The next DSO that I worked on with Pixinsight was the Heart nebula. Previously I had only imaged this in H-alpha and O-III, but there was one session when I also managed to get some S-II subs as well. Hence it made sense to reprocess this with Pixinsight to see if I could extract even more out of the data.
As with my narrowband images, I captured these with the ZWO ASI178MM-cooled with gain setting of 200, 3 minute exposures with the Astronomik 12nm H-alpha (22 subs), O-III (20 subs) and S-II (26 subs) filters. With the Samyang 135mm F2.0 wide open, this grabs enough light with an image scale that fills up most of the sensor.
First is the Heart nebula previously processed with StarTools using just H-alpha and O-III subs from previously.
Now the images processed with Pixinsight, albeit with S-II subs included. I also used DrizzleIntegration to quadruple the number of pixels as these images are significantly undersampled. These images were processed in this order: Hubble palette, H-alpha/O-III/S-II, modified Hubble palette. Hence with each image processed I have learnt from it and refined the process further, as it is a learning process after all.
For the processing, the first step was Dynamic crop of the image which is then copied and applied to the other two images. This is then followed by DynamicBackgroundExtraction (DBE) for each channel before combining them using PixelMaths. For the first 2 images, I did not perform DBE again on the combined image, but did so for the last image and I definitely feel this does make a very big difference. Hence in future, I will also be using the DBE tool for the combined images as well.
The remainder of the processing (which I copied and modified from the Pixinsight forum for their NGC2359 thread):
- Background Neutralization,
- Color Calibration,
- Multiscale Linear Transformation for noise reduction,
- Deconvolution with Dynamic PSF/Luminance-Mask/Deringing-Mask,
- Masked Stretch instead of Histogram Transformation,
- TGV Denoise for noise reduction,
- Curves Transformation to boost colour and saturation with Luminance-mask to protect background
- Curves Transformation to reduce colour and saturation in stars wit Star Mask protecting background and galaxy,
- SCNR to extract green from image.
- HDR Multiscale Transformation with StarMask to protect the stars,
- Local Histogram Equalization also with StarMask to protect the stars,
- Multiscale Median Transformation for noise reduction and sharpening,
- Morphological Transformation to reduce star size,
- Unsharp Mask to increase details in nebulae with Star Mask to protect stars.
- Final application of TGV Denoise.
So these are the final images.
The modified Hubble palette is something recommended on the Pixinsight forum, but I do need to experiment to see if I can find a combination which I like more.
Pixinsight is a great tool for processing of astro images, but is also very time consuming – I spent at least 8-12 hours just processing these three images. But hopefully what I have learnt (and saved into the process icons for future use) will make things work much quicker – and only limited by computer processing power. But I also learnt that the correct mask is crucial for the processing, and sometimes it is worth experimenting with this in order to try and get things perfect.