Content idea: How to save outputs from napari

[melissawm]

📚 Documentation

It would be great to have a how-to guide on saving outputs from napari.

A possible outline:

• What is the best/quickest way to save data from napari
• What are the possible output formats (csv? svg?)
• What elements from a napari state/operation can be saved (layers? camera angle/position? just the data? a subset of the data as selected by a plugin?)
• Plugins that allow you to save in popular formats (gif? tiff?)

Related issues:

• Add viewer.save method napari#1286
• File IO napari#379
• napari_write_tracks doesn't save tracks layer napari#3936
• Saving 3D stacks as animated gifs doesn't work napari#3746
• Meta issue: roadmap for exporting output from napari napari#1068

May be related:

• save napari to zarr file napari#490
• No plugin found capable of reading '/path/to/Shapes.svg' napari#3342

[chili-chiu]

Thanks @melissawm!

Currently, not all layer types have napari native saving capability, and there's no support for metadata saving (from my understanding).

Current native data saving options (File > Save Selected Layer(s)) -
Image layer: lossless formats (recommend tiff for interoperability) and lossy formats (not recommended for data)
Points layer: csv
Shapes layer: csv
Tracks, Vectors, Surface layer: no built-in saving option :(

Perhaps Zarr could be an integrated way to handle the metadata and the diverse layers @kevinyamauchi?

[kevinyamauchi]

Currently, not all layer types have napari native saving capability, and there's no support for metadata saving (from my understanding).

I believe this is correct, @chili-chiu . Even though, there currently aren't built-in writiers for all layer types, I think it would still be valuable to have clear documentation on what can/cannot be written without a plugin and how to do it. We could then potentially link to docs on making writer plugins and napari-hub?

Perhaps Zarr could be an integrated way to handle the metadata and the diverse layers @kevinyamauchi?

Yes, I think this is a great long term goal! With the forthcoming tables spec, we will have the data structure required to save most napari layer types in OME-NGFF. However, there is still to work on the OME side to create specifications for how the data structures are used to store polygons (shapes), points, tracks, and meshes (surface). After the October OME-NGFF community meetings, I would be happy to have a chat about what it would take to make specs for the remaining layer data!

[rjlopez2]

Thanks for this effort.
In the meantime what is the recommended way to let's say save and read an image layer preserving its metadata?
Let's say I have an image layer that contain some metadata dict with number of info/features I want to associate with the image (fps, acquisition, some operations on the image, etc). I want to be able to save the image with this info. The default napari writer doesn't seem to preserve the info when exporting in tif format.

[Czaki]

At this moment we do not know a plugin that properly reads all metadata.

In the context of a sequence of operations, we now not even save such information, so the writer does not have the option to save it as it is not passed to it.

[rjlopez2]

Thanks
In the case of sequence of operation, I manage to deal with that. I wrote a reader plugin that can read the image an also collect the info for a specific data format (.sif from Andor). As I do operations I add that information on top of the native metadata.
for instance here is an example

This is obviously not a general solution but is sufficient for now.
the idea would be that when I export an image, let's say in tiff format this information is saved along and eventually also read it back. Could you provide a suggestion how could I achieve this?
I am still new to Napari

[Czaki]

You need to write your own writer for that. There is tifffile library to interact with tiff. You may take a look on OME TIFF file format to check if it allow to fit your data.

Maybe @melonora has more information for this (or knows who call).

[rjlopez2]

ok
I will see what can I do.
thanks

[melonora]

You could put all your custom metadata in the comment of an OME-Tiff. I believe this would allow for also reading it back in though it might not be automatically parsed in a manner that is useful. For that you would require a custom reader. Let me know if you have additional questions.