Tuesday, 28 December 2021

The Value of Revisiting Old Data

Why do it?

The longer you are involved in a hobby such as astronomy and more particularly astrophotography, the more you find two things.  
  1. You start to amass large libraries of data and images which can quickly take up all your storage if you don't keep on top of it.  
  2. You develop your own techniques, and learn new techniques for both capturing the data and also processing the data.
At one point, I had got to the stage where I had accumulated around 3TB of data, a vast amount of which I was to never use again.  There is one argument to keep all this raw data for any number of reasons.  Maybe just in case you lose your finished images and you need to start over.  Perhaps you want to scour the individual sub frames and looking for moving anomalies which could turn out to be asteroids etc.  I did consider both these options, but realistically I couldn't justify using all that storage space for no real useful reason.  So now, once I have finished stacking and processing an image, I keep the stacked unprocessed image, the final processed image and maybe the master calibrated frames if I think they can be used in the production of other images using the same exposure settings etc.  This cuts down the storage space a I need for each image from 2 or 3 GB per image to maybe just a few hundred MB. 

Reprocessing old data.

Recently, I have put some time into learning other people's processing techniques in PixInsight with a view to adopting some of their methods into my own processing regime.  But the problem with this is that it is very hard to compare how it affects the final image.  But, because I have saved the unprocessed stacked version of each of my images, I can go back and open a copy of an unprocessed image from previous data sets and then have a second or third or fourth go of processing that same set of data ensuring I start from exactly the same starting point.  In that way, I can get a good comparison.
Back in February 2021, I took an image of NGC 2244 which actually turned out to be one of my best images of the year (in my opinion).  At the first attempt of processing it, I was trying out some new processing techniques involving shifting the colour pallet to produce a false colour image.  I decided to return to the unprocessed image and re-process it a couple of further times but using different techniques.

The original image of NGC 2244 as produced by me in February 2021 using my own workflow with slight adjustments on colour to produce a lightly different colour pallet.

Processing the Amy Astro way.

I'm a recent subscriber to the Amy Astro YouTube channel.  It's definitely worth looking out.  What caught my eye with this channel is that we use very similar equipment and both use PixInsight for processing.  In particular, there are two videos Amy has produced on how she processes her basic OSC images.  The first video looks at the calibration and stacking part of the process.  The second video looks more at how she processes the stacked data and arrives at the final image.  Many of the steps are similar to what I include in my own workflow, but I was intrigued particularly with Amy's noise reduction techniques.  I decided to follow her tutorial fully and see what I could produce.  So, leaving my normal workflow to one side, I followed her workflow in full as set out in her tutorial video.  The results were impressive and while the colours and brightness of the target didn't come out the same (not that I would expect them to using different techniques), the noise within the target and background was substantially reduced when compared with the original.  

This image was produced following the workflow presented in Amy Astro's tutorial on her YouTube channel.  While the colour is slightly different (which is to be expected), when viewed at high resolution, the amount of noise within this version of the image is substantially reduced when compared to my first effort.

Looking at this result, I can say that I'm definitely going to work on bringing the noise reduction techniques both in the linear and none-linear phase of Amy's workflow into my own workflow.

One More Thing...

While reading up on alternative workflows, I also come across the blog of Simon Todd, another astrophotographer who uses a multi-band filter and One Shot Colour imaging camera.  The subject of this blog was how to achieve a pseudo Hubble Pallet colour image (which is usually reserved for images taken with Mono cameras) with an image produced using a colour camera.  In this scenario, we are using slightly different filters.  Whereas I use a quad band filter from Altair, Simon's blog states that his data used in the blog post was acquired using a dual band filter but the process can also be used when dealing with an image taken with a tri band filter.  However, I thought the principle might be the same, so I decided to once again turn to my base image of NGC 2244 and give it a go.
In this image, I used my normal workflow that I have used for a while, but omitted the steps I usually take at the start of the processing steps in which I split and re-assign the colour channels.  In Simon's blog, he states that his technique is best done towards the end of the processing of the image when it is in it's none-linear phase and otherwise fully processed.

A false SHO (Hubble Pallet) image showing the type of effect Simon's process has on one of my data sets.  

While this version of the image is very different in colour to my other images, I still think the outcome is quite attractive.  You could argue that this image to trying to be something it's not, or can ever be due to the fact it is trying to replicate results achieved by using a different set of kit.  However, in my opinion, if I like the outcome, the image is good enough for me.  I really like the idea of taking an image one step further.  It won't work for all targets obviously, but as I am learning, there will never be one set and clearly defined workflow which will suit all targets.  The image above is a very rough and ready first attempt at the process shown in Simon's blog post.  I think there is more to be done in the early stages of the processing which would improve the outcome, but for a first test of the concept for me I'm quite pleased.

Next steps.

Having learned a new set of techniques, I'm going to look and changing my basic workflow to adopt some new steps.
  • Swap out the linear and none-linear noise reduction steps in favour of those from Amy's tutorial.  I especially like her use of mask creation to isolate either the target or the background in different stages.
  • Add some optional steps to my work flow which I can use specifically for nebulae to produce a second version of an image, in SHO.

As usual, thanks for reading.  Fingers crossed for some quality clear sky again soon.  
Cheers!

No comments:

Post a Comment