While thinking about how to best implement publishing of write nodes in Nuke, I’ve been thinking about child instances.
Basically in Nuke we have a final write node at the bottom of the node graph. This outputs the final renders for the node graph. Along the way down the node graph, there might be pre-render write nodes that “bakes” a part of the node graph to speed up the rest of the node graph.
This means there is an order of instances, and they are children of the main write node instance. Say an artist wants to publish the final write node, and maybe one of the pre-renders. Even if you presented the artist with an “final-render” and “pre-render” family, there could be multiple final write nodes in a graph, and the relationship between two instances aren’t apparent in the GUI.
What I would propose is to have a tree structure for instances in GUI, so artists can visually see how instances relate to each other.
Currently I not entirely sure where else this could be useful. My initial thoughts were that you can represent how a model instance would be a child of a rig instance, and so publish the model again if needed from the rigging work file.
Let me know what you guys think. There is quite possibly a better/different solution, that I would love to hear:)