I have found it is safest, and I presume you are using ICE Cache? ......to setup the instancing on the point cloud reading the cache in - i.e. I copy and paste the instance section from the original cloud to the in cloud. So you can either read in the cache into the mixer, or a cache node - then if mixer add a simple ICE tree that does just instancing.

Does this make any sense?

S.

On 2013/11/04 6:05 PM, Jonny Grew wrote:
Hello list,

Just wondered if anyone can shed any light on some issues we have with mb feather tools....

When we're caching out the pointcloud we're getting sporadic results in reading the cache back.

We have 3 separate mb feather tool setups (body, head and wings) all under different models. These all reference the same feather group in it's own model) and all nodes have been switched to say 'this_model' as advised within the documentation.

We're using 6 different feathers and populating the surface using the envelope weights on the 'envelope mesh' that determines which feather within the feather group goes where. This all works great when live but when caching out we're finding that either: 1/The instances aren't there (the particles are - you can see if you change the pointcloud display property to point)
2/Certain feathers are missing (ie feather 3 and 4).

You can force the instances to be visible by adding another ice tree on the empty cache read point cloud with a 'set instance geometry' node, however this only populates with the first feather in the feather group. - As the compounds are locked we're not sure what attribute we could use to ensure that the correct feathers are in the correct place.

It appears that when things do sporadicly work that they only do from a single scene. Exporting the model from this scene and caching causes even weirder results which means it looks like using reference models in our animation scenes is a no-go. We'll have to export animation action clips, apply them into this scene that sporadicly works and export the ICE Cache into our render scene.

Any pointers much appreciated on stating our blindingly obvious faults or hooky work-arounds.

Cheers
Jonny



Reply via email to