Closed Joelgranados closed 9 years ago
@chuong can you please test how small the images can get before they break your colour correction/unwarping/pot detection?
Thanks K
I have reduced them even more (12% of their original size) in my https://github.com/borevitzlab/timestreamlib/pull/153. Not sure if we can go furhter.
An idea is to get a sub-image and only work with one tray.(the one witth the color card), but that would entail more than just rescaling the values in the yaml files.
I think we have reduced the images enough
I have tested with 25% of the original sizes of the BVZ0036 timestream and it all seems to work. We may be able to get away with reducing the image sizes even more by using a higher jpeg compression factor.
We might also be able to reduce the size of the tests images even more if we generate dummy images with numpy.