Switching to a dedicated Astro Camera – ZWO ASI 533 MC Pro

Right after finishing up “my” shooting of M31 – the “Andromeda Galaxy” – with my Skywatcher 72ED and my Nikon D7500, I switched to a dedicated astro camera, a ZWO ASI 533 MC Pro. It has two big advantages over the standard DSLR:

  • It can be actively cooled so I can always take my images with the same chip temperature (and re-use my darkframes)
  • It does not have IR Cut Filters and therefore accepts light in wavelengths the unmodified Nikon D7500 cannot
The first result produced with the ZWO ASI 533 MC Pro – 80 x 70s, 21 x 180s

The initial integration of the images (80 x 70s, 21 x 180s for a total of 2:36h) shows a nice result. And zooming in shows the superiority of a camera sensor cooled down to -10°C (plus I used darkframes this time): almost zero background noise!

Note the very low background noise in this 3:1 zoom of the image.

The rest is the same basic processing workflow that I have done with the Nikon images: Color Calibration, Stretching, some Curve Transformation…done.

M31 as developed in PixInsight

A direct comparison of both images – the one from the Nikon on the left, the one from the ZWO ASI 533 on the right – shows: they both work well, the Nikon’s image is an accumulated 12+ hours, the ZWO ASI 533 is an accumulated 2+ hours of exposure time.

A direct comparison at 1:1 zoom – Nikon on the left, ZWO ASI on the right.

The overall image also shows: the sensor of the ZWO ASI 533 is smaller but the items appear larger when compared to the Nikon. Photographing all of Andromeda probably requires a Mosaic of 2×1 or 2×2 frames, depending on what should be achieved.

Just out of curiosity (and stepping ahead a few month) I am also showing a comparison with a monochrome camera, a ZWO ASI 1600 MM with a Luminance filter:

And adding a ZWO ASI 1600 MM Luminance Frame to the equation…

Mono Cameras are a whole different thing – and I will talk about them later. But for now, let’s conclude: taking Deepsky Images with a DSLR is (for broadband targets such as galaxies) completely fine! The advantage of a dedicated color astro camera is mostly the ability to cool the sensor (and reduce the background noise reliably and reproducible) and the lack of the IR (Infrared-cut) filter on the One-Shot Color Camera (OSC).

Much more important are a good and steady mount that allows for pinpointed stars when guiding for say anywhere from 70s to 180s and a good telescope (preferable with a motor focus but more on that also later).

Suggestion from my side: put your money in the mount first! Then upgrade the optics and finally the camera. Pick your targets wisely and match them to your equipment. And don’t save money on the processing side! Like in many other cases: buy wisely! Those that buy “cheap” will either give up or buy twice…

Posted in Astronomy | Tagged , , | Leave a comment

A final DSLR Masterpiece – M31, the “Andromeda Galaxy”

By September 2020, and after the initial positive results with astrophotography from my backyard, I had invested into some “upgrades”. Initially, I was out just with the tripod and the Nikon D90 and several of my Nikon lenses.

Next thing that happened: the Nikon D90 got replaced by a Nikon D7500 and shortly afterwards, the tripod was equipped with an Omegon LX2 mechanical star tracker. Which did not last long (I mean, it did its job but had limitations) – it was replaced with a Skywatcher HEQ5 Pro Mount and the Nikon lenses made room for a Skywatcher 72ED telescope, together with a Skywatcher 50ED Guide Scope and a ZWO ASI 120 MM Guiding Camera. Also, a switch was made from a simple intervalometer, controlling the Nikon exposures to using dedicated astrophotography software. The Nikon D7500 remained in place for some last set of photos of M31, the “Andromeda Galaxy”.

Images were taken during several sessions between September 2020 and November 2020 – a total of 12:28 hours of exposure time.

PixInsight’s Weighted Batch Preprocessing Script with the Image Information

This time, some darkframes “survived” the times so I decided to apply them globally, although they did not match the temperatures of each image session – remember, the Nikon D7500 is a regular DSLR so darkframes should match the temperature the light were taken at.

Pre-processing a large number of images, including the creation of three master darkframes, takes a little while – my computer was busy for some 26 minutes. Like before, the next step is to debayer the images and then do the star alignment – this time, I had sorted out the “bad images” so I spare you that step.

PixInsight’s DynamicBackgroundExtraction Process in action

With an Deepsky Object (DSO) that large, the background extraction needs to be performed more selectively using the DynamicBackgroundExtraction process which allows the user to select what is “background” and what not.

M31 after the DynamicBackgroundExtraction process is done

Next are then the BackgroundNeutralization and the PhotometricColorCalibration processes – once done, the colors have been adjusted more to their natural look.

The “grand spiral galaxy” after the color correction processes

The large number of source images has already made sure that the background noise is very low but running a bit of noise reduction makes it even better. Now, we are ready to stretch the image finally.

“Stretching” the image helps with the full dynamic range – now, we have a “true” image that would show like this even outside PixInsight.

What comes now – again – is a matter of “taste”. But one thing I want to show you is how to work with the “galaxy” and preserve the stars around it in their current state (or handle them individually).

There is a PixInsight (and also PhotoShop) process written by Russel Crowman (see his Website) which allows PixInsight to “separate” the stars from the background. If you tell the process to also generate a Star Image, you can work both on the background and on the stars separately.

The result of running StarXTerminator against the M31 image.

This now provides almost endless possibilities – the sample below shows in the upper left corner an image of the background that was HDR-tunes in Photoshop, the regular image center bottom and the star-mask (intensified in Photoshop).

PixInsight with a variation of the M31 background and intensified star mask

Whatever we do to the individual images – we can use PixelMath to combine them afterwards:

Using PixelMath to combine two or more images

So for the moment, this is the result in PixInsight that comes from the long hours of photographing M31 in 2020

M31 as developed in PixInsight

This was the last image that I produced using the Nikon D7500 DSLR – afterwards, I switched over to a dedicated AstroCamera, a ZWO ASI 533 MC – but that is a different story…

Posted in Astronomy | Tagged | Leave a comment

Another Go at M81 and M82

With the success of the “Leo Triplet” and with the Nikon D90 replaced with a Nikon D7500, I decided to take another “go” at the big galaxies in Ursa Major, M81 and M82.

This was also my first “multi-session” attempt, with the images being shot over several nights (and with slightly different focal lengths).

  • May 25, 2020: 37 images of 45s each with the lens settled at 250 mm, f/5.6
  • May 28, 2020: 42 images of 45s each with the lens settled at 300 mm, f/5.6
  • May 28, 2020: 12 images of 60s each with the lens settled at 300 mm, f/5.6

That gives a total of 91 images with a total of 71 minutes. One of the first questions people often ask is: “Can you combine images from multiple nights?” and the second one, a bit more complex, is “Can you combine images from different telescopes?”

Both is possible and the StarAlign process in PixInsight will take care of that. The key is the Reference Image: all other images will be aligned (and resized!) to match the stars in the Reference Image. So you should make sure that an image with the shortest focal length (in my case one of those from the first night) acts as the reference – that will shrink the other images (whereas doing it the other way round would “enlarge” the images from the first night, creating unwanted artifacts.

In my case, the StarAlign process took exactly 6 Minutes but only 64 images aligned. Why?

One of the images that failed to align

In this special case, the reason for the failure to align was a failure to sort out the obviously unusable images first – this one failed on tracking, producing “star trails” which then failed the alignment process. Actually, this is a good thing because it sorts out “bad data”.

But there is a better way of doing that: the PixInsight process for the initial visual inspection is named Blink. It simply loads all the images and then “flips” through them, either automatically or manually.

Blinking the Images – pay attention to the star trails!

As you can see from the video: there are good images, bad images and “not quite too bad” images in there. Usually, I would have done this before I aligned the images (saves processing time!) but now, we can also use Blink to examine the effects of the StarAlign process:

Blinking the aligned images – take a note of the frames that show a “distortion”

Did you notice the “distortions” some of the frames showed? And how the field of view shifted suddenly to an area of the image, showing a 45° turned image? That is what StarAlign does for you: it aligns the images, rotates them where required and resizes them as needed. After a bit of clean-up (removing the distorted frames) this is what we are left with:

Staraligned and cleaned up, these are the images to stack

At the end of the day, our “good enough to stack” images turn out to be some 58 frames out of 91. Did you notice the satellite trails on some of the frames? I mentioned that ImageIntegration is taking care of these later but let’s first stack the images without and rejection algorithm:

ImageIntegration with no rejection algorithm set

Not horrible, but also not what we want – let’s do ImageIntegration again with what is called the “Sigma Clipping” rejection algorithm.

Now, ImageIntegration is producing three results – let’s look at the rejection map first (high-clipping):

The “High-clipping” Rejection Map

The High-clipping Rejection Map shows all information that was not integrated into the final image based on the algorithm – you can clearly see the satellite trails here but also some of the “fuzziness” around the stars.

The second rejection map – the “Low-clipping” – can be used to crop the image later. It very clearly shows which areas of the image are covered by all stacked frames and which ones are not.

The “Low-clipping” Rejection Map with two possible cropping scenarios (placed by myself for illustration)

Last but not least the integrated image:

The integrated and auto-stretched image

Before we continue with any other activitiy, I am goind to crop the image to the area of interest just around the four galaxies visible in here:

PixInsight’s DynamicCrop in action

Why cropping now and not later? Because all other operations will a) be faster on a smaller image and b) when it comes to working with averages across the image, the algorithms will just include data that also remains in the final field of view.

Now that we got the stacked and cropped image, the first thing that should be done is to PlateSolve it. The ImageSolver in PixInsight needs to know some RA and Dec Coordinate that is within the image but in this case, it is easy: M81 provides the required information.

The ImageSolver set up to plate solve the M81 & M82 image.

It always pays off to have a look at PixInsight’s Process Console window:

The result of the ImageSolver Process

Here, we are getting some valuable information about the image:

  • The Resolution (3.397 arcseconds per pixel)
  • The Rotation (91.371°),
  • The Observation Time (taken from the first and last stacked image)
  • The calculated Focal Length (228.28 mm)
  • The Field of View, and
  • The Image Center coordinates as well as the corner coordinates

If you can platesolve the image, you can annotate the image:

The annotated Image with M81, M82, NGC3077 and NGC2976.

The rest is “cleaning & adjusting” – in the following order:

  • Background Extraction
  • Background Neutralization
  • Photometric Color Calibration
  • Noise Reduction
  • Stretching

And if one wants, PhotoShop and some additional filters can also be applied, all a matter of “taste”.

The final image, showing M81 & M82 – taken with a Nikon D7500 and a regular Telezoom Lens @ 225 mm

So, last thing to do is some information just for reference – first, the finding chart so you know where in space you need to look at:

The Finding Chart for M81 & M82

Then, some information on the four galaxies contained in this image:

  • Messier 81 (also known as “Bode’s Galaxy”) is a grand spiral galaxy some 12 million light-years from us. It measures roughly 90.000 light-years across.
  • Messier 82 (also know as “Cigar Galaxy”) is a starburst galaxy of approx. 37.000 light-years diameter.
  • NGC2976 is also a spiral galaxy, approx. the same distance than M81 and M82
  • NGC3077 is a small irregular galaxy, also approx. 12 million light-years from us.

All four galaxies are members of the M81 group which contains total of 34 galaxies in the constellations Ursa Major and Camelopardalis.

Posted in Astronomy | Tagged , , | Leave a comment

Photographing the Leo-Triplet with a Nikon D90 and 135mm Lens

After my initial attempts at M81 (and missing my target), I had a second attempt just a few days later on May 16, 2020. Again, the Camera in use was my Nikon D90 with a Nikkor Zoom Lens set to 135 mm.

This time, the target was the southern region of Leo, specifically the area where I assumed the “Leo Triplet” to be. The “Leo Triplet”, that are three galaxies very closely together: Messier 65, Messier 66, and NGC 3628.

The “Leo Triplet”, taken with a Nikon D90 and a 135mm lens (52 x 30s)

This time, my aim was spot on, which is not too difficult since a bright star, Theta Leonis (also known as “Chertan”) was a good aiming point. And this is the annotated version:

The annotated version of the “Leo Triplet”

Zooming in to a 1:1 ratio, this time, the structure of the spiral galaxies becomes “visible” (though you still need a good idea of what you are looking at) – but again: just a DSLR and a 135 mm lens.

The 1:1 Crop of the “Leo Triplet”

If you want to try for yourself – here is the finding chart:

The fining chart for the “Leo Triplet”, showing the orientation and size of the Nikon’s capture.
Posted in Astronomy | Tagged , , | Leave a comment

An 18-Month Journey into Astrophotography

By now, it has been 18 months since Covid-19 brought me back into Amateur Astronomy and Astrophotography. Time to recap a little bit and go back to the first sessions taken from my backyard.

I have kept all “not completely messed up” images taken – way back when with my old Nikon D90 – but my post-processing software and skills have developed a little bit.

A 45s exposure at 3.200 ISO, using the Nikon D90 with a Nikon 105mm at f/2.8

The single exposure above is one of a set of 33 – and the only information I have about it is that I tried to aim at the Ursa Major area. What exactly the image is showing? I don’t know. The easiest solution to get an idea of what is in that image is http://nova.astrometry.net/ – simply upload the file and wait for the result.

The results from nova.astrometry.net – “blind solving” in minutes

The beauty of Astrometry.net: it blind-solves without you having to install a platesolver locally. Once done, I can continue to work with PixInsight (which from my perspective is a “must have” when it comes to working with astrophotos).

With the center coordinates now known, I run the Imagesolver process in PixInsight, provide the coordinates for RA and Dec as well as the Focal Distance (105mm) and the Pixel size (of the Nikon D90 Sensor – 5.5 micrometer) and hit OK.

Processing Console of PixInsight after successful platesolving of the given image

Why performing the step (again) in PixInsight? So that I can update the image with the relevant data and re-use the information in other processing steps – such as creating the Finding Chart straight from within PixInsight.

The result of the FindingChart Process in PixInsight (as of Version 1.8.8-12)

Looking at this, it seems that I had aimed at M81 and M82 but missed by a tiny fraction. Back then, I was simply using my tripod and “visual aiming” – a close miss but still a miss. So what else is in that series of images (after all, I did spend time on it so why simply throw them out?)

First of all, the Platesolver has also “told” us that the image is rotated by 92.1° – but PixInsight can take care of that as well, so I rotate the image by 90° counter-clockwise. And have to repeat the Platesolver because the rotation screwed up the earlier solution. Then, the AnnotateImage process put an layer of object annotations over the rotated image.

The rotated image with the AnnotateImage Overlay (incl. the NGC Catalog)

Well – I have missed M81 and M82 but I got NGC3359 and and IC2574 (not in the cropped image above) in the overall image. Time to “develop” the image…

Pixinsight – DeBayering and Stacking

Since the images were taken with a color camera, the first thing we got to do now is taking care of the so called “Bayer Matrix“. PixInsight can do this – the process is called Debayer. Given that I do not have any matching Dark Frames, Bias Frames, and Flat Frames, this is the first step prior to aligning the images that can be done. The good thing about PixInsight is: it does not alter the original images but instead creates a set of new images so if something goes wrong (or if you later want to come back to old data) you have unchanged originals to start over with.

Now, we got 33 images of the same area – what we want to do is “placing them over each other” to enhance the signal. However, the Earth rotated during the process of taking these images and therefore, the stars in them do not “align” in each of the frames.

A process called StarAlignment is what solves this “problem” – again, it can be done inside PixInsight.

The PixInsight StarAlignment dialog with the Reference Image set and the remaining images places as Target Images.

Again, the aligned images are written to a dedicated location, creating yet another set of images along the line. The process can take a little while, depending on the number of images and your processing power (or rather that of your computer) – in my case, all images were aligned after roughly 2 minutes.

Stacking the aligned Images

With all the images debayered and aligned, we are ready to “stack” them on top of each other. While it is not “as easy as just that”, the ImageIntegration process in PixInsight takes care of that.

Other than adding the input images, the only other setting is the Rejection algorithm – this is what takes care of “unwanted information” such as satellite trails and other disturbances.

The rejection-high map, showing the “high-value” point removed during the stacking process.

The stacked image itself now looks like this:

The result of the ImageIntegration Process – 33 x 45s for a total of 25 minutes of total exposure

Noticed that “dreadful” background pattern with the image getting darker towards the edges? This is caused by the Nikon lens and if I would have taken flats, it would probably not show. But again, there is a solution in PixInsight called AutomaticBackgroundExtractor.

It is the “lazy approach” to a “not-so-great” image – I ask PixInsight to automatically place a grid over the image in regular intervals, determine the background brightness, construct a background map and then subtract the background map from the actual image.

The background map created by PixInsight for the image above.

Removing the background has a stunning impact on the image…

The image after background extraction

Even though I used the automated process, the impact is significant – the number of stars visible has suddenly jumped up and the circular pattern is (mostly) gone.

Performing Color Calibration & Noise Reduction

Next steps – although on this image not really required – is color calibration and then (necessary) Noise reduction.

A zoomed-in view of the image reveals the noise in the background

The camera sensors are never providing a clean and clear colorized area – instead, the background noise is fluctuating at a low level, represented by the various shades of dark blacks, blues, greens, etc.

This could be reduced significantly by using dark frames and flat frames (which I do not have to that session) but PixInsight comes with a series of (complex) Nose Reduction algorithms that require a bit of time to understand and test…

The reduced background noise after performing noise reduction

Stretching the Image

The data in the image is generally “very dark” – most of what was photographed is “black void”, after all. Which also means that the image’s histogram shows very little “dynamic range”. “Stretching” the image from its linear state to its non-linear state means adjusting the histogram without losing too much information. At the end, some slight transformation of the luminance curve increases the contrast, some morphological transformation can help to reduce the visibility of too many stars (to pronounce the ones that are remaining) and there we go:

The final image after some careful twisting of the histogram.

And the annotated version as well:

Annotated version of the image.

The “rest” is an artist’s freedom of image development – leaving the image “as is” for for example using PhotoShop filters to add star spikes and flares to the stars is up to every photographer herself or himself… there is no “right or wrong”, there is only “taste”.

Using PhtoShop and ProDigital Software’s StarSpike Pro 4 to add some color to the image

Conclusions: well, for a very first image, this was not bad. Especially not if you consider that the camera used was a regular DSLR with a 105 mm lens that is usually used for macrophotography. Also, the camera was mounted on a regular tripod, which means no compensation for Earth’s rotation at all.

There are some very faint objects in that image – but IC2574 and NGC3559 are the ones that are clearly visible. Obviously, at 105mm not in any detail. But one should consider: IC2574 is a dwarf galaxy some 13 million light years from us and IC2574 is even further out: approx. 50 million light years…

Posted in Astronomy | Tagged | Leave a comment