Combining RGB and Dualband datasets with Siril?

[WolfieGlos]

I have previously collected RGB data with my OSC camera on a target, but I have also been collecting Ha/OIII with an lenhance filter. What’s the best way to combine these datasets?

I have stacked the RGB, stacked the dualband data separately, and then stacked them together. It works, but is this the correct way of doing this?

Since I use Siril, I did try using the “extract Ha and OIII” script, and I thought I might be able to take the calibrated files and register them with the RGB calibrated files, but since the RGB and Ha files have varying numbers of channels (obviously), Siril won’t allow this and it needs a matching number of channels. So the only way I can see of doing this is to “stack the stacks”. 

[Elp]

It depends on the look you're going for. For me I'd do it manually in PS/GIMP, the RGB layer remains as it is as a base layer, in a separate file split the lenhance data into its red and green/blue and copy those layers as separate layers onto the RGB file and only enable their blend modes to be the corresponding colours (ie the ha channel will be mapped to red, the o3 to green+blue), then adjust the opacity, colour saturation etc and the blend mode of the layer until you get a look which is satisfactory.

Or use the lenhance as the base layer and supplement areas of the image with the RGB via a layer mask.

It all depends on which data shows how much of what.

It's much better to do stuff like this manually than use automation.

[ONIKKINEN]

Siril doesn't have an easy way to mix narrowband data to RGB, i would strongly recommend doing this in another "general" image processing software like Gimp or Photoshop. It will be significantly easier to get a result you like that way.

If you absolutely must use Siril then there are a few available methods, all of which are compromises to some degree.

The easiest way is what you were thinking of already: Stacking the stacks. Simply register the narrowband RGB stacked image to the broadband stacked RGB image and stack with average and no rejection. This result will make your broadband portion of the image worse than in just the broadband stack, and your narrowband portion of the image worse than in just the narrowband stack - so you get the best of neither but at least they are combined to an image now. You may try stacking with weighting set to "number of stars", which will give the broadband stack a much higher priority because it likely has a much higher star count.

The other method involves using the Pixelmath tool to combine the extracted channels in some way which you will then recompose into an RGB image. The Ha to red part is not too difficult, you can just mix red and Ha with the max(x,y) operator for example, which will take the maximum pixel values from both and combine them to an image - which will add the Ha intensive parts of the image to the RGB red, but also add all of the noise.

The Pixel math way rarely produces exactly the image you want it to, and even then it will take a great deal of trial and error. The compromise stack the stacks method is easy, but you wont get everything out of either dataset that way. In short, i recommend Gimp/Photoshop for blending the stacks.

[WolfieGlos]

Thanks @Elp and @ONIKKINEN .

So if I layer the two datasets in GIMP, they will have different rotations. Is there a way to align the two layers, so they have the same rotation, and stars are in the correct place? Other is it pure guesswork? That was why I tried to register both together in Siril, and why I couldn’t use pixel math. 

[ONIKKINEN]

1 hour ago, WolfieGlos said:

Thanks @Elp and @ONIKKINEN .

So if I layer the two datasets in GIMP, they will have different rotations. Is there a way to align the two layers, so they have the same rotation, and stars are in the correct place? Other is it pure guesswork? That was why I tried to register both together in Siril, and why I couldn’t use pixel math. 

To register the images to each other, first drag and drop both of the stacks into the "Conversion" tab in Siril. Give the sequence a name and then press the Convert button, this creates a sequence with just those 2 files. Here you have to have them in the same format, so both in RGB with no channel extractions or anything like that.

Then in the Registration tab choose the "Global star alignment" as your registration method and hit Go register, now they are as well aligned as they can be. You can choose which of the 2 stacks is the reference frame by having that as the first frame you input into the Conversion tab. If you want to align the Duoband to RGB, then drop the RGB image first or vice versa if you want to use the framing of your Duoband image.

[Elp]

9 hours ago, WolfieGlos said:

Is there a way to align the two layers, so they have the same rotation, and stars are in the correct place? Other is it pure guesswork?

It's not really guess work more trial and error, when you see two layers on top of each other you'll notice pattern stars. Change the opacity of the top layer you're trying to align to something like 40pc so you can clearly see the bottom "reference" layer and start rotating the top one. If you've images them quite close to each other session wise the rotation will be within less than 10 degrees, once you start rotating you'll see the stars forming a trailing type pattern so you're now close. Concentrate on a star pattern near or offset from the centre of the image and do minor angular adjustments. When you're close to bang on it comes down to single pixel movements, you may have to nudge the top layer in X and Y to align.

It's quite simple and takes less than five minutes usually, quicker if youve done it a few times. It takes longer when you're combining different optical setup data together and when you have to apply scaling but not much longer.

I'm just thinking off the top of my head, if you use the RGB composition in Siril, whether you can use the offset data it outputs within the dialogue box to use as your offset when you're doing this in PS/GIMP (with PS/GIMP the data doesn't combine until you choose to flatten the layers or do a stamp visible to create a new combined layer, leaving the old layers still individual and unedited, the non destructive workflow is a much much better way to work).

Edited March 3 by Elp

[WolfieGlos]

21 hours ago, ONIKKINEN said:

To register the images to each other, first drag and drop both of the stacks into the "Conversion" tab in Siril. Give the sequence a name and then press the Convert button, this creates a sequence with just those 2 files. Here you have to have them in the same format, so both in RGB with no channel extractions or anything like that.

So essentially the method you described to me previously for multiple sessions. This is what I had been trying, but I had extracted the Ha so it was a single channel and so it didn’t work. But yes, I guess I could combine all 3 channels from RGB and dualband. Thanks Oni.

13 hours ago, Elp said:

It's not really guess work more trial and error, when you see two layers on top of each other you'll notice pattern stars. Change the opacity of the top layer you're trying to align to something like 40pc so you can clearly see the bottom "reference" layer and start rotating the top one. If you've images them quite close to each other session wise the rotation will be within less than 10 degrees, once you start rotating you'll see the stars forming a trailing type pattern so you're now close.

That’s what I was afraid of, I did this for my 6 panel of the heart and soul, and I’d agree it was more trial and error than guesswork. I guess I was hoping there was a command that would do it automatically. Thanks Elp.