Aberration with autodev

[mgutierrez]

Hi all,

I've always used manual dev cause with autodev I got so much color aberration:

Selection_148.png (63.31 KiB) Viewed 4325 times

the halo above translates into an awful color aberration (it maximizes the CA of my doublet).

However, with manual dev I get a much more clean star:

Selection_149.png (24.17 KiB) Viewed 4325 times

I tried to tune and play with many parameters, but no luck. I would like to understand why.

Regards

[admin]

Hi,

I'm not sure what you mean by "clean star"? The star's core is, as expected, much better resolved in the AutoDev version, with a neatly tapering off stellar profile visible.

Are you saying you blow out the stars intentionally as a way to hide chromatic aberration?

There are various ways to deal with chromatic aberration. But if you feel one particular channel (for example the blue channel) is particularly poor, you can just choose to not use it for luminance at all by setting its exposure time to 0 in the Compose module.

[mgutierrez]

Hi Ivo,

Thanks for your reply and clarification.

I think my problem is the misconception I have about some topics.

Let's say that this:

Selection_150.png (48.09 KiB) Viewed 4276 times

translates into:

Selection_151.png (163.7 KiB) Viewed 4276 times

and a manual dev into:

Selection_153.png (133.06 KiB) Viewed 4276 times

CA is much less pronounced

...
BUT
...

Are you saying you blow out the stars intentionally as a way to hide chromatic aberration?

I think I've been doing this unconsciously (and wrongly!) just because I was getting less aberration...

As said, I think I still don't have clear some very basic concepts. For example, color aberration. Yes, I've read that is produced due to the lens being not able to focus all the colors at the same point. But, why does this result in a bloated star and blue halo? If I have aberration mainly in blue color, I guess I should get a defocused blue channel image... well, maybe is just that bloated star? is the blue halo only a product of the defocused channel?

you can just choose to not use it for luminance at all by setting its exposure time to 0 in the Compose module

But I'm not composing an image. This is an osc shot.

This post may be quite related with a previous one I posted asking the same about CA, which you kindly answered.

Regards

[mgutierrez]

Ivo, I've been playing around with the compose module, trying to understand what's going on with my data. I'm not sure where is technically the problem. Not sure which is the channel with the most aberration. More than finding a solution, I would like to interpret my data and tools that startools provides to diagnose what I have.

This is my image (stacked with dss): https://www.dropbox.com/s/wqryzmnkxkrhu ... 1.fts?dl=0

Thanks in advance.

Miguel

[admin]

Thank you for uploading that Miguel.

It indeed appears the blue (and also a little red) channels are out of focus, relative to the green chanel; a case of classic chromatic aberration.

In the Compose module, loading the stack as red, again as green, again as blue and setting "Luminance, Color" to "L From R(2xG)B, RGB (Color from OSC/DSLR)" will use the coloring as normal from the channels. But this will also allow you to specify exactly how much of the channels should be used for the luminance (detail).

Because the red and blue channels are out of focus, we can choose to only use the signal from the green channel for our luminance (detail). This will reduce/remove the "halo" caused by the red and blue channel. The only "halo" that remains now comes from the green channel which should be a little less, as it is a little bit better in focus;

The result is that the halo is a little bit less bright, but of course it unfortunately still remains discolored;

Selection_686.jpg (37.96 KiB) Viewed 4254 times

It is not the end of the world of course. There are some things we can do to reduce the effect. Firstly, in the Color module, we can choose Artistic, Not Detail Aware as the Style. This will desaturate highlights;

Selection_687.jpg (37.49 KiB) Viewed 4254 times

Next, we can also use the Highlight Repair parameter (this "blurs" the color information around very bright stars - perfect for when that color information cannot be "trusted" as is the case here);

Selection_688.jpg (37.43 KiB) Viewed 4254 times

Finally we can bring out be "big gun" and attack the chromatic aberration using the Filter module's Fringe Killer feature. To do that, create a "Fat Stars" star mask, then Grow the mask until the discolored halos are all covered. Back in the Filter module, set Filter Mode to "Fringe Killer". Now click on the color that you wish to remove in one of the stars. Do that a few times. Every time you click, you should see the star become more and more neutral in color.

Before;

Selection_690.jpg (70.56 KiB) Viewed 4254 times

After;

Selection_689.jpg (70.2 KiB) Viewed 4254 times

Hope this helps!

[mgutierrez]

Hi Ivo,

Thanks again for your kind reply and help. Let me answer step by step.

It indeed appears the blue (and also a little red) channels are out of focus, relative to the green chanel; a case of classic chromatic aberration.

This is very interesting. How can I notice the aberration I have in a particular channel? I would like to been able to detect it, and understand its nature.

But this will also allow you to specify exactly how much of the channels should be used for the luminance (detail).

Does that mean that the color sliders only affect the way the luminance is built? does not affect the color amount itself?

Regarding the focus, I have an electronic focuser, and use ekos for imaging and focus. So, I guess, it is focusing a particular channel, but don't know which one is getting the best focus.

The result is that the halo is a little bit less bright, but of course it unfortunately still remains discolored;

to be honest, I almost can't see any difference

Firstly, in the Color module, we can choose Artistic, Not Detail Aware as the Style. This will desaturate highlights;

Yes, I also noticed that!

Next, we can also use the Highlight Repair parameter (this "blurs" the color information around very bright stars - perfect for when that color information cannot be "trusted" as is the case here);

yep, also tried this and get a bit better results. Also fringe killer filter do its job as described.

Basically, my way to deal with this effect is by using the artistic color, highlight repair and fringe killer filter. And the result are even better if I use manal dev instead of autodev, but it seems it's not the way.

Regards