If using a delayed load shader on pointinstances where each point overrides the geometry file instanced:
How do you set the IFD bounds?

Are IFD Bounds transformed by the instance point?
Like you instance an unit box onto scattered points - would the IFD bounds be -m -1 -1 -1 -M 1 1 1
or
-m ($TX-1) ($TY-1) ($TZ-1) -M ($TX+1) ($TY+1) ($TZ+1)

if the later:
how do you query this in the delayed load?
I tried
point('/obj/geo1/READ“, instancepoint(), ”P", 0)

But in the IFD only a random number of points get the -m -M flags attached. Most of them are skipped during the generation …

Any tips or pointers are appreciated.

While writing this I realize of course I could override the IFD Bounds on the points directly like any other shader attribute - :roll:

Cheers,

Georg

rdg
If using a delayed load shader on pointinstances where each point overrides the geometry file instanced:
How do you set the IFD bounds?

Are IFD Bounds transformed by the instance point?
Like you instance an unit box onto scattered points - would the IFD bounds be -m -1 -1 -1 -M 1 1 1
or
-m ($TX-1) ($TY-1) ($TZ-1) -M ($TX+1) ($TY+1) ($TZ+1)

if the later:
how do you query this in the delayed load?
I tried
point('/obj/geo1/READ“, instancepoint(), ”P", 0)

But in the IFD only a random number of points get the -m -M flags attached. Most of them are skipped during the generation …

Any tips or pointers are appreciated.

While writing this I realize of course I could override the IFD Bounds on the points directly like any other shader attribute - :roll:

Cheers,

Georg

The bounds are on the geometry, not the transformed geometry. The instance point provides a transform.

mark
The bounds are on the geometry, not the transformed geometry. The instance point provides a transform.

Thanks Mark!
I heard that there is a bug registered causing delayed loads to ignore IFD bounds - whatever they are set to - can this be confirmed?

rdg

mark
The bounds are on the geometry, not the transformed geometry. The instance point provides a transform.

Thanks Mark!
I heard that there is a bug registered causing delayed loads to ignore IFD bounds - whatever they are set to - can this be confirmed?

There was a recent question, but I think there might have been a misunderstanding (possibly the same one about spaces) in the question.

I don't think there are any outstanding bugs with bounds on the delayed load.

Hi Georg, mark.

Unrelated to any instancing we actually had occasions in the past that caused delayed load bounds not to cook correctly or not at all when generating ifds and rendering these. This seemed to be fixed with a build update. Unfortunately it looks to have come back again. Neither the sop based bounds nor the explicit bounds seem stable with DLAs when generating IFDs with hbatch, this could be due to a cook problem in expressions etc but the errors do not occur when in Houdini itself only when generated by hbatch and rendered through mantra on the renderfarm . I will be submitting a bug to support on this tomorrow when I get back into work.

A related bug to instances and delayed loads I have found recently is that when instances do not compute procedurals correctly if there are any expressions in geometry procedural parm. For an example an if statement in that parm would not work and the sop bounding box would render instead of the geometry file. When the if statement was removed it rendered correctly. I will be submitting this bug to sidefx support along side the other.

Thanks
Nick