StarTools

Hi,

I'm not sure which stacking mode to select in DSS.

According to this thread, it should be Median: https://www.startools.org/links--tutori ... r-settings

But according to this tutorial I've been going through, it should be Kappa-Sigma Clipping for Lights, and Median Kappa-Sigma Clipping for flats and biases:
https://astro.ecuadors.net/processing-a ... ky_Stacker

And then according to this thread, it should be Kappa-Sigma Clipping for Lights, and then 'Median Combination' for flats and biases which I'm assuming means Median Kappa-Sigma Clipping: viewtopic.php?f=10&t=720

So, which is right?

That's all!

Thanks, Brendan

BrendanC wrote:Hi,

I'm not sure which stacking mode to select in DSS.

According to this thread, it should be Median: https://www.startools.org/links--tutori ... r-settings

But according to this tutorial I've been going through, it should be Kappa-Sigma Clipping for Lights, and Median Kappa-Sigma Clipping for flats and biases:
https://astro.ecuadors.net/processing-a ... ky_Stacker

And then according to this thread, it should be Kappa-Sigma Clipping for Lights, and then 'Median Combination' for flats and biases which I'm assuming means Median Kappa-Sigma Clipping: viewtopic.php?f=10&t=720

So, which is right?

That's all!

Thanks, Brendan

Hi Brendan,

Choosing a suitable outlier rejection algorithm depends on how many frames you have (as well as which methods are on offer). Without delving into how they work precisely, a decent rule of thumb is that using Median outlier rejection is most useful if you have ~ 20 frames or less, whereas Kappa Sigma rejection comes into its own with more than 20 frames to work with.

In general, have a look at this section on the website about DSS. You will also want to peruse the other information about starting with a good dataset.

Hope this helps!

Thanks Ivo. I always have (many!) more than 20 frames so that's what I'll go with from now on.

Presumably it doesn't matter too much that I've encoded my entire calibration files library with median?

BrendanC wrote:Presumably it doesn't matter too much that I've encoded my entire calibration files library with median?

It's probably not at problem, though Median stacking preserves the bit-depth of the source material and does not increase its precision (it just selects the middle sample and does not perform any averaging, except when the amount of samples is odd). Depending on the depth and fidelity of your signal this may become a problem.