rounding of floating point to integers

[chrisvdberge]

I'm wondering how PI is converting floating point values to integers. If I understood correctly, when we do an Integration with Combination method 'Average' we are working with floating point. But at the end of the stacking process we need to convert to integers for the RGB values right?
So theoretically we loose signal here.

Is that statement correct? And if so, how is the conversion done, ie; might we also be gaining signal since 2.5 will be rounded up as 3?
 

[chrisvdberge]

just to clarify my question;
I'm looking into the differences between stacking a subset versus stacking all subs at once.
To me it seems you lose signal and the noise wouldn't be as 'nicely' distributed as with more frames, but just trying to confirm some ideas on that here..

[NGC7789]

But at the end of the stacking process we need to convert to integers for the RGB values right?

I don't think this is correct. I think you stay 32 bit until you save or explicitly covert to another bit depth.