GLTFExporter: Add support for KHR_mesh_quantization.

In the past, my model was compressed with draco, and it was displayed on the web page normally, and it was also normal to export the file with GLTFExporter. However, due to the draco compressed model, the decompression speed is relatively slow, and the display is no problem if it is changed to GLTFPACK compression. But when I want to export the page, I am prompted that the bufferAttribute component type is not supported.

var loader = new THREE.GLTFLoader();
loader.setMeshoptDecoder(MeshoptDecoder);
loader.load(model_path, function(gltf) {
theModel = gltf.scene;
theModel.position.y = -1.2;
modelg.add(theModel);
});
var urlsave = ‘http://192.168.101.113:8080/customProduct/saveCustomProduct’;
var qhxhr = new XMLHttpRequest();
qhxhr.open(‘POST’, urlsave, true);
var blobfile = null;
var fdata = new FormData();

var btnsave = document.getElementById(‘savejm’);
btnsave.addEventListener(‘click’, saves);
function saves() {
var exporter3D = new THREE.GLTFExporter();
exporter3D.parse(modelg, function(glbfile) {
if (glbfile instanceof ArrayBuffer) {
saveArrayBuffer(glbfile, ‘customize1.glb’);
} else {
var scenejson = JSON.stringify(glbfile, null, 2);
saveString(scenejson, ‘customize.glb’);
};
}, binary = true, onlyVisible = true );
};
function saveString(text, filename) {

  save(new Blob([text], {
  	type: 'text/plain'
  }), filename);
};
function saveArrayBuffer(buffer, filename) {
save(new Blob([buffer], {
type: ‘application/octet-stream’
}), filename);
};
function save(blob, filename) {
fdata.append(‘file’, blob, ‘customize.glb’);
qhxhr.upload.addEventListener(“progress”, function(result) {
if (result.lengthComputable) {
var percent = (result.loaded / result.total * 100).toFixed(2);
console.log(percent);
}
}, false);
qhxhr.send(fdata);
console.log(“ff”);
};

Hope to support bufferAttribute component type in GLTFExporter.

[Mugen87]

Small update on this issue: When fixing a bug in context of interleaved data, I’ve also made a test with glTF-Quantized 2.zip since it contains interleaved data.

I’ve realized it will be a bit complicated to add support for KHR_mesh_quantization since the exporter actually uses buffer data at various places. At least when recomputing normals, data of typed arrays which are not Float32Array have to be decoded into floats and then encoded back after the computation.

Besides, it’s necessary to honor specific (byte) strides and padding values in order to write valid buffer data for quantized values.

[donmccurdy]

I suspect it’s any model using KHR_mesh_quantization that will cause this error. If so, this file should reproduce the issue:

glTF-Quantized 2.zip