Device & Canvas
Setup a Device, add a render loop and clear screen.
ts
import { createDevice, Device, FrameContext, PlatformId } from '@gglib/graphics'
import { vec4 } from '@gglib/math'
export default async (canvas: HTMLCanvasElement, tools: HTMLElement, platform: PlatformId) => {
// The `Device` is the entry point to the graphics API. It wraps either a
// WebGL2 or a WebGPU context (depending on `platform`) behind a single API,
// so the rest of the code stays mostly platform agnostic.
//
// Device creation is asynchronous, `.ready` resolves once the backend
// (and, for WebGPU, the adapter and device) has been acquired.
const device: Device = await createDevice({ canvas, platform, autosize: true }).ready
// `renderPass` is the recording interface for draw calls, state changes and
// clear operations. There is a single instance that gets reused every frame.
const pass = device.renderPass
const color = vec4(1)
function frame(ctx: FrameContext) {
// Animate the clear color so it is obvious the loop is actually running.
const hue = ctx.time % (Math.PI * 2)
color.x = 0.5 + 0.5 * Math.sin(hue)
color.y = 0.5 + 0.5 * Math.sin(hue + 2)
color.z = 0.5 + 0.5 * Math.sin(hue + 4)
pass.setClearColor(0, color)
// fill the render target with the color configured above.
// Nothing else happens in this example - no shaders, no geometry yet.
pass.clear()
// send the recorded commands to the GPU. This will keep all the
// state on the render pass for subsequend commands.
pass.submit()
// reset the render pass state to it's default. Internally this will
// also submit all pending commands before resetting
pass.flush()
}
// For a simple frame loops we utilize the built in frame scheduler.
device.schedule(frame)
// Tear down function that will be called, when the example is unmounted.
// Always dispose the device to free GPU resources and the canvas context.
return () => {
device.dispose()
}
}