Is it done by channel? That is, all R pixels are normalized based on data only in R pixels, and so on.
Or is each pixel normalized using the mean of all pixels, regardless of channel?
Is it done by channel? That is, all R pixels are normalized based on data only in R pixels, and so on.
Or is each pixel normalized using the mean of all pixels, regardless of channel?
That depends the option you use.
If equalize CFA is on, then the mean intensity of RGB channels is equalized in the flat CFA image.
So just to confirm,
So 2 questions:
Any benefit or downside to using Equalise CFA? Since its an option I suppose there are situations where it should not be used?
Looks like it should be used while making the master flat as well as the calibration. Is that correct?
Thanks - thats clear.
But I am still not clear on the actual algorithm.
Say I have a calibrated flat which has a mean of 30,000 ADU. Channel wise, I find R is 10,000, and B and G are 40,000 each.
Now when calibrating, does Siril normalize a single R pixel using the mean of 10,000 (for R channel), or 30,000 (for the flat as a whole)? (I am using mean for convenience of explanation).
Relevance: if it use the mean for the flat as a whole, in a situation where any channel is low signal, it could introduce lot of noise. In which case when making flats it is better to ensure all 3 channels are reasonably close in signal.
It computes coefficient in the center of the flat (to avoid vignetting effect) to get R, G and B at same level, with the use of the mean yes.
Flat must always have enough signal. This is why we recommend to shoot flat at low ISO.
And in this case, noise is not a problem at all.
Im using an astro camera OSC.
Issue is that with different filters and different light source, the mean of each channel varies a lot.
Anyway, will now try to make sure that all 3 are in roughly the same range, by varying the colour of the light source.
This algorithm is dedicated for flats with channels not in the same range.
This is also wanted for photometry as it does not alter wb.
Thanks