Live 2D STXM Data Streaming #375

Activity

1. Relm-Arrowny

   added a parent issue on Sep 7, 2026

   • P99 Acquisition & Visualisation Interface Overhaul #373
2. Relm-Arrowny

   added

   Data Vis

   core dependencyMore needed to know before we can fully scope this work.More needed to know before we can fully scope this work.

   on Sep 7, 2026
3. Relm-Arrowny

   changed the title [-]Implement live 2D STXM data streaming to display scanning updates in real time. (Mapping ?)[/-] [+] Live 2D STXM Data Streaming[/+] on Sep 9, 2026
4. Relm-Arrowny

   added

   Decision neededDecision is required before this is ready to be scopedDecision is required before this is ready to be scoped

   on Sep 9, 2026
5. DanPorter

   added

   workflows

   on Sep 15, 2026
6. DanPorter commented on Sep 15, 2026

   

   DanPorter

   on Sep 15, 2026

   Collaborator

   More actions

   Am I right in thinking that each "pixel" on the STXM image will be the sum of a detector image?

   So then the specific question is - where does that sum (or other processing) happen?

   Tiled/ Nexus file etc will only provide the data array from the detector, but supposedly there will be another service from the data analysis platform team to perform small operations like this.

   The other option is to have a BlueSky device that reads the last detector image and returns the sum, but I don't know if that is possible.
7. Relm-Arrowny commented on Sep 15, 2026

   

   Relm-Arrowny

   on Sep 15, 2026

   Last edited by Relm-Arrowny

   CollaboratorAuthor

   More actions

   Am I right in thinking that each "pixel" on the STXM image will be the sum of a detector image?

   So then the specific question is - where does that sum (or other processing) happen?

   Tiled/ Nexus file etc will only provide the data array from the detector, but supposedly there will be another service from the data analysis platform team to perform small operations like this.

   The other option is to have a BlueSky device that reads the last detector image and returns the sum, but I don't know if that is possible.

   Image statistic can be made to be emitted per point, as EPIC can provides them directly (NDPluginStats). Not 100% if we can do the same if we are hardware triggering. Epic is probably the best place to do it, as the data is already in buffer as it need to do the file writing anyway, so there should not be any I/O issue.
   I would definitely not do it in Bluesky, you are pushing the image data to bluesky which will generate a very high demand on the network and the bluesky memory/cpu.
8. DanPorter commented on Sep 16, 2026

   

   DanPorter

   on Sep 16, 2026

   Collaborator

   More actions

   OK great, Getting the data directly the the image statistics from the detector device in EPICS sounds like the best way to go then.

   Whilst we are waiting on the decision of where detector data will come from, we should still be able to create this component with a mock API call and ensure we tell the core team how we are expecting the call to look.
9. NKatti2011 commented on Sep 16, 2026

   

   NKatti2011

   on Sep 16, 2026

   Collaborator

   More actions

   Below contains notes from meeting with @EmsArnold and others.

   Rendering

   • Davidia may be the likely/intended solution. Evaluation of the client and server is in progress;
     it has not yet been assessed specifically for plotting scan data or for layers, so
     those should be included in the evaluation before we commit.

   Where the data comes from

   The available path depends on how the plan is set up:

   • If a Bluesky plan writes to a PV, BlueAPI can feed that back: but only scalar data.
   • A file-writing detector cannot be read back through BlueAPI.
   • RabbitMQ also carries scalar data.

   So whether something like an intensity value is available at all depends on the plan and
   on how much processing is needed before it can be displayed. Recommendation: all data
   comes through Tiled as the single point of truth (see #389), rather than the UI picking
   a different transport per data type.

   Maths and reshaping

   • Tiled plugins for maths and reshaping are not widely used, so the maths should not live
     there. It belongs in the standalone transformation package described in Reference Frame Coordinate Transformations #379, sitting
     alongside or just below the Davidia server, the exact placement still needs a decision.

   Tiled transport

   • Assume all data goes through Tiled, even though Tiled will not hold the file directly.
   • VISR currently polls Tiled continuously, which errors while a file is being written
     to. Websockets have recently been made available on tiled-staging, which should remove
     that problem:

   I have deployed the latest version of tiled on tiled-staging.diamond.ac.uk which
   supports subscriptions via websockets. You make a websocket connection which the tiled
   server accepts without any checks and waits for the first message from the browser
   client to send the access token in a message
   json.dumps({"type": "auth", "access_token": access_token})
   If the access token is correct you are let through. You can subscribe to new
   nodes (data) being created for a i15/cm123/1 visit, or to the root, which gives you
   updates for everything.

   — Zoheb Shaikh