Object Plugin doesn´t render with Physical Renderer

ThomasB

Hi as always shortly before the goal the disillusionment.
I am working on a Object Plugin and the code is getting a bit complicated but anywaysI
I have now noticed that the plugin does not render in the Editor-View and in the Picture Viewer it only renders parts of my returned object.. The object which is directly below the null object as clone is rendered other object which are under this null object in other null object are not rendered correctly. If I make it editable everything works fine and you can sie the hierachytree . it is just one level Hierachy nothing complicated.
I noticed this with my last object plugin as well. What am I doing wrong?

A rough description of the program:
It creates a virtual document into which it loads one model file at a time from disk with a specific structure depending on a menue choice entry . There is null object as Parent-Container in which are other objects as child all polygon objects.
Then another null object is created in which I subordinate and structure clones from the virtual document, then I load the material clones from the virtual document into the active document and read out the texture tags from the cloned object in the Null-Container and swap the original entries with the materials from the active document etc.
In the end this null-container will be cloned and output in GVO?
When I press the view render button, it flickers briefly but doesn't render.
I still use some GetEnabling mechanisms and in the message method there are also a few buttons that are used and in the GeDDescription -Method I create the dynamic Menu - Entries which reads the needed folder in the zip-File

Corona Renderer and Standard Renderer render both in the Editor-View and also in the Picture Viewer are working properly but Physical doesn´t . That really pisses me off. Also in the Editor everythin is visible correctly.
The insertion of the objects does not necessarily happens in the GVO it also hapens in the Message Method where for instance a function is called when a button was pressed which feeds the virtual Null-Object with the clones from the virtual document.
Standard Renderer also renders in the Editor but then not in the Picture Viewer the full hierarchy oh man
Best Regards

ThomasB

I found out that I set self.OptimizeCache() set in the init and forgot to ask in the GVO if dirty so I added:

dirty: bool = op.CheckCache(hh) or op.IsDirty(c4d.DIRTYFLAGS_DATA)
        if not dirty:
            return op.GetCache(hh)

Now it renders in the Editor View but not in the Picture Viewer. The base is rendered, so think I have to do the actions not in the Message, there I change just the Data-Sturcture but loading in the GVO yes.....I will try that out...

ThomasB

Yes that was the problem, I did the model structure after a message and returned the object in the GVO but I have to do this one function in the GVO now it works. Thanks

Manuel

Hi,

sorry for the late reply but i am happy you solved the problem.

The GVO is supposed to return the cache of the generator. This cache will be used as the object to be rendered. GVO is called from a thread and not the main thread. So, you are not allowed to add any kind of object in the scene. That is why you should add your object in the scene in the message function. Then you GVO function will kick in and will find the materials.

Cheers,
Manuel

ThomasB

@manuel
ok I understand, yes but there is no object added in the GVO, in the GVO I just trigger some function to exchange virtual objects which are not in any document, and there I also exchange the material links in the texture tags to the materials which are already part of the active document.
So the Message gets Postsetparameter for instance. Then in the active document some objects will be reordered and then the GVO will triggered is dirty and calls a function to exchange the virtual objects related to the reorderd structure. Then the texturtags material links will be read exchanged and then the virtual object will be returned.

I just have one function called carefully in the GVO which just triggers one time.
It adds a layer puts the node into the layer and loads the materials from the virtual document into the active document. then it puts the layer into a hidden description Baselink of the node, so that it can keep track. then it switches the variable to True.

        if not self.start_refresh:
            self.load_modules_new(op)
            if op[c4d.PY_LAYER] == 0 and op[c4d.PY_LAYER_ID] is None:                
                self.create_layer(op, doc)
                self.add_new_materials_layer(op)
            elif op[c4d.PY_LAYER] == 1 and op[c4d.PY_LAYER_ID] is None:                
                self.create_layer(op, doc)
                self.add_new_materials_layer(op)
            c4d.CallCommand(12147)
            self.start_refresh = True

May i also do this things in the Init() method, because I don´t know where to call this.
Also will the virtual document be destroyed when I close the node in the Object Manager?

Manuel

hi,
creating, inside GVO, a virtual document, and do operation there, is a valid workflow and we do it sometimes internally.
Problems occurs when you have material, links etc.

I was just trying to emit some hypothesis on why object can render in the viewport and not in the picture viewer.
If i understand correctly, your workflow looks safe.

About the last question, when you create your virtual document, you do it like doc = functionToCreateTheDoc(). When you exit the scope where you are, the python garbage collector will destroy the variable doc. If no other variable is using the same memory address, the memory will be cleaned.

Cheers,
Manuel

ThomasB

@manuel
the virtual document I already create in my __init__ function.
yes but I am a little bit unsure in terms of the layer and the material which just insert only the first time in the active document in the GVO. I would do it in the __init__ but I have no access to op, and doc. I tested a lot of scenarios, deleting the layer etc, he instantly creates a knew and puts the mats back into it, if the user presses an "Actualize" Button.
The most save workflow would be to let the user press a start button to create the layer, load the models into the virtual doc and then let the plugin do it´s job, if the layer accidently was deleted, the GetDEnabling method will hide the elements and the grayed out start button will be highlited again.
Cheers
Tom

Manuel

Hi,

did you tried the message MSG_MENUPREPARE? I would use that to "setup" anything in the scene when your nodeData is created.

Cheers,
Manuel.

ThomasB

@Manuel
The message system is a bit overwhelming for me I just get step by step forward.
Where and how do I call this?
node.Message(c4d.MSG_MENUPREPARE) ?
sorry for this newbie question

Manuel

hi,

That is a valid question. One step after another, we are here to help.

You can override the function NodeData::Message() to react to some message Cinema4D is sending to all NodeData. This is exactly what you are already doing to react to button pressed by the user. In the same function, you can react to the message MSG_DESCRIPTION_COMMAND for buttons, or MSG_MENUPREPARE when the user click on the menu to add the Node in the document. This is for example where phong tags can be added and attached to your object. This could also be a good place to load materials and create a material tag for your object.

We have two manuals explaining the message system:

on in the python documentation
one in the c++ documentation

they are no difference between python and c++ but those manuals have different type of information.
The c++ manual show how to add an annotation tag when your object is added to the scene.

Cheers,
Manuel

ThomasB

@manuel
Thank you very much!