GPUQueue: writeTexture() method

Secure contextAvailable in workers

The writeTexture() method of the GPUQueue interface writes a provided data source into a given GPUTexture.

This is a convenience function, which provides an alternative to setting texture data via buffer mapping and buffer-to-texture copies. It lets the user agent determine the most efficient way to copy the data over.

Syntax

writeTexture(destination, data, dataLayout, size)

Parameters

• destination

  • : An object defining the texture subresource and origin to write the data source to, which can take the following properties:

    • aspect Optional

      • : An enumerated value defining which aspects of the texture to write the data to. Possible values are:

        • "all"
          • : All available aspects of the texture format will be written to, which can mean all or any of color, depth, and stencil, depending on what kind of format you are dealing with.
        • "depth-only"
          • : Only the depth aspect of a depth-or-stencil format will be written to.
        • "stencil-only"
          • : Only the stencil aspect of a depth-or-stencil format will be written to.

        If omitted, aspect takes a value of "all".

    • mipLevel Optional

      • : A number representing the mip-map level of the texture to write the data to. If omitted, mipLevel defaults to 0.

    • origin Optional

      • : An object or array specifying the origin of the copy — the minimum corner of the texture region to write the data to. Together with size, this defines the full extent of the region to copy to. The x, y, and z values default to 0 if any of all of origin is omitted.

        For example, you can pass an array like [0, 0, 0], or its equivalent object { x: 0, y: 0, z: 0 }.

    • texture

      • : A GPUTexture object representing the texture to write the data to.

• data

  • : An object representing the data source to write into the GPUTexture. This can be an ArrayBuffer, TypedArray, or DataView.

• dataLayout

  • : An object that defines the layout of the content contained in data. Possible values are:
    • offset Optional
      • : The offset, in bytes, from the beginning of data to the start of the image data to be copied. If omitted, offset defaults to 0.
    • bytesPerRow Optional
      • : A number representing the stride, in bytes, between the start of each block row (i.e., a row of complete texel blocks) and the subsequent block row. This is required if there are multiple block rows (i.e., the copy height or depth is more than one block).
    • rowsPerImage Optional
      • : The number of block rows per single image of the texture. bytesPerRow × rowsPerImage will give you the stride, in bytes, between the start of each complete image. This is required if there are multiple images to copy.

• size

  • : An object or array specifying the extent of the copy — the far corner of the texture region to write the data to. Together with destination.origin, this defines the full extent of the region to copy to. See destination.origin for examples of the object/array structure.

Return value

None (undefined).

Validation

The following criteria must be met when calling writeTexture(), otherwise a GPUValidationError is generated and the GPUQueue becomes invalid:

• mipLevel is less than the destination mipLevelCount.
• origin.x is a multiple of the texel block width of the destination format.
• origin.y is a multiple of the texel block height of the destination format.
• If the destination format is a depth-or-stencil format or sampleCount is more than 1, the subresource size is equal to size.
• The destination usage includes the GPUTextureUsage.COPY_DST flag.
• The destination sampleCount is 1.
• destination.origin.x + the destination width is less than or equal to the width of the subresource to write to the destination GPUTexture.
• destination.origin.y + the destination height is less than or equal to the height of the subresource to write to the destination GPUTexture.
• destination.origin.z + the destination depthOrArrayLayers is less than or equal to the depthOrArrayLayers of the subresource to write to the destination GPUTexture.
• The destination width is a multiple of the texel block width of the destination format.
• The destination height is a multiple of the texel block height of the destination format.
• destination.aspect refers to a single aspect of the destination format.
• That aspect is a valid image copy destination according to depth-or-stencil formats.
• The destination is otherwise compatible with the format.

Examples

In Efficiently rendering glTF models, a function is defined for creating a solid color texture:

function createSolidColorTexture(r, g, b, a) {
  const data = new Uint8Array([r * 255, g * 255, b * 255, a * 255]);
  const texture = device.createTexture({
    size: { width: 1, height: 1 },
    format: "rgba8unorm",
    usage: GPUTextureUsage.TEXTURE_BINDING | GPUTextureUsage.COPY_DST,
  });
  device.queue.writeTexture({ texture }, data, {}, { width: 1, height: 1 });
  return texture;
}

This can be used to define standard textures for use in material libraries:

const opaqueWhiteTexture = createSolidColorTexture(1, 1, 1, 1);
const transparentBlackTexture = createSolidColorTexture(0, 0, 0, 0);
const defaultNormalTexture = createSolidColorTexture(0.5, 0.5, 1, 1);

Specifications

Browser compatibility

See also

• The WebGPU API