Loading options

[PiMaV]

We should discuss if and then how we would implement loading features like timeline copping and masking BEFORE the actual loading.

Background: We enable subsetting, and rsizing at the moment, which works fine. Using that, we can create masks and get a feeling for the intersting time window. Now i want to use this mask (and the intersting time window) to load my data again, but with higher resolution or bigger subset, which is not possible at the moment. We have the "keep in RAM" option, but thats not the same. With that, we load everything first and then free up RAM by reducing again. What we actually want is not to blow up RAM in the first place during loading; same is true for masking.

Lets discuss this as a future feature

[irkri]

• If we have a number of image files, we can load them one-by-one, and mask every file before loading the next. This way, only one file at a time is completely in RAM.
• So, somehow there is a feature for loading npy files partially. I don't know, how exactly this works, but it might be worth testing. stack overflow thread numpy memmap docs
• With videos, I am not sure how to do this. Probably we would need to load everything first before masking, but I don't know.

[PiMaV]

Thats not exactly what i mean. We should think about an option to "save" the loading parameters, the mask and the time line crop. Then we would use this parameterfile when loading a bigger dataset. Each image manipulation is done for every file before being added to the matrix.

This should be discussed in person not here