Support writing Document containing multiple Buffers as GLB

[Makio64]

Describe the bug
On this particular model I cant convert the gltf to glb. but it works perfectly using the gltf-pipeline
It says there is no buffer but I dont understand whats wrong, works fine with other model i tested.

To Reproduce
here the model : https://www.dropbox.com/scl/fi/calx036yfn0h3qt9x67vw/camera_curve.gltf?rlkey=vltp12bbg7s021wmur0wa2kjp&dl=0
here my code for converting :

let io = null
const initNodeIO = async () => {
	let dracoDecoder = await draco3d.createDecoderModule()
	let dracoEncoder = await draco3d.createEncoderModule()
	io = new NodeIO()
		.registerExtensions(ALL_EXTENSIONS)
		.registerDependencies({
			'meshopt.decoder': MeshoptDecoder,
			'meshopt.encoder': MeshoptEncoder,
		})
		.registerExtensions(KHRONOS_EXTENSIONS)
		.registerDependencies({
			'draco3d.decoder': dracoDecoder,
			'draco3d.encoder': dracoEncoder,
		})
}
initNodeIO 

async function convertGltfToGlb(inputPath) {
	const document = await io.read(inputPath)

	// write the glb
	const outputPath = inputPath.replace(path.extname(inputPath), '.glb')
	await io.write(outputPath, document)

	console.log(`Converted ${inputPath} to ${outputPath}`)
	return outputPath
}

Expected behavior
Convert the model with no problem.

Versions:

• Version: 3.9.0
• Environment: Nodejs Windows & Mac

[donmccurdy]

@Makio64 is this the error you're seeing?

GLB must have 0–1 buffers.

Perhaps the error should be more specific here, but the problem is that this model has many (13) buffers. Normally that would mean 13 external .bin files. Cinema4D is exporting with 13 large base 64 Data URIs, which is kind of a strange choice.

In any case, GLB requires the buffers be combined, and that's easy to do. Use unpartition() before exporting to .glb, it's lossless and won't hurt anything if the model doesn't have this problem —

https://gltf-transform.dev/modules/functions/functions/unpartition

[donmccurdy]

The CLI does this automatically as well:

glTF-Transform/packages/cli/src/session.ts

Lines 79 to 81 in eaacc26

	if (this._outputFormat === Format.GLB) {
	await document.transform(unpartition());
	}

[Makio64]

Thanks, yes the more specific message would have help, especially if it point to the unpartition.
Maybe it could be a good idea to do the unpartition automatically if there is too many buffer to convert ? And add a warning explaining an unpartition had been done on the way ?

[donmccurdy]

I think it's important that io.write never modifies the input, so these should give the same result regardless of order:

// (a)
io.write('out.gltf', document);
io.write('out.glb', document);

// (b)
io.write('out.glb', document);
io.write('out.gltf', document);

Perhaps it wouldn't be too difficult, when writing to GLB, to ignore the buffer layout specified by the document without modifying the document. I'll mark this as a feature request if someone would like to look into that here:

glTF-Transform/packages/core/src/io/writer.ts

Lines 445 to 454 in eaacc26

	json.buffers = [];
	root.listBuffers().forEach((buffer, index) => {
	const bufferDef = context.createPropertyDef(buffer) as GLTF.IBuffer;
	const groupByParent = context.accessorUsageGroupedByParent;
	const accessorParents = context.accessorParents;
	const bufferAccessors = buffer
	.listParents()
	.filter((property) => property instanceof Accessor) as Accessor[];
	const bufferAccessorsSet = new Set(bufferAccessors);

This may affect implementation of Meshopt and Draco compression, in which case it could turn out to be simpler just to log a more specific warning.