Skip to content

2023-10-10.md

Latest commit

 

History

History
103 lines (79 loc) · 4.42 KB

2023-10-10.md

File metadata and controls

103 lines (79 loc) · 4.42 KB

2023/01/10 - Microscope development meeting

Attendees

Attending:

  • David Miguel Susano Pinto
  • Ian Dobbie
  • Tom Fish

Notes

  • additional metadata to cockpit images (microscope-cockpit/cockpit#676)

    Ian has spent some time working on microscope-cockpit/cockpit#676 and microscope-cockpit/cockpit#391 which add extra metadata to images saved from snap and mosaic. Tom says that this will be quite useful since at the moment metadata for mosaic images are saved on a text file but often get lost when files are moved around. Tom and David to review and comment.

    This only affects images from snap and mosaic. Images for experiments get their metadata collected at the beginning of the experiment. One thing discussed was that the timestamps used in experiments files are computed as difference from the first image. But we are not sure if this is per camera. What happens when cameras are in different computers?

  • DeepSIM code

    A draft if the DeepSIM paper is ready. The code used in DeepSIM are specific branches in Danny's fork. Ideally these changes would have been developed with us but that didn't happen. It is not clear what some of the changes do, many seem specific to DeepSIM and would not work on other systems, and there's just too many of them to review.

    • Python-Microscope fork

      Code is at https://github.com/dstoychev/microscope/tree/deepsim/. There's effectively two changes: a fix for the case when pattern 131 on Alpao is skipped (python-microscope/vendor-issues#12) and a new Alpao specific flush_patterns method. David will try to get his hands on an Alpao mirror to test this but the flush_patterns should have gone into the ABC

    • Microscope-Cockpit fork

      Code is at https://github.com/dstoychev/cockpit/tree/deepsim/. Ian took a look a quick it and reckons that many changes are DeepSIM specific and can't be merged. Also, some changes seem to be incompatible and there's just too many merge conflicts. No one is willing to do the work to merge this.

    • Microscope-AOtools

      Code is at https://github.com/dstoychev/microscope-aotools/tree/deepsim-matthew-danny/. Nick had a look and reckons most changes are at the UI level. Still, there are a lot of changes and we're leaving it to Nick to decide what to do. However, at least some of the changes here are dependent on the changes in Python-Microscope so there's no point in accepting these as-is without accepting changes on Python-Microscope too. It is quite possible that these changes are also dependent on the changes to Microscope-Cockpit.

  • Digital Z stack experiment (microscope-cockpit/cockpit#792)

    This was never merged but Ian really though it was. Implementation is described in microscope-cockpit/cockpit#792 and the need is described in microscope-cockpit/cockpit#691. David will try to take a look.

  • zoom() in PiCamera (python-microscope/microscope#285)

    Ian implemented this because he thought it would be faster than dropping the extra region in Microscope. It wasn't any faster. He left the code in case it was useful but it is not because it isn't any faster. It is also a device specific method, not part of the ABC. This method will be removed.

  • setting camera trigger type (microscope-cockpit/cockpit#816)

    The settings dict is meant to provide a lower level list of device properties from the device vendor. Originally, we exposed anything we though that could be useful there such as trigger types but we don't do that anymore. Cockpit does not have an interface to select ROI and trigger type and binning, it previously used the settings dict. As they get removed from the settings dict, we need a way to set them from Cockpit. A possibility to add them without adding too much clutter is to make the settings dict window in cockpit have two parts, the top is the Microscope/Cockpit unified properties, and the bottom is the low level settings.

  • Saving data as Zarr (microscope-cockpit/cockpit#881)

    This has been discussed a few times. The main alternative before was OME-TIFF but the new format in town seems to be OME-Zarr. At least this has a native Python interface that could make it easier. Ian will try to save snap images as a test.

    At the end, the main reason we continue stuck with dv format is that we mainly used this for SIM experiments and the best SIm reconstruction is in SoftWoRx.