Loading data

How to load remote OME-Zarr, local filesystem directories, and multichannel datasets — both from the library API and the demo viewer.

Local filesystem loading

Load a local .zarr or .ome.zarr directory using the File System Access API:

import {
  KilnViewer,
  LocalZarrDataProvider,
  promptForZarrDirectory,
  preValidateLocalZarr,
} from 'kiln-render';

// Show the native directory picker
const handle = await promptForZarrDirectory();

// Optional: validate before loading
const issues = await preValidateLocalZarr(handle);
if (issues.length > 0) {
  console.error('Unsupported dataset:', issues);
  return;
}

const provider = new LocalZarrDataProvider(handle);
const viewer = await KilnViewer.create(canvas, provider);

Browser requirement: The File System Access API is currently only supported in Chrome/Edge. promptForZarrDirectory() throws if the API is unavailable.

Previously granted handles can be restored across page loads:

import { getStoredHandle, requestPermission } from 'kiln-render';

const handle = await getStoredHandle();
if (handle && await requestPermission(handle)) {
  const viewer = await KilnViewer.create(canvas, new LocalZarrDataProvider(handle));
}

See the Data Guide → Local OME-Zarr for details and limitations.

Multichannel

Multichannel OME-Zarr datasets (up to 4 channels) are detected automatically. Per-channel colour, window/level, and visibility can be controlled via the renderer:

// Check channel count
const numChannels = viewer.renderer.numChannels; // 1–4

// Set channel colour (RGBA, 0–1)
viewer.renderer.setChannelColor(0, 0.0, 1.0, 0.0);     // channel 0 → green
viewer.renderer.setChannelColor(1, 1.0, 0.0, 0.0, 0.5); // channel 1 → red, half intensity

// Set channel window/level (normalised 0–1)
viewer.renderer.setChannelWindow(0, 0.3, 0.4); // centre=0.3, width=0.4

// React to auto-leveling from OMERO metadata
viewer.onChannelWindowsChanged = () => {
  // Update your UI with new window values
};

See the Multichannel documentation for compositing details, the multichannel demo, and known limitations.

Pre-validating remote datasets

Check a remote URL for compatibility before starting a load:

import { preValidateRemoteZarr } from 'kiln-render';

const issues = await preValidateRemoteZarr('https://example.com/scan.ome.zarr');
if (issues.length > 0) {
  // e.g. unsupported dtype, missing multiscales metadata, etc.
}

Demo viewer — loading custom datasets

Add the dataset URL parameter to load your own data:

https://kilnrender.com/app/?dataset=YOUR_DATASET_URL

Example:

?dataset=https://ome-zarr-scivis.s3.us-east-1.amazonaws.com/v0.5/96x2/kingsnake.ome.zarr

OME-Zarr

Compatible multiscale OME-Zarr datasets require no Kiln-specific conversion — just point to a URL.

Supported formats:

• OME-NGFF v0.4 and v0.5
• Single-channel and multichannel datasets (up to 4 channels — see Multichannel)
• uint8, uint16, int16 and float32 input (float64 is read as float32); everything except uint8 is stored as r16float on the GPU

See the Data Guide for full format requirements.

Local datasets (File System Access API)

Load local Zarr datasets directly from your filesystem using the "Load Local" button.

Browser requirement: Local dataset loading requires the File System Access API, which is currently only supported in Chrome/Edge. Safari and Firefox do not support this feature.

How it works:

1. Click "Load Local" button
2. Select a .zarr or .ome.zarr directory
3. Grant read permission when prompted
4. Dataset loads with auto-leveling based on OMERO metadata (if available)

Note: When you load a local dataset, all URL parameters are cleared to ensure the new dataset loads with fresh defaults.