Hi James, as stated in my previous message, "our" solutions are "low tech" approaches that try to eliminate a fixed pattern due to specific CMOS manufacture.
The flat-field approach is for planets. Proper correction needs specific flats for each filter, not for eliminating dust differences between filters but to correct for pattern response of the CMOS that varies along the spectrum. I need to apply different flats for UV, B, G, R, IR807, IR1000, CH4 18nm, and CH4 5nm filters. Flats are obtained full-frame and SharpCap applies them automatically (and auto-corrects for the FOV used) when a specific filter profile is originated and then chosen. Afterwards, image alignment to produce stacks in AutoStakkert eliminates any residual pattern. I say residual because when we calibrate a single image with a master flat, noise present in the image is always seen.
For H-alpha sun, where the needed contrast enhancement may be higher, some residual pattern may remain. In my case, I use a 178mm for H-alpha solar imaging. Since no dust or Newton rings are present in my system, and the full solar disk FOV is uniform, I even do not use flats. To eliminate the pattern described for this camera, which is really present, I do the following.
I configure AutoStakkert, so that my full-frame SERs are stacked with the 2x resample option activated. Then, after full processing of the resulting image stack (sharpen, gamma...etc) I do the following in photoshop:
(D) FINAL PROCESSING of ORIGINAL and “CONV” IMAGES (to eliminate camera pattern and restore original camera sampling)
1) File – Open – Open Image.
2) Filter – Blur – Motion Blur (angle=0 degrees; Distance=3 pixels).
3) Filter – Blur – Motion Blur (angle=90 degrees; Distance=3 pixels).
4) Image – Image Size (Width=50%; Height=50%).
5) File – Save As (TIFF; 16 Bits/Channel).
In short, first 2x increase image size in AutoStakkert, then return to original sampling in Photoshop after "motion-blur correction".
As stated, this is a low tech approach that may not fully use the camera's possibilities, namely that so important highest resolution that could be available. In fact, 2x resampling during stacking, and afterwards maintained, has been described to have its advantages. This effect is lost with the "motion-blur correction" needed to fully eliminate the pattern.
The other, and most desirable "high tech" solution, apparently is not one of ZWO's priorities.
And one has to understand that.