Combining Frames of Different Focal Length (Plz help) [Deep Sky] Processing techniques · Alien_Enthusiast · ... · 9 · 500 · 5

I don't know about DSS, but something like this works like a breeze in Pixinsight or Siril. 

I describe the Siril way, because its free to use and you could try it out for yourself. 

You stack every session separately with the "OSC_preprocessing" script and then look in the "process" folder of your working directory for the "pp_light" files. Those are the preprocessed lights of your session you currently looking into. You can take all the pp_lights from your sessions, copy them into a separate folder named "lights" and choose this as your new working directory in siril. Now you can use the "OSC_preprocessing_withoutDBF" (without Darks, Flats, Bias; because those lights are already calibrated) to stack them into one big master file. Siril does not care about Sampling or FOV, it just stacks, but later on you will see the high SNR region (the center) vs the lower SNR outer regions and the clear SNR cuts of the different FOVs. I always cropped to the FOV that matched every session. 

Regarding drizzling, as far as I can tell Siril has no proper Drizzling implementation, so you would have to use your Seestar sessions at native sampling rate. Maybe someone else has more Siril experience to share. 

Hope this can bring you closer to your goal, good luck!

Regards

Tobias:
Regarding drizzling, as far as I can tell Siril has no proper Drizzling implementation, so you would have to use your Seestar sessions at native sampling rate. Maybe someone else has more Siril experience to share.

Of course it can:=commandnameregister [url=reference external=https://siril.readthedocs.io/en/latest/genindex.html][/url]=highlight-text notranslate=highlight

register sequencename [-2pass] [-noout] [-drizzle] [-prefix=] [-minpairs=] [-transf=] [-layer=] [-maxstars=] [-nostarlist] [-interp=] [-noclamp] [-selected]

=lineFinds and optionally performs geometric transforms on images of the sequence given in argument so that they may be superimposed on the reference image. Using stars for registration, this algorithm only works with deep sky images. Star detection options can be changed using SETFINDSTAR or the Dynamic PSF dialog. The detection is done on the green layer for colour images, unless specified by the -layer= option with an argument ranging from 0 to 2 for red to blue.=line =lineThe -2pass and -noout options will only compute the transforms but not generate the transformed images, -2pass adds a preliminary pass to the algorithm to find a good reference image before computing the transforms, based on image quality and framing. To generate transformed images after this pass, use SEQAPPLYREG. -nostarlist disables saving the star lists to disk.=line =lineThe option -transf= specifies the use of either shift, similarity, affine or homography (default) transformations respectively.=lineThe option -drizzle activates the sub-pixel stacking by up-scaling by 2 the generated images.=lineThe option -minpairs= will specify the minimum number of star pairs a frame must have with the reference frame, otherwise the frame will be dropped and excluded from the sequence.=lineThe option -maxstars= will specify the maximum number of star to find within each frame (must be between 100 and 2000). With more stars, a more accurate registration can be computed, but will take more time to run.=line =lineThe pixel interpolation method can be specified with the -interp= argument followed by one of the methods in the list no[ne], ne[arest], cu[bic], la[nczos4], li[near], ar[ea]}. If none is passed, the transformation is forced to shift and a pixel-wise shift is applied to each image without any interpolation.=lineClamping of the bicubic and lanczos4 interpolation methods is the default, to avoid artefacts, but can be disabled with the -noclamp argument.

I would recommend you try to stack everything to one image instead of stacking multiple stacks and then stacking them, because you get better signal to noise ratio with 1 stack than with multiple stacks (maybe not the Seestar data - more later).

Batch 1 and 2 are quite similar in field of view and Siril will stack them to 1 image easily, you just need to manually stack the frames instead of using the script. You can still grab the pp_light files (they are your calibrated frames) after running the script and use those for registration and stacking manually. You need to choose which reference frame you want to have, and with it your pixel scale. The reference frame and the pixel scale applied to every image is chosen by the first frame you input into the sequence in the "Conversion" tab. You should choose the widest field of view for this purpose (the 150mm data).

Siril will also stack the Seestar data to batch 1 and 2 without any complaints should you choose to do so, but i am not so sure its a good idea. You can try, but you will have some visible artifacts around the area where the Seestar data ends and the next dataset starts (such as background issues, different SNR - tricky to process out to an image that looks like it was one scope from the get go which i am assuming is the goal, granted this issue also exists with batch 1 and 2 because of the difference in field of view, but its not so drastic).

As for Siril Drizzle which was mentioned before, it is not "true" drizzling like with DSS/APP/PI, it is a simple 200% upscale applied to the subs at registration. You dont need to do this, nor is there probably any use for it for what you wanted to do here (or in general - few cases where one would want to do that anyway). You can also manually stack at any scale you want to with some trickery, here is how: Choose your registration frame and resample it to whatever you want, lets say 133% and then use that as the first frame in the Conversion tab. Then run Global star alignment in the Registration tab and every sub that comes after that will be resampled to fit that so you can in fact choose any pixel scale you want to this way and dont have to go with any of the 3 options you have.

Oskari Nikkinen:
As for Siril Drizzle which was mentioned before, it is not "true" drizzling like with DSS/APP/PI, it is a simple 200% upscale applied to the subs at registration. You dont need to do this, nor is there probably any use for it for what you wanted to do here (or in general - few cases where one would want to do that anyway)

That implemented in Siril (from Iris and that from IRAF) is actual drizzle. Which is only meaningful if the original images are undersampled, which isn't the case here.

Oskari Nikkinen:
And i am being pedantic here, but i think Drizzle is not really useful even for undersampled images, at least if we are slightly undersampled (the 2 pixel fwhm that Siril recommends to drizzle with is fine, no need to drizzle). Drizzle was created for the Hubble space telescope, so this leads us to the question of if we are operating HST or not? If no, then dont drizzle, easy guideline.

It's the same operation that was used in Iris under the same name so "simplified drizzling" is the proper name (since does not reject, weigh or normalize results) but I expect the same affine remapping between under-sampled and final image with a drop of 0.5 is operating here. Need to look at the code though.

And, yes, you are being pedantic here as several forms of actual drizzling, more or less refined with respect to the original code developed for the Hubble's WFC, are in use in astronomical image processing codes and operating *on* the HST images (as opposite operating *the* HST, unless you are NASA) is no prerequisite for it to be used and useful as long as the two main requirements, i.e., 1) enough dithered frames to cover the whole of sub-sampled image space and 2) FWHM < 2.5 are met.

I suspect most of the people that uses drizzle do not operate under both conditions but that it's another story.