One of the most time consuming step is the registration of the reference images with the science image.
For example, you use PS1 as reference images, and you have a science image that overlaps 20 PS1 images. You chos to perform a subtraction on each of the 20 PS1 frames.
So you need to cut your image in 20 sub-images to match the PS1 frames. Imagine you also have a mask frame so you have 3 images to register per frame.
At the moment, what is done:
- you loop over the 20 frames
- for each frame you take the header of the reference image as the reference frame for the registered images
- run SWarp on each of the 3 images (science sub-image, PS1 image, mask) using the reference header
So Swarp is run 60 times in this example.
For sure there is a way to at least register the 3 images (science sub-image, PS1 image and mask) in one step with the correct Swarp configuration. But I haven't found it yet.
May be there is a configuration allowing to do run SWarp only once, if we can provide multi reference headers or something like that.
This will save quite a lot of time for large CCD, i.e. 9kx9k
This step is performed in gmadet/registration.py
One of the most time consuming step is the registration of the reference images with the science image.
For example, you use PS1 as reference images, and you have a science image that overlaps 20 PS1 images. You chos to perform a subtraction on each of the 20 PS1 frames.
So you need to cut your image in 20 sub-images to match the PS1 frames. Imagine you also have a mask frame so you have 3 images to register per frame.
At the moment, what is done:
So Swarp is run 60 times in this example.
For sure there is a way to at least register the 3 images (science sub-image, PS1 image and mask) in one step with the correct Swarp configuration. But I haven't found it yet.
May be there is a configuration allowing to do run SWarp only once, if we can provide multi reference headers or something like that.
This will save quite a lot of time for large CCD, i.e. 9kx9k
This step is performed in gmadet/registration.py