StarTools

I have searched the forum looking for some more detail regarding not having DSS align the RGB channels as part of it's stacking efforts. I have always thought that having DSS do this for me would be a little better than doing this manually with the Startools Lens module. I always use DCRAW to prepare tiffs for DSS and I don't believe I see any normalization of the data as a result of the RGB aligning. If you could explain the reason for deselecting "Align RGB Channels" in DSS I would be very interested in your answer.

Thanks for creating such a wonderful product...

Peter

Hi Peter,

Align RGB Channels may indeed be left checked/on (do let me know if there is information on the website or in the forums that say otherwise - I may have made a mistake somewhere! ).
It's the only "Per Channel Color Calibration" and "RGB Channels Calibration" that needs to be set to off.

The reason for this, is that they linearly stretch the data in the different channels. This causes noise levels to also linearly scale in those channels, so you end up with different noise levels in the different channels. It also has consequences for color saturation (as is also stated on the DSS website).

Hope this helps!

Ivo,
Thanks for clarifying. I have poured thru the forum and the recommendation to NOT align RGB channels when using DSS is widespread. The most obvious is this topic

http://forum.startools.org/viewtopic.ph ... +RGB#p2914


I have written my own programs to calibrate my images using DCRAW and I have seen how dramatically DSS will alter the RGB values when using it's background calibration methods. I fully understand your concerns about how the DSS background calibration alters the noise profile of the RGB data.

Peter

Thanks for bringing this to our collective attention! I made an edit in the thread you linked.

I guess this setting warrants some further explanation. As far as I understand the inner workings of DSS, this setting aligns 2 out of the 3 channels to the other (1) channel. It stands to reason that this would usually be the green channel, as it tend to have the best fidelity and focus when dealing with instruments (e.g. DSLR/OSCs) with a color filter array on top of the sensor.

I'm not 100% sure why this would be needed, however, as you wouldn't really expect the image to shift between channels, except in the case of the more uncommon forms of chromatic aberration.
I guess my advice would be to experiment with this setting, if your dataset is suffering from color fringing around stars, where different colors are prevalent on different sides of the star.

Indeed, the Lens module addresses these sorts of shifts, but I'd be interested to learn what could cause such shifts other than chromatic aberration. I have observed similar color fringing caused by incorrect or sub-optimal implementations of debayering algorithms, but I would expect DSS/DCRAW has this sorted out!