Closed hybridherbst closed 21 hours ago
The interactivity extension will add a few virtual helper properties to the glTF asset object model. So the endTime
input socket could be connected to pointer/get
pointing to something like /animations/0/endTime
.
Thank you! Is there a preliminary list of those somewhere?
Is the ability to bind graphs with assets at runtime considered a design goal of KHR_interactivity?
The graphs are stored inside glTF assets and use JSON pointers that make sense only within the same asset.
Thanks. To translate this into a concrete answer, that means it's not a design goal for graphs to be transferrable between assets. They always need to be exported from some generator for the final asset.
(Note the pointers in the example in my original post are chosen so that they would be transferrable between animated assets, I understand that won't work in a general case.)
It's possible that some future interactivity revision would add support for truly separable graphs but it's not a goal for now.
That said, graphs could be generic enough to work with different assets even with the current design in some cases. For example, a graph could query the total number of animations, find the one with a specific duration, store its index in a custom variable, and then start the animation using that index.
resolving this: to summarize nodes do not have implicit value properties and need to have inlined values or be connected to out values of other nodes
Not sure if this should rather go into the spec readme, or here, please let me know. I find it confusing that currently, nodes must declare all their properties, and there are no default properties.
For example, this doesn't seem to be valid:
Is the consensus here that graphs need to be so tightly coupled to the scene that they even contain info regarding, for example, the duration of each animation as part of the nodes? (e.g. would have to set "startTime" to 0 and "endTime" to "specific value for different files").
Some questions around this: