fruitflybrain / neu3d Goto Github PK

It might be useful to put some dividers (instead of folders which needs to unfold them to see them) to group, for example, opacity settings for highlight mode, opacity settings for pinned mode, size for radius, size for soma radius, size for synapse. This can make the settings for opacity and size a bit more understandable.

As title, update typing definition in typings/index.d.ts

addJSON should only resolve when threejs objects have been rendered.

For example, in these lines, the this.loadObjCallback should return a promise. The promise of the addJSON callback should in turn depend on that promise.

The current side effect of this bug is that examples/activity.html bugs out.

To replicate:

1. Load a few neuron skeletons
2. Pin any given neuron
3. Try changing BG opacity or wireframe opacity

Add capability to highlight both neurons and synapses.

Since the scene only contains objects that do not necessarily need to be rendered in an activate animation frame, could we benefit from using static rendering?

The specific changes are:

1. remove animate() function. Specifically, remove call to requestAnimationFrame().
2. change TrackBallControl to OrbitControl.

A preliminary test suggests that the performance difference is minimal, but worth a look.

A thread to trace performance enhancements.

as title

As title, perhaps consider removing Propman dependency since it is not actively maintained.

as title. add neu3d.dispose logic

• During Click-and-drag, disable raycaster such that the highlighted neurons remain unchanged as we pan/zoom the camera. Only update raycaster result after click-and-drag event is over
• Disable adaptive resolution during trackball events to improve performance during zoom/pan

This Code Block

    $('#' + this.container.id).on({
      'dragover dragenter': function (e) {
        e.preventDefault();
        e.stopPropagation();
      },
      'drop': function (e) {
        let dataTransfer = e.originalEvent.dataTransfer;
        if (dataTransfer && dataTransfer.files.length) {
          e.preventDefault();
          e.stopPropagation();
          $.each(dataTransfer.files, function (i, file) {
            let reader = new FileReader();
            reader.onload = $.proxy(function (file, event) {
              if (file.name.match('.+(\.swc)$')) {
                let name = file.name.split('.')[0];
                let json = {};
                json[name] = {
                  label: name,
                  dataStr: event.target.result,
                  filetype: 'swc'
                };
                ffbomesh.addJson({ ffbo_json: json });
              }
            }, this, file);
            reader.readAsText(file);
          });
        }
      }
    });

was written to enable drag-dropping files onto the neu3d widget relies on container.id being specified because it uses JQuery.

Change this to use JavaScript's native method, see here

Neu3D currently handles loading data of a few formats, here we standardize the input data formats.

Input format

MorphologyData sent for multiple objects should be a dictionary with the rid of Morphology data as the key,
and each value is a dictionary with the following fields:

• class: the type of data, e.g., 'Neuron', 'Synapse', 'Neuropil', etc.
• orid: the rid of the parent node,
• morph_type: the type of data
  and fields according to the morph_type.

MeshDict Object (output items in MeshDict)

• ffbo_json: {[rid:string]: data_to_be_loaded}
• type: 'morphology_json' | 'gltf' | string
• highlight: boolean
• visibility: boolean
• background: boolean
• color: THREE.Color | string (RGB hex)
• label: string
• colormap: string;
• colororder: 'random' | 'order;
• showAfterLoadAll: boolean;
• reset: boolean

if loading from file

• filename: string
• filetype: 'json' | 'swc' | 'syn' | 'gltf'

if loading from Obj

• dataStr: string

Morphology Transformations

The following attributes will be applied to all except gltf files.

• radius_scale = 1.
• x_scale = 1.
• y_scale = 1.
• z_scale = 1.
• x_shift = 0.
• y_shift = 0.
• z_shift = 0.
• xy_rot = 0.
• yz_rot = 0.
• xz_rot = 0.

Attributes added by _registerObject

• rid: string
• object: RenderObj
• pinned = false
• position: THREE.Vector3
• opacity: float

SWC (loadSWCCallBack)

Each row of the swcString is assumed to be a length-7 array with each item corresponding to the following entries

type

• 0: undefined
• 1: soma
• 2: axon
• 3: (basal) dendrite
• 4: apical dendrite
• 5: custom
• 6: custom

parent = -1 indicates root

Mesh (loadMeshCallBack)

• vertices: Array
• faces: Array

MorphJSON (loadMorphJSONCallBack)

MorphJSON is assumed to be a json of 7 fields (as below) where each field is an array of corresponding items.

identifiers used for Neurons:

• 0: undefined
• 1: soma

Synapses need to take the special format to start listing all presynaptic sites with parent -1 and followed by postsynaptic sites using sample of its presynaptic sites as parent, e.g.

GLTF (loadGltfCallBack)

• dataStr

As title.

Change following code block to enable multiple file upload when clicking on the upload button.

    let fileUploadInput = document.createElement('input');
    fileUploadInput.id = "neu3d-file-upload";
    fileUploadInput.setAttribute("type", "file");
    fileUploadInput.style.visibility = 'hidden';
    fileUploadInput.onchange = (evt) => {
      $.each(evt.target.files, function (i, file) {
        let reader = new FileReader();
        reader.onload = $.proxy(function (file, event) {
          if (file.name.match('.+(\.swc)$')) {
            let name = file.name.split('.')[0];
            let json = {};
            json[name] = {
              label: name,
              dataStr: event.target.result,
              filetype: 'swc'
            };
            ffbomesh.addJson({ ffbo_json: json });
          }
        }, this, file);
        reader.readAsText(file);
      });
    }