Okay, here goes. I’m going to try my best to remember all the steps I took but I’ll probably leave out a lot of details. This image is actually going to be displayed in an upcoming art show called Stellar that is space themed. AFAIK I’m the only astrophotographer participating out of maybe 30 some artists.
First, capture. I had planned to go to the dark site for three nights. The targets don’t get up above the horizon until near midnight so I had plenty of time to perfect my polar alignment and play with the new scope (again, first light for this 8″). I brought my 6″ along just in case something was horribly wrong with the Orion and thankfully there was nothing major. Only issue I could find was that the secondary nuts on the outside of the tube were tightened far too much from factory and actually bent the spider vanes slightly, but enough to make for some big ugly diffraction spikes. I twisted them back into shape as good as I could but I’m going to have to take them out and use some tools to get them perfectly flat. No matter though, onward!
The first night was plagued with dew, humidity was near 100%, not something I’m accustomed to dealing with in the mountains of southern California. It got so bad the PRIMARY actually dewed up, never had that happen before. I decided to instead spend the time building a decent model and getting the polar alignment perfect using EQAlign. I then did some testing with the new scope to estimate what my exposure times would have to be. I started with the usual, 10 minutes @ ISO800. Completely over saturated! That’s a good thing! I decided to try some 5 minute exposures @ ISO800. Still too long. Awesome! Went for 3 minutes, perfect. Definitely overexposed but without clipping the highlights, just the way I like it.
Spent some time planning out the framing. I should mention that was wasn’t my intended target for the weekend, I was originally going to do a mosaic of the Elephants Trunk Nebula area. I discovered that my north eastern horizon was right in the light dome of Palm Springs… I’ll have to come back to the Elephants Trunk from a different site. But the south east looked good so I decided to go for a mosaic of M8 and M20, something I’ve always wanted to do. I haven’t imaged those targets since I first started in AP a few years back so I was excited to see what I could pull out now. Anyway, I noticed that to do a mosaic of M20 to M8 I could keep the X and Y axis of the sensor chip lined up with RA and DEC without having to rotate. AND I can slew in DEC from M20 to M8 and they both fit in the frame. How easy is that?! So I took three frames with a decent overlap of around 25% so I could plan the imaging run for the weekend out. Went to bed at that point because as soon I was done blow drying the optics they went up dewed up all over again… the temp was below freezing too, again, not something I expected at this time in the year but the mountains always surprise me.
Spent the next two nights capturing the data. I split the center frame up over both nights. The quarter moon set at midnight the first imaging night and around 1 am the second night. Before sunset second night I realized I had enough time to get an hour per frame of hydrogen-alpha data and still have plenty of time to capture the final frame in RGB after the moon set. Did some quick tests to determine my exposure time and set it to 10 minutes @ 1600 ISO. This proved to be a mistake as I was overexposing the cores past the dynamic range of the camera but I didn’t notice during the tests exposures because I was so concerned about the background detail being raised cleanly above the sensors noise floor. I had to set my alarm and wake up every hour to change the frame. Around 2am or so I woke up to remove the HA filter, refocus and reframe the shot. Skies were still perfectly clear, seeing looked excellent and the temp was around 40 degrees. Plenty low enough to keep noise under control while doing 3 minute exposures.
Had to wake up a final time around 3:30 to do a meridian flip, refocus and frame again. It’s been awhile since I imaged with a steel tube and I was concerned with focus shifting due to temperature, even a tiny bit at f4 could spell disaster. Got another couple hours of sleep, woke up at 5 am to pack the car and drive three hours to a full day of work!
After work I was far too tired to attempt any processing (in and of itself a HUGE lesson. It’s very tempting to get excited and process your image after you’ve pulled an all nighter. I get far too sloppy without enough rest) so I stuck the camera in the fridge and had it take dark frames all night, I didn’t have any 3 minute frames in my library since I’ve never had a scope fast enough! I did organize and group all the images into the proper directories along with their associate flats and such.
Preprocessing (calibration) and image integration: I decided to do the HA data first since processing monochrome data is quite a bit faster than RGB and that way I could get a good feel of the processing needed for the RGB image. I used Dark Library to scan my dark files for some that were within 2C of temp difference and wrote those down. Took about 100 bias frames (why not, it’s quick and easy) and checked my flats to make sure the dust motes would match.
Pixinsights new preprocessing script was used over my traditional method of using Deep Sky Stacker and I am glad I did, it gave far better results than I could get out of DSS and I’ve got a bit of experience with every possible setting on DSS… Pixinsight worked perfectly using the defaults.
I lined the three frames up to see if I would have to do any dynamic background extraction and only a very tiny amount was required. Again, this was the hydrogen alpha data so I shouldn’t be seeing any drastic gradients from light pollution, just the moonlight. I usually use around 4-6 points per row with a pretty low tolerance (.5-1.0) because the last thing I was to do is remove any nebulosity nor do I want to remove the glow from the milky way. A few attempts and trial and error and I was on my way. I used star alignment to register and merge the images together. I found that I didn’t have to move on to GradientMergeMosaic, frame adaptation in star registration was all I needed. So in the HA image there is zero feather edging between seams and you better believe I was pleased with that.
After merger I did a touch of deconvolution to better resolve the faint wisps of dust. I used the DynamicPSF tool to determine the right psf parameter to use and then spent quite a bit of time building an appropriate deringing support image (basically a star mask). Finally got the StarMask settings right, made the mask and then had to play with the deringing parameters for quite awhile until it looked perfect. Never get sloppy with thise! No matter how fantastic an image looks if there are deringing artifacts on stars they stick out like a sore thumb.
After deconvolution I think I did my initial histogram transformation. Played with this awhile until I was satisfied with the dust levels in the background and of course the nebula was blown out. HDRMultiscale Transform took care of that. I don’t remember what exact settings I used but they were not wildly different than the default, I think a bit less aggressive if anything. I kept the nebula brighter than the background gas and dust as this is what you see when you look in a scope. Then I used a star mask and ATWT to enhance contrast on both small scale structures a tiny bit on really large scale structures, this helped the background to really pop. I am sure I did another curves adjustment somewhere along the line.
Once the hydrogen alpha image was to my satisfaction I then proceeded to start work on the RGB image. I did the same initial steps that I did to my HA image… DBE, deconvolution and then merged the images together. The seams were still visible, even with frame adaptation enabled so I went ahead and merged the images with GradientMergeMosaic using a feather amount of 5 pixels. Looked good so on we go.
Went ahead to color calibration. I don’t bother with the camera white balance, since the camera is full spectrum I just leave it on a custom daylight setting, since we are working with raws it doesn’t matter anyway. After color calibration I started my histogram transforms until I got the RGB image looking similar to the HA image. Spent quite a lot of time building a good star mask and applying a morphological transformation to reduce star size, this was done using morphological selection rather than minimum as I feel this gives me more control. All the stars were a big bigger than I would have liked since again, this scope was faster than I expected. Collimation could have been off a tiny bit, I had used a barlowed laser but I’m sure the focuser moved around a bit from the weight of the camera. Star shape was still good and everything was definitely in focus, something made easier by the dual speed focuser (I’ve never used one before!), especially when getting HA focused. This scope is fast enough that I could actually use liveview to focus with a bahtinov mask through an HA filter!
Anyway, I could not be as heavy handed as I would have liked to be with the star sizes as they would have blended into the background and made weird looking worm like structures everywhere. Time to move on.
I registered the HA image to the RGB image, had to play with the settings a bit to made sure every part of every frame lined up. I was easier than I expected this stage to be and the Baader MPCC did a fantastic job of keeping coma in line at the edges of the frames.
HA and RGB were then blended using PixelMath. There is a great HaRGB script for Pixinsight written by silvercup but I feel I have more control using pixelmath. This is were my memory is hazy. I know I blended the images on two or three different scales with slightly different ratios. I also blended the HA with the red channel, around 25%ha or so. The blend on the lum channel may have been as much as 70%ha but I really don’t remember. After the blend I needed to do a little bit of work getting the color to look right again. Saturation was enhanced in the background and I believe I raised the blue saturation to better show the blue-grey dust around M20.
After this I did a very minimal noise reduction using ACDNR on only the darkest areas, the rest of the image was clean enough. Masks are your best friend with noise reduction.
I probably did quite a bit more tweaks in curves and another iteration of morphological transformation to better get the dust to pop out from behind the stars. Also had to do a bit of work blending the cores of the nebula. The highlights were clipped in the HA image so that was a problem.
So there you have it, if anybody wants me to clarify anything please let me know. I’m probably going to end up reprocessing the image completely from scratch again… I’m waiting on a color calibration card from X-Rite so I can calibrate my monitor as this image is going to be printed and put on display in a big lightbox. Never had to do that before so I’m a bit nervous.
Biggest lessons you can take out of this:
1. Star size reduction does wonders to enhance background detail
2. Don’t get sloppy with your flats if you’re doing a mosaic
3. Fast scopes are magical to use
And the Hydrogen-Alpha: