Flats etc revisitted

2025/11/28. I've collected 700+ images of M31 over the last few nights and am finally determined to systematize how I'm handling them. Here's the current workflow:

Debayer. Start with the CR3 files and debayer as usual in PI (Auto/VNG/combined RGB output color).

Calibrate. Today's darks are just ten or so averaged, no normalization, no exclusion. The flat is an adventure. It's a blue sky flat made through the doube acrylic defuser, debayered, averaged, format converted back to RGB, and desaturated using the PI "curves transform". Then it's tweaked against a subset of live data to produce a decent result with neutral -- not too bright, not too dark -- corners. In this case, I applied a convolution filter twice (PSF around 600) to remove dust shadows (which are inconsistent between sessions) preserving only the light transmission of the 400mm. Then I adjusted the curves (a linear transfer, raising the black point to ~30 and lowering the white point to ~225) to prevent over- and under-correction.

Align. Remember to turn drizzle data off, otherwise just take defaults and go.

Integrate. Winsorized Sigma Clipping, Scale + zero offset, no signal and noise evaluation. -4 Sigma, +3 Sigma, Winsorization cutoff 5.000 (whatever that one is).

I did these in batches of 80-130 frames each to be stacked subsequently following additional adjustment / calibration.

And then things got interesting... the middle 30% of so of the data had been acquired with the camera rotated about 40 degrees. Eventually, I rotated the minority exposures to approximate the majority and aligned and stacked them as usual. I still had to crop pretty severely and do some local cosmetic adjustments, but it all worked. I really need to put a scale on the lens so I can set, note, return to specific camera angles.

703x30s, 400mm F2.8
Canon R6, ISO 1600 (mostly, a few 3200)

:: top ::