NGC 253 and the noise battle

[SimmoW]

Had a good imaging session last night in the 'burbs, guiding went well after my purchase of a few heavy lead fishing weights and a couple of ankle weight holders, for fitting to the rear of my Tak that has had balance issues in the past. Its very front-heavy.

At some points, the dec graph in PHD was totally flat, never seen that!

Alas I think I'm already reaching the limits of my dslr, its noise in these temps is becoming unbearable. Check out the original non-noise processed image! Second file. This is a good example of what Pixinsight can do to rescue (partly) a pretty bad image.

higher rez image here (https://www.flickr.com/photos/48232585@N00/15374543148/)

[raymo]

Hi Simon, what ISO were you using? what camera? 253 itself looks better
in the second image, but the background has only changed from large
amorphous patches to much smaller more sharply defined ones.
Neatimage will do as good a job, or better, and it's free. One of the
benefits of a larger scope when used with a DSLR, is that you can use lower ISO settings for the same imaging results.
You could also cool your DSLR, cheaper than moving up to a CCD camera.
raymo

[SimmoW]

Thanks for the ideas Raymo. I'm using my usual Canon EOS 40D, at ISO800. Might try 400 in future. I'd love to try cooling but have no idea where to start and the commercial options are so expensive I might as well buy a proper CCD!

What cooling options are you working on?

[raymo]

I've got a plastic clip on lid lunch box about 200 x 150 x 100 deep. I've cut a hole in the bottom that is a neat fit on the camera's t-ring. I've stuck a small thermometer on the inside of the lid. drilled small holes for the camera's power and intervalometer cables in each side of the box.
Mount the box with the 200mm dimension horizontal. Lay a flexible
frozen cold pack in the box and put the lid on. Experiment with how long is needed to get x amount of cooling before commencing imaging. Also
see how much of a problem condensation on the sensor might be.
Might work, might not. Being as I don't take separate darks, and leave
the camera to do it, it won't matter if the temp fluctuates.
raymo

[SimmoW]

Hmm, sounds interesting. And dangerous if it leaks!

Here's a reprocess I've just completed, I removed all the darks which at these temps and long exposures (22 by 3 mins) pretty much add to the noise, not subtract. Re-stacked and fiddled a bit more with the noise reduction, better and more natural I think.

It's here (https://www.flickr.com/photos/48232585@N00/15376189318/)

[codemonkey]

Nice one Simmo! The repro looks really great compared to the original ones, you've done a great job with that.

Looks like you're really pushing the boundaries of what you can get out of an uncooled dSLR now, in our temps. Not sure how the 40d goes noise wise either; there's likely to be newer models with better noise performance which could be picked up at relatively little cost (e.g. the Nikon D5100, although that comes with its own set of problems with software support). But then you've got it Ha modded so that's another additional cost.

A home made cooler box does sound like a good investment as you might be able to significantly improve your results with little outlay. I believe Orion also make some cooler boxes for Canon dSLR for a few hundred $ that don't require any ice (peltier driven I think) so they reduce the risk there.

[codemonkey]

Also, I like the FOV here. For some reason I actually prefer galaxies to be smaller in the frame... can't actually put my finder on why.

[Rex]

Hi Simon, great result mate! I really like the final one, it looks much more natural all round. I too like the FOV you get with your system. It almost looks like a lense shot. Great stuff!

[RickS]

The last one is looking pretty good, Simon.

Are you using a mask to protect the brighter (higher SNR) areas when you do noise reduction? For noisy data I usually do a light noise reduction with MultiscaleLinearTransform and a linear mask before stretching and then some additional noise reduction afterwards, often with TGVDenoise. For a troublesome background you can build a RangeSelection mask and then hit it hard.

Cheers,
Rick.

[SimmoW]

Thanks Rick and everyone. Yes I like my scope's FoV, yeah feels like a beefy lens sometimes!

Rick, I'm not familiar with some of the tools you mention, so I'll google, thanks for the suggestions. I mainly use ACDNR with its lightness mask, nothing too complex, so I'm always up for more suggestions.

I may reply back with some queries if Google doesn't help, with such terms as 'linear mask'.

I think my cam is pretty bad, as I just saw a friends image taken on the same night with a Canon 60da, so much better noise, significantly improved.

[RickS]

Simon, the multiscale processes in PixInsight like MLT are all capable of noise reduction and do it very well. The advantage is that they can be applied with different settings at each wavelet layer, so you can do stronger noise reduction at small scales which is where the noise is usually strongest. I find they work a lot better than ACDNR at reducing noise without smearing detail. Here's a tutorial: http://pixinsight.com/examples/M81M82/index.html#Noise_Reduction

Cheers,
Rick.

[SimmoW]

thanks heaps Rick, I'll go thru that tutorial.

Do you do the noise reduction at the linear stage or after stretching?

[RickS]

You can use MLT on linear data if you're gentle then follow up with some further nr if needed after the stretch (maybe use TGVDenoise the second time.)

Cheers,
Rick.

[SimmoW]

Thanks Rick, I'm playing with it now. TGVDenoise is very powerful. I found that with the MLT tool, I couldn't remove the small flecks of noise, only TGV could do that. I tried using a range selection mask, but it wasn't accurate enough for many of the stars, so I'd probably resort to using a star mask or as I'm doing lately, a star mask and then using the clone stamp tool to further mask the main object. I'll be fiddling plenty, thanks mate!

[raymo]

Re the cool box; it wouldn't matter if the warming cold pack generated some condensation, as the cold pack is about 25mm below the camera, and any water would drip out of the bottom of the box
through a couple of small holes I've drilled, and onto the ground.
raymo

[cometcatcher]

But doesn't the cold air leak out of the holes?

[raymo]

The holes are tiny, just large enough to let water drip through. There is not much spare space in the box, and the flexi pack is about 180 x 100
x 30mm, so should take quite a while to heat up to near ambient.
raymo