I have two cameras that have different pixel sizes, 2.9 and 1.49 microns. Is there a way to rescale the output of one to match the other? I don't use PI so please don't give me that solution. I do use Siril and Seti astro which both have astrometric based mosaic ability. I suspect the answer is within that function. Basically I'd like to be able to accurately rescale one image to another without combining them. I could do a simple resize but I'm not sure that would be accurate enough. Any ideas?
Combining image scales
Well Written Engaging
I thought it might be done in registration. I found this in the manual:Astrometric registration
Introduced in version 1.3, this is the prefered mode for assembling mosaics or images with little overlap. It can also be useful to register stacks issued from different set-ups (different optics, different cameras, different field of views etc...).
It does not have an entry in the registration method drop-down as the information to export the registered images has already been computed when platesolving the sequence. All you need to do is to Apply Existing Registation.
Undistortion will be applied as defined when platesolving the sequence, meaning if the images were plate-solved using a SIP order larger than 1, then undistortion will automatically be included. Unless you have a perfectly optically flat field, it is usualy a good idea to platesolve using SIP, as shsown below, with and without distortion correction.
Introduced in version 1.3, this is the prefered mode for assembling mosaics or images with little overlap. It can also be useful to register stacks issued from different set-ups (different optics, different cameras, different field of views etc...).
It does not have an entry in the registration method drop-down as the information to export the registered images has already been computed when platesolving the sequence. All you need to do is to Apply Existing Registation.
Undistortion will be applied as defined when platesolving the sequence, meaning if the images were plate-solved using a SIP order larger than 1, then undistortion will automatically be included. Unless you have a perfectly optically flat field, it is usualy a good idea to platesolve using SIP, as shsown below, with and without distortion correction.
Helpful
Registration via the homographic transformation works just fine if the images aren't too dissimilar in optical distortions.
Well Written
Based on that I think I have the bare bones of a script:
This assumes that I have the two images I want to match scale in a folder named input_images.
#convert and organize images
cd input_images
convert light -out=../sequence
cd ../sequence
#plate solve the images
seqplatesolve light_ -nocache -force -near -distro=ps_distortion
#choose reference frame
load light_1 #change this depending on which image you want to match
save cam2_reference #defines this image as the reference frame
#WCS registration
seqapplyreg light_ -framing=reference -ref=cam2_reference.fit -interp=lanczos4
#save rescaled image
load r_light_1 #load registered and scaled image, change image number as needed
save rescaled_1
This assumes that I have the two images I want to match scale in a folder named input_images.
#convert and organize images
cd input_images
convert light -out=../sequence
cd ../sequence
#plate solve the images
seqplatesolve light_ -nocache -force -near -distro=ps_distortion
#choose reference frame
load light_1 #change this depending on which image you want to match
save cam2_reference #defines this image as the reference frame
#WCS registration
seqapplyreg light_ -framing=reference -ref=cam2_reference.fit -interp=lanczos4
#save rescaled image
load r_light_1 #load registered and scaled image, change image number as needed
save rescaled_1