Cortex and RMS 20.2

Hradec

unread,

Apr 20, 2016, 3:56:19 AM4/20/16

to cortexdev

So... I'm trying to use cortex procedurals with RMS 20.2. After a simple patch in cortex (prman 20.2 doesnt have RiProceduralV, so I just patch it to use RiProcedural for now), I got IECoreRI building fine!

The problem is in maya. I cortex procedural node in maya, but when I render using renderman, the procedural is not being exported to the rib.

I get no error's or anything. I can see the attributeBeing/End section for the node, but no Procedural call in it.

Have any of you guys ever tried running cortex in maya with RMS?

here is a rib archive of the procedural:

##RenderMan RIB

#Created by RenderMan for Maya 20.2 (@1521997 Aug 25 2015)

#Wed Apr 20 12:52:20 AM PDT 2016

#bbox: -13.51105881 -2.535037577 -13.51110744 13.51109219 10.30395931 13.45505619

version 3.04

ScopedCoordinateSystem "world_ref"

AttributeBegin

Attribute "identifier" "string name" ["generic"]

Attribute "identifier" "float id" [3]

ConcatTransform [ 1 0 0 0 0 1 0 0 0 0 1 0 0 3.24274 0 1 ]

AttributeBegin

Attribute "identifier" "string name" ["genericShape"]

Attribute "identifier" "float id" [8]

Sides 2

ShadingInterpolation "smooth"

Attribute "user" "int receivesShadows" [1]

Attribute "visibility" "int camera" [1] "int specular" [0] "int diffuse" [0] "int transmission" [1]

Attribute "shade" "string transmissionhitmode" ["shader"]

Attribute "grouping" "string membership" ["+shadow"]

##RLF Inject SurfaceShading -attribute sets@,initialShadingGroup,

AttributeEnd

Hradec

unread,

Apr 20, 2016, 5:21:23 AM4/20/16

to cortexdev

Ok... I've dug a bit more and I notice that there's no code in Cortex to tell RMS to export an ieProceduralHolder node.

The way RMS works is by executing whatever mel code is defined in an .rman__torattr___preShapeScript string attribute found in the node.

So, I need to define a generic mel function that serializes the node parameters as a string to be passed as args to the RMS mel command, like this:

RiProcedural "DynamicLoad" "iePython.so" $bbMin[0] $bbMax[0] $bbMin[1] $bbMax[1] $bbMin[2] $bbMax[2] $ags

were $args is the serialized parameters string.

Could you guys point me how to generate the args string in cortex? (I known cortex has the code to do that, but I can't remember how!)

Also, I need 2 more things:

1. where should I drop this mel code to be sourced by maya when cortex is loaded?

2. Where can I hook up some initialization code for ieProceduralHolder to automatically create the .rman__torattr___preShapeScript that calls the export mel code?

sorry guys... its being a while since I code for cortex and I can't remember where things should go!!

cheers...

-H

john haddon

unread,

Apr 20, 2016, 11:41:24 AM4/20/16

to cortexdev

Hi Hradec, welcome back!

Support for RMS would be a great contribution - let's see what we can figure out.

Could you guys point me how to generate the args string in cortex? (I known cortex has the code to do that, but I can't remember how!)

It might be a bit confusing to follow, because the 3delight exporters are more complex than the RMS ones and it's all in C++, but for reference here's the equivalent code for 3dfm :

https://github.com/ImageEngine/cortex/blob/master/src/IECoreMaya/DelightProceduralCacheCommand.cpp

What might be more directly illuminating though is this line from the RenderManRender node in Gaffer :

https://github.com/ImageEngine/gaffer/blob/master/python/GafferRenderMan/RenderManRender.py#L132

1. where should I drop this mel code to be sourced by maya when cortex is loaded?

You can put it in a new file called "mel/IECoreMaya/ieRMS.mel", and then make sure that is sourced by "mel/IECoreMaya/IECoreMaya.mel".

2. Where can I hook up some initialization code for ieProceduralHolder to automatically create the .rman__torattr___preShapeScript that calls the export mel code?

How about just putting it in the ProceduralHolder constructor, in "src/IECoreMaya/ProceduralHolder.cpp"? Since it's just a string attribute, it doesn't sound like it will introduce a dependency on RMS, or otherwise get in anyone's way.

sorry guys... its being a while since I code for cortex and I can't remember where things should go!!

No need for apologies - it's great that you're looking at this stuff. We haven't built against PRMan for a long time, and my impression is that 3delight and PRMan have been steadily diverging, so do let us know if you hit any problems with IECoreRI.