How to process images that have different binning?

[Singlin]

I use Pixinsight.

My R,G,B images are 2x Binned.

My Luminance are 1x Binned.

I have a set of Bias,Darks and R,G,B falts that are all 2x binned.

I am sure I will need 1x binned versions of my Bias,Darks and Flats for my 1x Binned Luminance files.

How  do I batch process the whole lot in Pixinsight?

[Steve 1962]

Hi

You’re right - you’ll need 1x1 binned calibration frames to go with your Lum data.

Because you’ll have two sets of calibration frames, I don’t think you can process the whole lot as one batch, but you should be able to calibrate your subs in two batches -one batch for 1x1 and the other for 2x2.

Use the Batch Preprocessing script.

Once you’ve calibrated and cosmetic corrected your subs - you can use the same script to align them all and integrate them as a third batch. I’d use a 1x1 binned image as the reference frame because PI should expand the 2x2 subs to match.

Hope that helps.

Steve

[dannybgoode]

I asked a similar question here and got some useful answers also:

 

[tooth_dr]

APT can do this sort of thing for you, and align and resize to reference 1x1 frame.

[carastro]

I use Registar to re-size the binned images to match the unbinned ones.

Carole 

[ollypenrice]

In pretty well any image processing programme you can just go to Resize and type in the same pixel dimensions as those of the unbinned image.

Olly

[billyharris72]

My take on this would be that you should get some benefits from the preprocessing, but not others, once you have scaled the images to match. Assuming you choose to upsample the binned images your calibration frames won't help with fixed pattern noise at the individual pixel level - you could try downsampling the luminance and see what happens. In terms of each type of frame.

Flats - should be fine. I've seen people denoising or median blurring flats, and I don't see this causing problems (you wouldn't do it for scientific  imaging obviously). You might lose a little accuracy, but it won't matter in practice.

Darks and bias will have info on the bias and dark current level and distribution (e.g. if one side or corner has stronger unwanted signal), but might need to be scaled to your image depending on how it's been binned (if the ADU is the pixel average rather than sum then no need). I don't use Pixinsight but believe Pixel Math is what you want here if it's needed.

If you bin the light frames you'll obviously lose resolution, but might be an option for plan B - this approach should allow the calibration frames to deal with hot pixels as well.

Overall, maybe not the ideal workflow but we can only work with what we have. Experiment and see what you come up with.

Billy.