Open marpet opened 3 years ago
Feel free to improve, but I think we can only deal with less resamplings if we use another algorithm for cloud shadow detection.
This could be an option
Feel free to improve, but I think we can only deal with less resamplings if we use another algorithm for cloud shadow detection.
Or a bit more precise, we only need the correct resampling for the geometries from the "S2 resampling OP". If this can be achieved in any other way, the "S2 resampling OP" can be dropped and it will a) faster and b) easier to decrease resampling steps.
When using S2 data it is resampled multiple times. It should be investigated if this can be avoided and handled in a better way.
Also if an already resampled product is provided the internal S2 resampled product says that the input is not compatible. This is due to an issue in the S2TBX (https://senbox.atlassian.net/browse/SIITBX-433)