|
As mentioned above, ideally your flat frame should be processable (and processed with) in raw form, before de-Bayering has taken place. This preserves as much raw data as possible, as the de-Bayering process averages out adjacent pixels to artificially create the colour data. Maximum resolution and colour saturation is preserved by ensuring that any manipulation to the data such as darks, bias, flats, etc, is done prior to interpolating the colour data.
On the question of colour tint of a flat frame, given the average DSLR uses an RGGB colour matrix, you want to make certain that your green component is not being ignored as it is 50% of your data. If the flat has to be skewed towards one part of the spectrum, I'd suggest green over red.
As Alex mentioned, you want the histogram peaks to be somewhere around 50%, with no clipping at either end. I would suggest that the histogram of the flat presented is likely clipped in either the green or blue data, or both. Being clipped or skewed significantly towards the end of the histogram will result in artefacts such as increased noise in those colour channels due to a lack of dynamic range to work with. Yes, a flat frame does add noise to your image, so ideally you shoot a bunch and median combine them to minimise the impact.
Regards,
Eric
|