How to Handle Binned RGB?

[kerrywaz1]

I have some binned (2x2) RGB data I want to combine with unbinned luminance. What is the best way to do this in PI?

Thanks,
Kerry

[Nocturnal]

Hi,

I've never done this but I imagine it's something like this:

- open the StarAlignment process
- make the L frame the reference
- register your RGB frames
- open the LRGB combination tool
- combine the 4 frames

[pfile]

i've done this before; this sounds right.

the only nuance might be: what algorithm does StarAlignment use to upscale the 2x2 binned image? if Juan is around maybe he can comment if it makes sense to rescale the image with the Resample process to take advantage of the advanced interpolation algorithms in that process.

[Juan Conejero]

The general procedure that Sander has described should work without problems. You can however facilitate (and speed up) things if you disable the 'Use scale differences' option of StarAlignment (Star Matching section). By default SA runs optimized for images without scale differences (the most common situation). By disabling this feature you'll save an initial unsuccessful registration attempt for each binned image.

if Juan is around maybe he can comment if it makes sense to rescale the image with the Resample process to take advantage of the advanced interpolation algorithms in that process.

To upsample images, both tools —StarAlignment and Resample— will use the same algorithm: clamped bicubic spline. Another option would be using the IntegerResample tool to upsample the binned images without interpolation, with the help of an ImageContainer. This would apply a 'debinning' step to all binned images prior to registration. I don't know if this would be better than resampling by interpolation. Something to try and compare results?

[Nocturnal]

Doesn't the star alignment process resample the image in almost every case? Or does it detect that upsampling is or isn't needed and then skips that step if possible? I'm thinking that different filters may lead to very slight scale differences. In that case a single resample would be beneficial over 2.

[sreilly]

I usually image my RGB frames at 2x2 bin and luminance at 1x1. My usual process is to process the luminance image and save. Align all the RGB data and combine into a RGB image then process as usual. When I'm happy with both I'll open the finished RGB and Luminance image and use the Dynamic Alignment tool to align them. Use the luminance image as the 1st image (reference image) and the RGB as the 2nd image. After picking at least 3 point and they are corresponding to each image, drag the triangle button (new Instance) onto the RGB image. This will generate an aligned RGB image. Save the aligned image and then use the LRGB Combine tool to create the LRGB image. With the aligned RGB image and luminance image open, un-check the R, G, and B boxes. For the luminance box, left checked, click the right down button and you should see the luminance image listed. Choose that image and then drag the new instance button onto the registered RGB image.

I'd be curious to hear if this is any better, different or otherwise than the other procedures mentioned. Always good to learn.

Steve