M45: Advanced Processing in StarTools 1.3 Part 2

Guides, tutorials, tips & tricks.
Post Reply
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

M45: Advanced Processing in StarTools 1.3 Part 2

Post by admin »

We now turn our attention to the colour information that was acquired for our image of M45. We intend to combine this colour information with the luminance information that we processed in the first part of this tutorial.
Fortunately, processing colour information is a lot easier.

We start off with creating an RGB composite by loading the R, G and B
img_029.jpeg
img_029.jpeg (45.21 KiB) Viewed 10795 times
Upon 'Keep'ing the result we have to indicate once more that the data is stretched (although, again, it clearly isn't!), as we will be pre-processing the data in the Layer and Repair modules again which are otherwise not accessible during 'tracking' mode.
We further Ignore the 'stacking artefacts' warning, but again make a mental note of it.

The colour data too suffers from the CCD blooming that we noticed while we were processing the luminance data. So, in exactly the same way, we process this data to get rid of the blooming;

Mask, Auto, Stars, set Threshold to 90%, Do.
Repair, set algorithm to 'Debloom (Vertical Streaks)'
Do, Keep.
Layer, Mask, Grow x2, Keep.
Center on one of the Seven Sisters stars.
Set 'Filter Kernel Radius' to 4.0 pixels, Mask Fuzz to 4.0, Layer Mode to 'Lighten'.
img_031.jpeg
img_031.jpeg (67.5 KiB) Viewed 10795 times
After 'Keep'ing the result, we turn our attention to the stacking artefacts. After launching the 'Crop' module, we notice that it has remembered the cropping parameters we used for the luminance frame - it makes our life just that little bit easier! More importantly - colour and luminance combine would fail if the dimensions are not exactly the same.

We're now ready for post-processing this image - we click 'Track' and confirm that ''This image is still linear and unstretched'.

We first turn to 'AutoDev' to see what we got. It appears we have a significant blue bias. We 'Keep' the stretched result for now.
img_033.jpeg
img_033.jpeg (162.6 KiB) Viewed 10795 times
To get rid of any colour bias and gradients, we launch 'Wipe'. We choose 'Fill Mask' to include all pixels for Wipe's consideration. You can also tell Wipe to ignore parts of the image or use it PixInsight DBE-style, selecting individual samples if needed. We, however, just use the default settings that Wipe opened with. We do set the 'Dark Anomaly Filter' to 5 pixels. The latter makes sure that any small dark anomalies (ex. dead pixels) don't interfere with Wipe's ability to detect the correct background level. We click 'Do' to let Wipe do its thing.
img_034.jpeg
img_034.jpeg (95.7 KiB) Viewed 10795 times
From what we can tell, the result is a big improvement. The blue bias has been removed and Wipe has helped us a great deal towards final colour calibration.

We run AutoDev again ('Redo global stretch') to better see what Wipe has done and to re-allocate dynamic range.
We increase 'Ignore Detail <' to around 1.5 to make it ignore noise. The result looks good - we don't see any areas where Wipe backed off due to the presence of non-celestial dark anomalies. Such areas would have looked like coloured halos. We are happy with this and 'Keep' the result.
img_035.jpeg
img_035.jpeg (182.95 KiB) Viewed 10795 times
Believe it or not, but this concludes our processing of the colour data - we told you it'd be short and sweet! :D
We still have final noise reduction to go though - we click 'Track' and select 'Stop tracking, do final noise reduction'.

The trick with colour data is to apply heavy noise reduction. The reason why we can get away with this is that the human eye is much less sensitive to loss of colour detail than it is to loss of luminance detail. Since all the luminance detail will come from our luminance frame (see tutorial 1), we can be heavy handed with our noise reduction. In fact, we kind of have to - astronomers typically allocate only a fraction of their time to shooting RGB data because of the aforementioned human insensitivity to colour detail. This data set too has had much fewer exposure time allocated to RGB, so it is quite noisy. So in the De-Noise module, we set all scales to 100%. We set both 'Color Detail Loss' and 'Brightness Detail Loss' to 100% and 'Keep' the result.
img_036.jpeg
img_036.jpeg (72.11 KiB) Viewed 10795 times
We save the result to a file. We now have the colour part of our data set out of the way, and we are ready for combining colour and luminance data.

Combining the two couldn't be easier.
We launch the 'LRGB' module and load the luminance file for 'Luminance' and load our colour file for R, G and B.
img_038.jpeg
img_038.jpeg (128.64 KiB) Viewed 10795 times
The combination shows some faint signs of colour. All we need to do is boost that colour and we're done!
We 'Keep' the result and indicate that 'This image has been stretched'.

To make a minor correction to our clours, we launch the Color module. When prompted we click 'Fill Mask' to let Color set all the pixels in our mask (LRGB doesn't set a mask for us by default).
From here' it's all to taste. We choose to set 'Saturation' to 450% and increase the red ratio a tiny bit to 1.05. We further set 'Cap Green' to 'To Yellow'. The latter effectively makes sure predominantly green pixels don't exist in our image, assuming such pixels are colour noise. We can do this since the colour green is very rare in astronomy, except perhaps for OIII-rich nebulas such as the Trapezium in M42.
img_044.jpeg
img_044.jpeg (126.18 KiB) Viewed 10795 times
We 'Keep' the result and are left with the following very high resolution result;
M45_ST13.jpg
M45_ST13.jpg (1.41 MiB) Viewed 10795 times
From here, all depends on personal taste and artistic vision. We, for example, would 'Bin' this image to a quarter of its resolution (50%), 'Mirror' it horizontally and vertically, apply the 'Isolate' preset in the 'Life' module with a 50% 'Strength' and a 130% 'Saturation', after which we would run the Flux module to let it perform some automated sharpening. We would then use Wipe's 'Cast' preset (setting 'Mode' to 'Correct Color and Brightness') and finally bump up saturation in the Color module to 130%;
M45_ST13_PersonalTaste.png
M45_ST13_PersonalTaste.png (4.13 MiB) Viewed 10794 times
Ivo Jager
StarTools creator and astronomy enthusiast
Post Reply