vtf-js

A javascript IO library for the Valve Texture Format.

Overview

vtf-js is a canvas-independent library for reading and authoring Vtfs. The library supports encoding and decoding VTF versions 7.1 - 7.6, including Strata-format compressed Vtfs. Mipmap generation is automated by default, however interfaces for manually specifying image data for each mipmap are also available (see VDataCollection).

The following formats are supported by default.

• RGBA8888
• BGRA8888
• BGRX8888
• ABGR8888
• ARGB8888
• RGB888
• BGR888
• RGB565
• BGR565
• IA88
• UV88
• A8
• I8
• P8 ¹
• DXT1
• DXT3
• DXT5
• R32F
• RGB323232F
• RGBA16161616
• RGBA16161616F ²
• RGBA32323232F

Additional Setup

While vtf-js should work mostly out of the box, there are several things that this package needs help with: DXT encoding, and Strata Deflate/ZSTD compression.

vtf-js includes a DXT decoder by default. Unfortunately, native encoding is not supported at this time. To enable this functionality, the vtf-js/addons/squish module can be used to register the libsquish-js library for encoding/decoding.

Support for Deflate decompression is provided by default using the standard Compression Streams API, however this compression interface is limited in functionality, and a compression level of -1 is enforced when encoding due to technical limitations.

To add external compression support, import one of vtf-js/addons/compress/node or vtf-js/addons/compress/fflate to choose a compression interface, or manually register compression/decompression methods by calling setCompressionMethod from vtf-js/utils.

// import { ... } from 'vtf-js';

// For node (using built-in zlib bindings)
// import 'vtf-js/addons/compress/node';

// For web (using fflate)
import 'vtf-js/addons/compress/fflate';

Examples

Create from image data

import { Vtf, VFormats, VFilters, VFrameCollection } from 'vtf-js';

// ...

const frames = new VFrameCollection([image], { mipmaps: 3, filter: VFilters.NICE });
const vtf = new Vtf(frames, { version: 4, format: VFormats.DXT5 });
const out = vtf.encode();

Read image into an object

import { Vtf } from 'vtf-js';

// ...

const vtf = await Vtf.decode(inbuffer);
const slice = vtf.data.getImage(
	0,  // Mipmap
	0,  // Frame
	0,  // Face
	0   // Slice
);

// The image's data may be a uint8, uint16, or float32 array depending on the format.
// The convert method normalizes the image into the desired format.
const data = slice.convert(Uint8Array);

Re-encode an existing image

import { Vtf, VFormats } from 'vtf-js';

// ...

const vtf = await Vtf.decode(inbuffer);
vtf.format = VFormats.RGB565;
vtf.version = 6;
vtf.compression_level = 4;
const out = await vtf.encode();

1. As the Vtf format does not specify a singular method of defining palettes in the file, vtf-js interprets P8 images as a single-channel greyscale image. ↩

2. This codec is registered only in environments where sec-float16array is supported. ↩