Material duplicate (and delete) problem

[haggi]

Hi,

I'm using assemblies with maya and maya references. I want the assemblies to use the shaders in the scene. For this reason I exchange the material name in the assembly like this:

myShadingGroup

to

::myShadingGroup

This lets the assembly search in the current scene for a material called myShadingGroup, works really fine.
Now I have a problem: I use references in maya with namespaces. This leads to shading groups called:

myReference:myShadingGroup

Of course the assembly will be unable to find the correct shaders because they now have a namespace. Because mulitple references in maya will use the same assembly, there is no way to change the material name in the assembly. In maya I cannot simply rename the shadingGroups because they are locked. Then I had a simple idea:

Why not create a new material with the correct name in a geometry shader and hope that the assembly uses the new correct one. So I tried it like this:

The original material name is something like this: myReference:myShadingGroup
The newName is the same without namespace: myShadingGroup

miMaterial *newMat = mi_api_material_begin(mi_mem_strdup(newName));
miMaterial *origMaterial = (miMaterial *)mi_db_access(matTag);
newMat->displace = origMaterial->displace;
newMat->shader = origMaterial->shader;
newMat->shadow = origMaterial->shadow;
newMat->photon = origMaterial->photon;
miTag newMatTag = mi_api_material_end();
mi_db_unpin(matTag);

And it worked! I was happy. But now a problem occurs:
To clean up my modifications, I collect the new created materials and try to delete them after rendering in an output shader:

for( size_t matId = 0; matId < materialTagList.size(); matId++)
{
        miTag newMatTag = materialTagList[matId];
        miMaterial *material = (miMaterial *)mi_db_access(newMatTag);
	material->displace = miNULLTAG;
	material->shader = miNULLTAG;
	material->shadow = miNULLTAG;
	material->photon = miNULLTAG;
	mi_db_unpin(newMatTag);

	const char *newName = mi_api_tag_lookup(newMatTag);
	mi_info("Trying to delete material %s tag %d", newName, newMatTag);
	mi_db_delete(newMatTag);
	const char *newNameA = mi_api_tag_lookup(newMatTag);
	mi_info("After delete material %s tag %d", newNameA, newMatTag);
}
materialTagList.clear();

As you can see, I get the miTag, retrieve the material, set the components to miNULLTAG to be sure that the material only and nothing else is deleted and then I delete it from db. But, it still exists. In the last few lines I try to get the name from a theoretically invalid tag. But I get a valid result. If I now try to render again everthing goes wrong with the shaders because the database seems to be corrupt now.

Maybe this approach is naive, but it was such a nice idea, but it seems my method is wrong, if it is valid at all.
Any ideas how I can delete my self created materials? Maybe there is something like an db_update or a complete different approach?

[gunter]

Haggi, mi_db_delete will delete the scene element however the symbol table entries
associated with your material are still present. mi_db_delete is a great function however
mi_api_delete is much more appropriate in your case.

Gunter

Hi,

I'm using assemblies with maya and maya references. I want the assemblies to use the shaders in the scene. For this reason I exchange the material name in the assembly like this:

myShadingGroup

to

::myShadingGroup

This lets the assembly search in the current scene for a material called myShadingGroup, works really fine.
Now I have a problem: I use references in maya with namespaces. This leads to shading groups called:

myReference:myShadingGroup

Of course the assembly will be unable to find the correct shaders because they now have a namespace. Because mulitple references in maya will use the same assembly, there is no way to change the material name in the assembly. In maya I cannot simply rename the shadingGroups because they are locked. Then I had a simple idea:

Why not create a new material with the correct name in a geometry shader and hope that the assembly uses the new correct one. So I tried it like this:

The original material name is something like this: myReference:myShadingGroup
The newName is the same without namespace: myShadingGroup

miMaterial *newMat = mi_api_material_begin(mi_mem_strdup(newName));
miMaterial *origMaterial = (miMaterial *)mi_db_access(matTag);
newMat->displace = origMaterial->displace;
newMat->shader = origMaterial->shader;
newMat->shadow = origMaterial->shadow;
newMat->photon = origMaterial->photon;
miTag newMatTag = mi_api_material_end();
mi_db_unpin(matTag);

And it worked! I was happy. But now a problem occurs:
To clean up my modifications, I collect the new created materials and try to delete them after rendering in an output shader:

for( size_t matId = 0; matId < materialTagList.size(); matId++)
{
        miTag newMatTag = materialTagList[matId];
        miMaterial *material = (miMaterial *)mi_db_access(newMatTag);
	material->displace = miNULLTAG;
	material->shader = miNULLTAG;
	material->shadow = miNULLTAG;
	material->photon = miNULLTAG;
	mi_db_unpin(newMatTag);

	const char *newName = mi_api_tag_lookup(newMatTag);
	mi_info("Trying to delete material %s tag %d", newName, newMatTag);
	mi_db_delete(newMatTag);
	const char *newNameA = mi_api_tag_lookup(newMatTag);
	mi_info("After delete material %s tag %d", newNameA, newMatTag);
}
materialTagList.clear();

As you can see, I get the miTag, retrieve the material, set the components to miNULLTAG to be sure that the material only and nothing else is deleted and then I delete it from db. But, it still exists. In the last few lines I try to get the name from a theoretically invalid tag. But I get a valid result. If I now try to render again everthing goes wrong with the shaders because the database seems to be corrupt now.

Maybe this approach is naive, but it was such a nice idea, but it seems my method is wrong, if it is valid at all.
Any ideas how I can delete my self created materials? Maybe there is something like an db_update or a complete different approach?

[haggi]

Works! Great, thanks a lot.