how to import .emd (an HDF5 type) files obtained with FEI Velox?

[Mingjian Wu]

Hi guys,

I am recently working with Microscopy data acquisition and processing with the new FEI Velox software, which is build on python.
However, the software is just awful to work with. Very inconvenient, easy to crash... but the datasets I need is only obtainable with it.

Does anybody know how to import .emd (an HDF5 type) files obtained with FEI Velox?

Thank you!
Mingjian

[Francisco de la Pena]

If the new FEI format is a standard HDF5 format you should be able to open it with e.g. h5py:

>>> import h5py

>>> import numpy as np

>>> f = h5py.File("you_velox_file.emd")

Does the above work? If it does, then you can manually import data into HyperSpy. Could you send us an emd file to try to add support for it in HyperSpy?

Thanks!

Francisco

--
You received this message because you are subscribed to the Google Groups "hyperspy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hyperspy-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mingjian Wu]

Yes! the codes works.
Beside, I can also explore a bit of the .emd file content with the HDFViewer tool.
Please find in the attachment an example file (obtained via the Velox DPC module) of a vacuum area scan.
It contains four 1024x1024 images from the four segments of the DPC detector.

I think the complete file structure description should be more complex than this example, e.g. spectrum images, multi-detector data, in-situ (time serial) data obtained by Velox.
If nessesary, I can obtain later other type of data (as mentioned above) with reduced size for testing.
BTW, I also send an inquiry to the Velox developer engineer about the file structure, and I will post it if it allows by FEI.

Mingjian

PS. the file is too large, please find the file via this dropbox link:
https://www.dropbox.com/s/or8xba9qf690jiv/vac.emd?dl=0

[Francisco de la Pena]

These are excellent news, finally an EM manufacturer uses an open standard format to save the data. I hope that others will follow. 

The structure seems pretty straightforward, but any official information from FEI would of course greatly help us supporting the format.

May thanks!

Francisco

[Mingjian Wu]

BTW, I also found out EMD (electron microscopy dataset) was initially developed by Colin Ophus in NCEM.
I consulted him, but he replied that the .emd obtained by Velox seems slightly different to his EMD.
But anyway, the nature of its open standard is already a great progress, although the Velox software itself is still in a very infant state.

Mingjian

[Francisco]

I've just opened an issue about this. Anybody interested on this feature can subscribe to the issue to get notifications.

Thanks again,

Francisco

[Magnus Nord]

Thanks for the dataset! It would be very nice if you could get in touch with the people developing Velox, to either get the specifications, or get them to communicate with us (and Colin) directly. So we could make sure the EMD-files from different software stays compatible.

Magnus

[Mingjian Wu]

Here I have added few more example files obtained from Velox with different combination of detectors, spectrum image, spectra, etc.
https://www.dropbox.com/sh/mqt3ykvwflojq8b/AABNLwwBn1IyEgzPvpBxStgka?dl=0

I willl post later if I got any news from FEI.

[Francisco de la Pena]

These files will be very useful to implement the feature. I have mentioned the new files in the issue on this topic. Given that this is a development discussion, feel free to post any further information directly there.

Many thanks!

Francisco

[bmam...@gmail.com]

Dear Wu, 

I am a new user of both TEM and hyperspy. I am trying to import some of the Velox maps data into hyperspy for offline analysis. I know this post is 2 years old, but i was wondering if you could share the example files again, as they seem to have been deleted from dropbox.

Thanks!

BM

[Francisco de la Peña]

Dear Wu,

There is now an implementation of a reader of this format in HyperSpy, see PR #1662. If you don't mind installing a development version of HyperSpy you can simply install the code in the PR and, if it doesn't work for you, let us know by adding comments to the pull request. If you want to do this and you experience issues during the installation don't hesitate to ask for help in our gitter chat. Otherwise, this feature should be available in HyperSpy v1.4 due early next year.

Best regards,

Francisco

--
You received this message because you are subscribed to the Google Groups "hyperspy-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hyperspy-users+unsubscribe@googlegroups.com.