RGB Compositing

Last week, I received my first cooled CMOS camera (ASI533MC) and I’m hurry up trying to get a deep field on M42 before it disappear until next session. At the moment, I’ve captured 160x60" frames and counting. But Orion is now very low on the West and the atmospheric refraction is very apparent. You can see here the blue-red gradient generated by the refraction of the left-right direction on this frame.


If you split the image into RGB components and look at the RGB channels, you can see how the star’s centroid is not coincident (as expected) and apparently it “moves” from blue to red


So, my plan was to load the three channels on the Image composition tool and use the stars registration to align them and minimize the atmospheric dispersion.

But it didn’t work. I’ve selected one of this highly dispersed star and after pressing the auto align, nothing happens. Also, the manual controls are disabled, so I can’t guess any value. As you can see from the screen shot, the stars was perfectly identified (orange circle).

How should I use this feature?

No need to use the rgb tool.
Select the star. Go to the rgb Chan. Right click and you can align.

1 Like

Oh man, you always find a way to surprise me with SiriL’s capabilities. I was not ready for a “one-click” solution hahahaha.

But unfortunately it didn’t works. When applied to my image, the channels didn’t move. I’ve zoomed the image until the point that any individual pixel is a big square and you can see that the channels are aligned on the max pixel value, but it doesn’t seems that the geometric center is fine… Obviously the size of the star depends on the stretching. On the first one, I’ve used outstretch, so is maximal

With a less extreme stretch the asymmetry is more apparent

But, I know, this could be a visual effect, so I’ve let SiriL’s to compute the PSF center on each channel before and after applying the correction. And nothing changes

So, the question remains. How can I align the RGB channels? Ideally I will like to be able to use the manual corrections on the composite window (perhaps after an automatic try).

By the way, if you want to experiment by your self, I’ve attached a little crop of the frame

crop.fits (511.9 KB)

Thank in advance

There is no manual option in the RGB tool.

In fact, if the algorithm can’t align it is because the centroid center is computed as it.
You could maybe try to align with the planetary algorithm.

Sorry, but I don’t understand. As you can see from the screenshots the coordinates for the centroid in each channel are different. But no translation is applied. Why? Shouldn’t it be moved until the PSF center is in the same location?

Yes but the alignement is not subpixel. It is done at pixel level.
If you take a look to the third screenshot, all computed shifts are at 0.
Did you try planetary algorithm on the star? That could works.

Yes I tried and the result was the same: no shift between channels

OK, so your shift is under pixel resolution :slight_smile:

Yes they are around 0.5 px for each channel: blue to left and right to right using green as the reference (so red is a little over 1 px from blue).

I’ve imported a tiff into PlanetarySystemStacker an it has been able to compute and apply those shifts. But it is a shame, because PSS doesn’t support 32bits files.