Skip to main content


The postProcessGLTF function transforms parsed GLTF JSON to make it easier to use.

  • It adds loaded buffers and images to the glTF JSON objects
  • It creates typed arrays for buffer views


Postprocessing is done by default by the GLTFLoader:

import {GLTFLoader} from '';
const processedGLTF = await parse(..., GLTFLoader,);

To turn post processing off, and then optionally post process via postProcessGLTF function:

import {GLTFLoader, postProcessGLTF} from '';
const gltf = await parse(..., GLTFLoader, {gltf: {postProcess: false}});
const processedGLTF = postProcessGLTF(gltf);

After post-processing, the gltf scenegraphs are now easier to iterate over as indices have been resolved to object references:

const scenegraph = processedGLTF.scenegraphs[0];
for (const node of scenegraph.nodes) {
// no need to resolve indices
if (node.mesh.primitives) {
// Ditto
// ...


postProcessGLTF(gltf : Object, options? : Object) : Object

  • gltf is expected to have json and buffers fields per the GLTF Data Format Category.
  • options.uri - Set base URI (for image loading)

The GLTF post processor copies objects in the input gltf json field as necessary to avoid modifying the input JSON, but does not do a deep copy on sub-objects that do not need to be modified.

General Post Processing

Replace indices with references

The first thing that postProcessGLTF does is replace glTF indices with object references to simplify iteration over the scenegraph.

Background: The GLTF file format describes a tree structure, however it links nodes through numeric indices rather than direct references. (As an example the nodes field in the top-level glTF scenegraph array is an array of indices into the top-level nodes array. Each node has a mesh attribute that is an index into to the meshes array, and so on).

Adds id to every node

The postprocessor makes sure each node and an id value, unless already present.

Node Specific Post Processing


The following fields will be populated from the supplied gltf.buffers parameter (this parameter is populated by the loader via options.loadLinkedResources: true):

  • buffer.arrayBuffer -
  • buffer.byteOffset -
  • buffer.byteLength -


  • - Typed arrays (Uint8Arrays) will be created for buffer views and stored in this field. These typed arrays can be used to upload data to WebGL buffers.


The accessor parameters which are textual strings in glTF will be resolved into WebGL constants (which are just numbers, e.g. 5126 = GL.FLOAT), to prepare for use with WebGL frameworks.

  • accessor.value - This will be set to a typed array that is a view into the underlying bufferView.


  • While it can be very convenient to initialize WebGL buffers from accessor.value, this approach will defeat any memory sharing on the GPU that the glTF file specifies through accessors sharing bufferViews. The canonical way of instantitating a glTF model is for an application to create one WebGL buffer for each bufferView and then use accessors to reference data chunks inside those WebGL buffers with offset and stride.


  • image.image - Populated from the supplied gltf.images array. This array is populated by the GLTFLoader via options.loadImages: true):
  • image.uri - If loaded image in the images array is not available, uses gltf.baseUri or options.baseUri is available, to resolve a relative URI and replaces this value.


  • ...texture - Since each texture object in the material has an ...index field next to other fields, the post processor will add a ...texture field instead of replacing the ...index field.



  • parameters - see table

Sampler parameters (which are textual in glTF) will be resolved into WebGL constants.

glTF constantWebGL constant



  • sampler - will be resolved the the corresponding image object.
  • source - will be resolved the the corresponding image object.