insightsoftwareconsortium / itk-viewer Goto Github PK

Possibly leverage ome-ngff-validator rendering

• Establish connection
• Show remotely rendered pixels
• Continuously update rendering parameters of remote renderer (resolution, camera projection, camera pose)
• Usable mouse camera control client side

Bigger picture we pursue LinkML,

But for now harmonize

ngff-zarr-types.ts

with

https://github.com/thewtex/ngff-zarr/blob/main/ngff_zarr/zarr_metadata.py

https://github.com/InsightSoftwareConsortium/dam

Generates div's to contain view ports Viewport actors.

Default implementation uses Spectrum SplitView?

After this PR #132

• check 2D image data
• Check 45 degree rotated image data
• Check up direction for each axis.
• Show slice index in GUI slider, rather than 0 to 1 float. If slicing mode is Z, the setSlice is in world coords not index???
• Check that GUI slider allows selection of min and max slice. Does not round out top or bottom slice.
• Add pick axis GUI

1. Install micromamba, set up a python environment, install hypha
2. Have an npm script to start the local hypha server (python command line command
3. Start before the dev script with concurrently

Camera Types

• Orbit mode
• Pan mode
• Synced camera: follows movement of other camera (with offset?)

Cameras are owned by Viewer. Cameras are as assigned to viewports.

Save app state. Boot up app with saved state.

State on URL?
How to traverse actors and aggregate snapshots?

https://pnpm.io/workspaces

Operate in a "data" web worker?

Per discussion in #10

https://openseadragon.github.io/

https://github.com/InsightSoftwareConsortium/itk-viewer/blob/dbd666a8d045f612eba5ee02492181f96ea612ef/packages/element/index.html#L10C18-L10C95

• src/index.ts
• tsconfig.json
• npm run build:tsc

Add ImJoy setup and run methods on the Viewer class.

Per: https://github.com/oeway/webrtc-hypha-demo

• vite.config.js
• package.json scripts

Would be intersting to subvert the NPM focused Changesets workflow. But only if easy.

Need to do something about this:

That code errors when trying to load an ome.tif file. Need to fix that.

Then later optimization possiblities:

Right now, the remote-renderer package does not actualy request chunks from the returned Zarr store. remote-renderer really wants the image metadata (spacing, array size, etc.) to puppatear agave-renderer. Would be sweet if this chunk of code simply generated NGFF-Zarr metadata for the single image scale raw input, without calling to_multiscales, whose output will never be used.

Or we take bull by horns and feed the NGFF-Zarr output of remote-image::remote_zarr.py to agave-renderer service.