Northwest Astro

Badly Processed M81

Better Processing on M81

Take a quick look at the two images to the right. One is a bad example of how to process M81. Maybe not bad for a first attempt (which it was for me), but not a great galaxy in general. It has very little in the way of features, and you really only get a vague idea of what shape and details it has. One is a better example of how to process the same target. While it is noisy, it still has much more detail visible in the galaxy. You can see the characteristic dust lanes near the core, and more detail in the spiral arms. Overall, it is a much better image.

Both images come from the exact same data. 16 exposures of 5 minutes each, aligned and stacked. In fact, they both come from the same final stacked image. So why is there such a huge difference between the two?

When processing, I’m finding out the hard way that you can’t be too heavy-handed when you do the final stretching work and post-processing. If you start doing that, then you risk losing the very detail you are trying to extract from the noise.

Don’t fall for the temptation to do curves adjustments in one or even two passes. Take time on it, and do it very gradually if you have to. And resist the temptation to ‘clean up’ the image too early. Noise reduction, darkening the sky background, and so on should be left for last once you have exposed the target. Do it too early and you will lose important faint details, or will simply wash out other details.

Now, that isn’t to say you can perform miracles with bad data. While the second image is better processed, I can only do so much due to the noise in the stacked image, and the fact that the galaxy’s data is buried in the noise. My method of collecting Darks for the DSLR seems to not really do much to combat the noise. So I’ll need to take longer exposures to also help get the signal above the noise. On top of that, the focal reducer used tended to bloat the stars and probably had an adverse affect on the galaxy as a result. Since this image was originally taken, I’ve moved to a different focal reducer which has helped there.

The upside is that with extra practice on existing data (one reason to always keep around raw data as long as possible!), I’ve gotten better at processing, and should be able to tackle M81/M82 again in the near future when the weather cooperates and make a second real attempt.

M42 Capture

I have been having trouble getting guiding to cooperate during the last couple nights out when the skies have been clear, and have been trying to reduce various sources of vibration and other error (that I have control over, anyways).

This weekend was actually pretty clear, and so the scope came out. I finally decided to try training the PEC on my mount for the first time. I used PHD Guiding through the LX200 to train the PEC on one of the stars in Orion. After training, I turned the scope to M42, imaging through the LX200 at f/6.3, and guiding using the Orion 80ED refractor at f/7.5. Guiding has definitely improved, and I am seeing a lot fewer ‘jumps’ in the guide star. Haven’t even bothered to try to measure the periodic error yet.

Above is a single 5-minute exposure at ISO 800 taken through the LX200R, showing the full DSLR frame. Stars still aren’t quite round, but they are worlds better than the last time I tried to take 5-minute exposures (they were previously long streaks, even when guiding with the SCT, and imaging with the refractor). Getting the periodic error down a bit has made it easier on the guiding, and I don’t see large adjustment spikes anymore.

For anyone else who is just starting out and having problems getting their exposures to last a bit longer: make sure you have trained your PEC, if your mount supports it. Even though training with an autoguider isn’t perfect, it is hands-free, and does help in a big way versus an untrained mount.

When pairing up two cameras for astrophotography, I’ve run into issues where the cameras and scopes are mismatched in such a way that autoguiding software, even with an automatic calibration tool can’t quite configure it so that you get good guiding. Instead you get all sorts of streaking introduced by the autoguider compensating in ways it doesn’t need to.

Almost all of it comes down to really just trying to make sure the light coming into the imaging camera doesn’t move a whole lot. The first step is to figure out the image scale for each camera. The formula for calculating that is:

Arcseconds/Pixel = (206.265 / (focal length in mm) ) * (pixel size in microns)

In the case of a Canon Rebel XSi that I use, the pixels are 5.2 microns in size (µm). The guide camera I am using has pixels that are 4.65 microns. I use a 10″ LX200 and an Orion 80ED. Say I am running the LX200 at f/8 and the Orion 80ED at its native focal length as a guide scope.

In this case, the XSi is running at 0.53 arcseconds/pixel, and the guide camera is running at 1.59 arcseconds/pixel. This isn’t too bad, as we aren’t capturing beyond the resolution limit of the LX200, nor the Orion 80ED, but we are really close in both cases. We do have one problem though. One pixel of movement in the guide star is equivalent to 3 pixels of movement in the imaging camera (as a ratio of one to the other).

It gets worse too, seeing will affect the size of our guide stars, and will make the centroid of the star appear to move a little bit. For us to image at this focal length well, we pretty much have to be guiding at 1/3rd of a pixel. That isn’t possible, but attempting to use settings that would adjust at these subpixel ranges will likely cause us to overcorrect too much (as a result of seeing, 1/3rd of a pixel is still within the realm of a good subpixel guide program, in perfect seeing conditions). 

So right off the bat, we know that we will likely be unable to guide at subpixel levels with this pairing very often, and will need to lower the sensitivity of the autoguiding software to get better results on average nights. And in fact this is exactly what I encountered the first time I tried to guide with this setup. The guide software kept trying to overcorrect and lead to exposures I couldn’t use. After I turned down the sensitivity, I was getting much better guiding, but still subject to some flexture (likely from the wedge and the 80 lbs of weight bolted to it). Although I will need some more nights before I can nail down exactly how much flexture I actually have.

AT66ED

LX200R

47 Megapixel

As a sort of ‘well, duh’ moment, I finally got the new scope collimated properly for the first time since I got it last night. The results were immediate and obvious. First thing I did after collimating was to point it at the moon that was just starting to wane as Jupiter had already sunk too low.

The resulting detail was pretty amazing. I went ahead and grabbed a shot from the AT66ED and the newly collimated LX200R to compare between the two. I also posted a full 47 Megapixel version of the LX200R image that lives outside the gallery for those curious about the full-resolution mosaic.

Tonight I was able to get out early enough to get a little visual observing in on Jupiter, but no chance to capture any video to see the difference it will make when trying to stack up an image. The views I got were rather impressive though. Jupiter’s bands were well defined, and I could make out the northern bands quite clearly. Also I caught a moon’s shadow on the face of Jupiter, clearly defined as a black spot, the first time I have been able to see that level of detail visually. Hopefully this will become a treat once I get the chance to point the scope at Jupiter for imaging purposes.