-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pydantic Rework #47
Pydantic Rework #47
Conversation
…e into thin-wrapper
…attice into thin-wrapper
I'm trying the plugin and I get this error when I try to start the plugin in napari:
|
|
deconvolution update: But, now I get a different error:
Looks like we need to convert psf files into numpy arrays.. |
The workflow rework is now in 91c7ea9. I'll document it more thoroughly in the docs PR, but the gist is that we now enforce that the workflow has exactly one parameter called
I couldn't reproduce this, at least after my changes. Did you use the preview or save button? |
This is hard to reproduce because I don't have access to anything cuda compatible myself. Were you running the crop workflow? Because I think I can identify a point where I neglected to convert back to numpy there. |
|
I've figured out the error and I've posted it here: tlambert03/pycudadecon#64 To get around a few dependency errors , I've used following installation commands in python 3.10
restrict napari to 0.4.* install napari-aiscimageio without updating dependencies (otherwise this caused a lot of issues with scikit-image and tifffile) pycudadecon |
or pycudadecon v0.5 from pip and you'll be fine. (but the numpy bit is up to you) |
When using Preview, I still get the deskewed image without processing It works with save... |
I can confirm that |
So to clarify, I'm going to:
|
Right, this is a bit tricky. The workflow can produce multiple outputs, including multiple images or no images. Should I add the "first" image output by the workflow, or all of them maybe? The non-image stuff can get thrown away I assume. |
No, not in pyproject.toml. I believe when we run a conda install with pycudadecon, numpy gets upgraded to v2. To prevent this we need to pin it only during pycudadecon installation. So, command to use for pycudadecon:
yes
yes
No, its working now with pycudadecon 0.5, so we don't need to specify this anymore.
We would like only the 'final' image output, if thats what you meant by 'first'.. With non-image stuff, I wanted them to be displayed as a table in napari, but I couldn't really get it working using napari-spreadsheet at this line in napari_lattice here. Not an immediate priority for now and we could include this as a separate PR.. |
Hopefully fixed in 3ddc78c. |
I'm getting another error now.. This works when I run Save, but when I use Preview, I get the following error with an empty array.
Workflow I tried:
This workflow should return a binary image.. I was using it in Crop and Preview Workflow. |
Oh wait actually it's the |
Sorry, which format? |
Yeah that's it. What happened was that your workflow returned a CL array, which doesn't know how to be summed. I've now fixed that with a test. |
Confirm its working! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All Changes look good and most of core functionality of the plugin and CLI are confirmed to be working...
Great work @multimeric
Pydantic
CLI
argparse
totyper
for the CLIGUI
Other
XArray
, which gives us labelled axis (closes Using xarray instead of dask array #41)Future Work
save_dir
a mandatory parameter, so that the user has to consider the output location. This will involve making theoutput
model optional, and also movingchannel_range
andtime_range
into theDeskewParams
. This was deprioritised due to the large impact this change would have