Question about the Implementation of quantize Method which calls prune and dedup

[gz65555]

I've noticed that the quantize method implementation involves calls to both prune and dedup functions. My understanding was that quantize is primarily intended to reduce the file size by limiting the precision of vertex attributes. However, the inclusion of prune and dedup seems to suggest additional functionality.

I have a model where I intend to use the quantize method to compress the data, but I do not wish to remove duplicate materials, which is the current behavior due to the dedup function being called within quantize.

[donmccurdy]

Quantization requires modifying data throughout the scene, sometimes including materials. To clean up any unneeded data, quantize partially runs the prune and dedup functions.

Probably the cleanup could be made more precise - to deduplicate materials only if materials were modified, for example - but that isn't implemented currently.

See:

glTF-Transform/packages/functions/src/quantize.ts

Lines 127 to 134 in f261f4b

	await doc.transform(
	prune({
	propertyTypes: [PropertyType.ACCESSOR, PropertyType.SKIN, PropertyType.MATERIAL],
	keepAttributes: true,
	keepIndices: true,
	}),
	dedup({ propertyTypes: [PropertyType.ACCESSOR, PropertyType.MATERIAL, PropertyType.SKIN] }),
	);

[gz65555]

@donmccurdy thank you for explaining. Perhaps, consider adding more specific options to control the behavior of prune and dedup. For instance,

doc.transform(quantize({
 dedup: false,
 prune: false
}));

[donmccurdy]

Could I ask why you want duplicates of materials in the model?

Disabling these steps means quantize would add redundant data to your file – possibly quite a lot. I think we would need a workaround that removes any extra data added by the quantize step without removing extra data that might have been there before. Please feel free to open a feature request or PR for this – I'm open to the idea, but I don't have plans to work on it myself right now.

Another workaround would be to make a locally modified copy of the quantize() function and run that in your pipeline instead.

[gz65555]

Well, the real-world scenario is that designers have a series of models, and we optimize the material size through quantization. These materials of glTF have the same properties in glTF, so they are removed. However, we actually need to adjust these materials at runtime to achieve different effects in different locations.

[gz65555]

This is the issue: #1185. We can discuss solutions under this issue, and I'll open a pull request over the weekend if I have time.

[donmccurdy]

Resolved in #1185.