Closed ddenberg closed 11 months ago
I believe I fixed the issue. I think the block size was too large.
I changed the block size to 64, 64, 16, but I am unsure how this will affect performance. Perhaps someone could recommend a reasonable block size? I am unfamiliar with the data format used to save the labels.
To enable parallel reading/writing, as well as a more responsive UI, the entire label dataset has dimensions as specified, but is chunked based on the specified block size. So in this case, as you initially had, you lose a lot of those benefits if you have the block size the same as the dimensions of the label dataset, since it means in order to read/write, you have to access the entire dataset, as opposed to a specific block. In your case, I think 64x64x16 is reasonable.
Hi! I'm interested in using Paintera to manually label a dataset I'm working with, however I'm running into an issue with creating a label source. I am using windows 10, and I installed Paintera via conda.
I've successfully loaded my raw data (hdf5 format). Because I'm not working from an n5 data set, I set the following in Sources > New > Label Source:
I get the following pop up error after selecting "Create":
Also in the console the following warnings appear:
Also, here is the stack trace from the pop-up error if that helps: