Lateral chromatic aberration

[Ken82]

Perhaps someone can help me better understand this-

So as an example a Refractor that has a focal length in red of 501mm and a focal length of 500mm in blue and 500.5 in green. So you get lateral chromatic aberration. 
 

Left is "regular" or longitudinal chromatic aberration and right one is lateral chromatic aberration.  With lateral chromatism stars will be elongated towards the edge of the field in lumiance, but each color will be "round" - although a bit displaced towards the edges.

 

So as the diagram,  blue is a shorter focal length so is pointing inwards.

But here’s what I don’t understand- after using star alignment (pixinsight, registar etc) the red and blue images have been scaled to about the same as green 500.5mm. So In theory the channels should line up ???

But they don’t, you still get blue pointing inwards. Why does this happen if they are the same scale?  Also in the centre of the image you also get the blue/purple to one edge? 
 

Thanks ken 

[vlaiv]

Longitudinal chromatic aberration is change of focal length with respect to wavelength of light.

Lateral chromatic aberration is change of focal length with respect to wavelength of light and angle of incidence.  For rays coming in parallel with optical axis / principal ray - there is no change in focal length. Further away from center you are - more is focal length decreased or increased for respective wavelengths.

You are right - if you use stacking software that is capable of scaling subs in order to align them (there are different operations that can be used to align subs - translation, rotation, scaling, ... and not all can be implemented in alignment algorithm, so scaling might be missing as usually focal length does not change), and given following conditions:

1. focal length depends linearly on angle / distance from the center

2. sensor response to uniform across wavelengths

you should see no lateral chromatic aberration as color separation and only slight elongation of stars should be visible (and if you use narrow band images - there should be no elongation).

Problem is that these two conditions are never met in reality.

If first is not met - then there won't be single "scale" that one needs to use to enlarge or shrink particular channel. Channel will be sort of warped and if you align by some stars - others will still show lateral chromatic aberration.

Second one just means that stars won't have "peaks" in the same place because camera has different sensitivity in different wavelengths. Once you align peaks of stars - "skirts" will go either way and if red skirt goes one way and blue goes the other - you will still see chromatic aberration.

[Ken82]

Thanks Vlad,

It does appear to me then that the first point isn’t being met as a single scale by the software isn’t solving this issue unfortunately. 
 

interestingly I also tried out registar,  advice from @ollypenrice which I’m now convinced does an ever so slightly better job of alignment on axis. (Still far from perfect) the purple fringe is still on the right hand side. Off axis and around the image it doesn’t have any effect. 
 

I know I’ve asked previously (sorry can’t remember who I asked) but could it be possible Sensor tilt is a factor here which is why it’s not scaling correctly? I recall I was told the colour fringing would point in one direction if so and didn’t pursue this. 
 

thanks ken 

[Ken82]

Ok I think I may have found something, I’ve had a look at FWHM and eccentricity across each image and things appear to change quite a bit- 

these two images were taken on the same night. The one on the left does appear to show slight tilt ?? But then it’s near perfect again later the same evening. 
 

 

[ollypenrice]

I'm out of my depth here but I, personally, would probably have chosen term 'radial' chromatic aberration over 'lateral.'  Hey-ho. I'll gladly be put back in my box!

For me the thing about a Registar co-registration of three channels over a regular alignment 'x-y and rotation' is that Registar resizes each channel to the reference one. If one channel has a longer FL the object will obviously be larger in that channel, meaning that, if everything is perfectly collimated, the 'long' channel will give round RGB stars in the middle with an 'x-y-rotation' alignment but the long channel stars will be centered radially outwards from the short channel stars in the corners. Registar's 'x-y, rotation and resizing' seems to me to correct this. Any tilt will throw in the complication of the 'radial' effect becoming eccentric. And then I really am out for the count...

Olly

[Ken82]

13 hours ago, ollypenrice said:

For me the thing about a Registar co-registration of three channels over a regular alignment 'x-y and rotation' is that Registar resizes each channel to the reference one. If one channel has a longer FL the object will obviously be larger in that channel, meaning that, if everything is perfectly collimated, the 'long' channel will give round RGB stars in the middle with an 'x-y-rotation' alignment but the long channel stars will be centered radially outwards from the short channel stars in the corners. Registar's 'x-y, rotation and resizing' seems to me to correct this. Any tilt will throw in the complication of the 'radial' effect becoming eccentric. And then I really am out for the count...

Olly

Yes I think registar is doing a good job of this. It does resize all channels to the chosen dimensions.

There is actually a script for realignment of RGB In the registar folder which is designed for exactly this issue. 

On inspection of the RGB focal lengths after star alignment in pixinsight I believe pix is also doing the same thing.

I can actually manually move the red and blue channels over each other in the centre but of course because this isn’t further scaling the remaining image looks awful. 
 

The problem is although the software is resizing each channel the RGB still doesn’t line up. 
 

I suspect my images require multiple scaling at different parts. I found out that the tilt I appear to have moves from corner to middle of image depending on direction the scope is pointing (as my previous post). So this clearly doesn’t help any scaling/alignment of images. I guess this could be one of three things - adapter, sensor tilt or focuser? Actually maybe excluding sensor tilt as the sensor would be fixed and not move. I’ll need to have a further look. 
 

It never rains but it pours ☹️