Appearance

Loading VTK.wasm ​

Everything starts with a single call: loadAsync(options) returns a VtkWasmRuntime — a loaded WebAssembly module that acts as the factory for sessions.

js
import { loadAsync } from "@kitware/vtk-wasm";

const runtime = await loadAsync({
  url: "https://gitlab.kitware.com/api/v4/projects/13/packages/generic/vtk-wasm32-emscripten/9.6.20260228/vtk-9.6.20260228-wasm32-emscripten.tar.gz",
});

First you need loadAsync on the page. Pick whichever fits your setup (see Adding VTK.wasm to a Project):

  • HTML Script Tag — no build step; loadAsync lives on the global vtkwasm object.
  • Bundler — import { loadAsync } from "@kitware/vtk-wasm".

For the complete, generated list of exported functions and classes, see the API Reference.

Choosing options ​

loadAsync takes one options object. The most common choices are conceptual; the full list with types is in the API reference.

  • Where to load from — url points at a directory or a .tar.gz bundle. Skip it when the module is already loaded as a script. wasmBaseName (default "vtk") and urlIsGzip adjust how the bundle is located.
  • Rendering backend — rendering: "webgl" (default) or "webgpu". WebGPU requires async execution and is switched on automatically.
  • Execution mode — exec: "sync" (default) or "async". Async unlocks WebGPU and non-blocking calls but requires JavaScript Promise Integration (JSPI) support in the browser.
  • Console output — pass print / printErr to pipe C++ std::cout / std::cerr to JavaScript.

Runtimes are cached ​

Runtimes are cached per (url, wasmBaseName, rendering, exec). Calling loadAsync again with the same options returns the existing runtime instead of fetching and instantiating a second copy, so it is safe to call from multiple places.

Creating sessions ​

A runtime does nothing on its own — it creates sessions:

js
const session = runtime.createStandaloneSession(); // in-browser rendering
// or
const remote = runtime.createRemoteSession();      // server-driven rendering

Releasing a runtime ​

Call runtime.dispose() (or use a using declaration) to drop it from the shared cache. Note that Emscripten cannot reclaim a runtime's heap before a page reload — disposing the sessions is what actually frees C++ memory.

Reference: loadAsync · VtkWasmRuntime