Closed Ciaran1981 closed 5 years ago
Yes, this is a little easier with the new altum-support branch. warp_matrices should be all that is required now.
I'll post an example later this week days of using one alignment to stack and export a whole ImageSet to Tiff files and copy the metadata so that those files can be processed in Pix4D (or anything that can process TIFFs).
I've added code to make this easier and added an example of this. When run the Alignment notebook all the way through, the warp_matrices will be printed at the last step. These can then be set and used in the Batch Processing notebook to apply that warp across a whole imageset. To see how this works or apply this idea to other work, follow the code through the process of saving a reflectance stack in the capture.save_capture_as_reflectance_stack(...)
method.
It's important that the same warp method is used in both cases -- cv2.MOTION_AFFINE used to be the default, but now it's cv2.MOTION_HOMOGRAPHY, which we recommend using instead. If you see matix is not square
or matrix sizes must match
errors it's probably a disconnect between these.
More of a question than an issue.
Which variable is required to apply a 'good' alignment transform to another capture and indeed the remaining dataset?
Is it simply a case of using 'warp_matrices' from a good aligment to the next capture?