Should uvNode be optional in `texture_node.sample(uvNode)`?

[ycw]

Description

Currently uvNode in texture(texture_node, uvNode) is optional, so shouldn’t texture_node.sample(uvNode) follow the same behavior? At the moment, however, the uvNode parameter in .sample(uvNode) is required.

Version

dev

[Mugen87]

IMO, calling sample() without an argument is somewhat obsolete sense since you can can directly use texture_node like demonstrated in #32153. If you need to sample the texture with updated uvs (e.g. because of an offset computation), use sample() instead.

@sunag How do you feel about this? I can understand why users would call sample() without an argument but the resulting syntax feels not right, tbh (meaning calling sample() but without defining the sampling location).

[sunag]

sample() shouldn't be used without uv as a parameter; this only works because if a custom UV isn't assigned, it will use the texture's default UV.

I think we can improve the warnings regarding this, as we did in #30783.

[ycw]

Understood. Would you consider making the uvNode input of the sample node required as well?

three.js/src/nodes/utils/SampleNode.js

Lines 83 to 91 in 94f3283

	/**
	* Helper function to create a SampleNode wrapped as a node object.
	*
	* @function
	* @param {Function} callback - The function to be called when sampling. Should accept a UV node and return a value.
	* @param {?Node<vec2>} [uv=null] - The UV node to be used in the texture sampling.
	* @returns {SampleNode} The created SampleNode instance wrapped as a node object.
	*/
	export const sample = ( callback, uv = null ) => nodeObject( new SampleNode( callback, nodeObject( uv ) ) );

[sunag]

When using sample in isolation, not in conjunction with a TextureNode, the first parameter should be a function and the second an optional UV.

I think the best approach would be to remove the second uv=null parameter and support Node as optional, e.g: sample(function|Node), for that, we should consider the possibility of use texture.sample(function|Node) which should respect a defined UV.

This should correct this ambiguity.