Sky Flat Issue

[SanjeevJ]

Hi there,

I noticed an issue with using Sharpcap for creating master flat using sky flat approach. The issue is simple -- the individual flat frames and bias frames (and even the master bias generated) are pretty good but the calibrated master flat over corrects.

I tested a couple different ways/ First I generated a master bias in SC Pro at pre-sunrise conditions. Next I collected 50 flat frames at 100 gain and 300 ms exposures for the zwo 2600 camera using SC's flat tool.

1. stacked the frames in Sharpcap's process, to generate a master flat
2. used the WBPP process in PI to stack a master flat

I compared both with a synthetically generated flat that I worked on iteratively to calibrate about 5 different stacks of star fields and faint nebulae.

Master flat generated in Sharpcap consistently over brightened the edges.

Master flat generated in PI was perfectly on point, within a couple ADU's for the edge in the raw stack.

So something in the stacking process works differently in SC Pro vs WBPP in PI. It could be the default settings especially for variance handling but I am not sure.

Windows 11 25H2 Pro
ZWO 2600 MC Pro
SC 4.14 from April 20, 2026

[admin]

Hi,

are you using SharpCap's master flat in other software? If so then that is the problem - SharpCap pre-subtracts the bias from the master flat before saving, which I don't think the other software will be expecting, so then the subsequent processing using a bias level again would be inappropriate. I'd usually recommend running the entire processing pipeline in one application (ie everything done in SharpCap, or everything done in PI or AstroAst or DSS, etc).

Another common cause of over/under subtraction is using a dark frame that is not properly matched to the light frames when using the flat correction (this applied either for flat correction inside SharpCap or outside). If there is no dark frame, SharpCap will 'try its best' using the bias level stored with the flat frame data, but that only works out if the camera bias level has not changed since flat capture and the thermal noise is practically zero.

cheers,

Robin

[SanjeevJ]

I can confirm its neither of the issues. I have tried using SC generated master flats in both SC and PI. SC sky flats are over correcting. In PI integration I treat the flat as a master calibrated flat with no correction needed. Once the master flat is generated, I am using identical process in both SC and PI for calibrating the lights.

Its not the actual frames being captured, since literally the same frames captured by SC for sky flat processed with the same bias master (also captured by SC) creates a more accurate master flat in PI at least for RASA 8 + ZWO 2600 MC Pro.

The only thing I can think of is that I have historically forced flats to be a couple seconds at home using a quality light panel and ND-95 type sheets. That has worked perfectly in SC Pro, and I have used those master flats for deep image processing in PI. Sky flats tend to be anywhere from 100 to 300 ms captures for me since I need to pick a time when no stars show up (has to be bright enough) but still dark enough to not quickly saturate at F2.

Not sure if there is something with the duration that makes the difference to how SC handles vs say WBPP in PI. I would guess not, but I cant think of any other reason why SC Pro would be over correcting only sky flats. I am doing mono.

[admin]

Hi,

do you get the same result if you use the flat in SharpCap with the 'Folder Monitor Camera'? You can re-run your lights through that camera, set up the flat and dark and see what happens with SharpCap's processing. At least in that situation I have access to all the relevant algorithms and code and can properly see what is going on.

As to general flat correction checks - my experience has been that there are five things that are the typical causes of bad flat correction :

1) Mismatch between lights and darks - different offset, gain, exposure or other camera settings
2) Mismatch between flats and dark flats/bias - same causes as above
3) Changes to the optical setup between capturing flats and lights - rotating or removing camera or significant focus change or adding/removing filter
4) Obstructions beyond the end of the telescope when using a flat panel (like a side-by-side mounting where the neighbouring scope intrudes into the light cone)
5) Excessing stretching - no matter how good the flat correction it is never 100% exact and if you stretch far enough you will see the residual over/under stretch error.

points 1 and 2 come down to the flat correction equation:

Flat corrected = (light - dark) / (flat - darkflat or bias)

Both the top and the bottom of the division must have no offset - ie. they should evaluate to zero at points with no illumination. If that is not the case then flat correction will fail.

cheers,

Robin