a focal length of 700mm and a field of view using an APS-C DSLR of about 1.3° by 1.1°. That's too small a field to make this practical--The Cluster has a size of about 15° by 15°, which is an area about 140 times the TV-102 field. When overlapping is considered it gets even more impractical. I would like to finish this within my lifetime!

Another option is my AT65 (422mm FL) with a field that's 3° by 2°. This means I'd need 38 images not counting the area lost to overlap needed for proper alignment. This is better, but still quite a task, and nearly impossible given the fickle weather around here.

How about a 200mm lens? Its field is 5.1° by 3.9° and gives about 12 images needed before overlap is figured in. That's not bad. A big galaxy like M84 manages to be about 85x74 pixels, also acceptable. It's time to pin down the cost of overlapping.

Consider a single line made of identical overlapping images. We take W as the image width and A as the overlap between one image and the next.

The width of sky covered by the first image in the line is W. Each additional image in the line adds (W - A) to the line length, making it easy to write an expression for the total line width W:

W =W + (N-1)(W-A)

We can solve this for N:

N = (W - W)/(W-A) + 1

Pretty simple, huh? We can do the same for height, letting M equal the number of images of height H in a column, B be the overlap distance, and H the total height of the column.

M = (H - H)/(H-B) + 1

We can fiddle with this a little more to express the overlap as f, the fraction it is of the width or height. This is useful because we'll almost certainly try to use the same fractional overlap in both directions.

f = A / W = B / H

N = (W - W)/[W(1-f)] + 1
M = (H - H)/[H(1-f)] + 1

One more change, let's define the constant F equal to 1/(1-f), giving
N = F(W/W - 1) + 1
M = F(H/H - 1) + 1

Notice that the right hand side of these expressions doesn't necessarily guarantee that N and M are whole numbers. It's up to us to round them up or down to an integer value depending on how well the resulting grid of subframes covers the target area.

One more alteration improves the convenience:

N = F(W/W - 1) + 1
M = F(H/H - 1) + 1

We should check this pair of equations for correct behavior in the case of no overlap. In that case f = 0 is zero and F = 1. This gives 

N = (W/W - 1 + 1 = W/W
M = (H/H - 1 + 1 = H/H

The number of subframes, N times M, is equal to (WH) / (WH), the area of the mosaic divided by the area of a subframe. This is exactly right in the limit that the subframe is much smaller than the mosaic and we can ignore the need for N and M to be whole numbers.

For the 200mm lens, we have (expressing everything in degrees) W = 5.1 and H = 3.9. For the Virgo Cluster both H and W are 15, so

N = F(1.94) + 1
M = F(2.84) + 1

Let's assume a 1/3 overlap rule, so F = 1 / (1 - 0.333) = 1.5.  This gives us

N = 1.5(1.94) + 1 =  3.91 (round up to 4)
M = 1.5(2.84) + 1 = 5.26 (round up to 6)

So--as a first guess--24 images are needed to make a rectangular mosaic of the Virgo Cluster.This could be as few as 15 (3x5), 18 (3x6), or 20 (4x5) depending on the fit and composition considerations. I'll leave that for Part II.

How about a 135mm lens that gives a 9.5° by 6.3° field?

N = 1.5(0.58) + 1 =   1.87 (round up to 2)
M = 1.5(1.38) + 1 =   4.47 (round up to 5)

Only 10 images, but the galaxies would be awfully small.

And for my AT65, with a a 3.0° by 2.0° field?

N = 1.5(4.0) + 1 =   7
M = 1.5(6.5) + 1 =   10.75 (round up to 11)

77 images, a bit much for my taste, and really difficult to get done by ALCON 2018.

Just for fun, let's do the case for the TV-102.

N = 1.5(0.58) + 1 =   16.8 (round up to 17)
M = 1.5(2.38) + 1 =   19.95 (round up to 20)

That's a mosaic with 340 subimages. That's not going to happen.

Time for Mosaics

I imaged many of the smaller/brighter AL Bright Nebula list objects in this season's sky last year, so this year I'm left to clean up the more difficult objects. These can be more demanding in terms of the sky I have to work with (tree-locked and bright), or by virtue of their larger size. In the former category are reflection nebulae and Lynds brightness category 6 objects.

Here's an example of the former, IC 4954 in Vulpecula:

This was shot in H alpha because it has a weak emission in that, and the narrowband filter helped to reduce the light pollution and moonlight. This is given a 2 for brightness.

For faint there's DG 191, shown in the previous post. There's no official brightness for this, but I estimate it as a Lynds 4. Most of it is visible in weak Ha emission, but the lower right quarter or so has a substantial reflection nebula component. Because I wanted to catch the blue tint I shot this in RGB. I should have gone with HaRGB, but I wanted to complete this in one night.

Sh 2-134 was too large for a single frame at 400mm. I combined two images to create a minimal mosaic using Photoshop:

I used Photoshop's default settings for this and it turned out pretty well. There is a transition line visible along the top of the lower frame, so it could use some reprocessing.

Next up is the huge LBN 270 (3 degrees across) in Cygnus, which will require a 2x2 mosaic. I managed to get the southwest frame last night during a couple of hours of clear sky!

Eventually I'd like to patch together a mosaic of Sh 2-240 (Simeis 147 in Taurus), a large supernova remnant in Taurus. That might take five or six frames!