We are already working off a shared network resource that all the machines have access to but the textures still fail to be read by the render slaves.
We also have access to the nightly builds. Can you recommend a particular nightly build that you know that the “transfer missing assets,” option in distributed rendering is working?
We need to have this option in order to get our work done.
In that case can you show me the error messages from the slaves when they fail to load the textures? What OS do you use for the master machine and the slaves?
We also have access to the nightly builds. Can you recommend a particular nightly build that you know that the “transfer missing assets,” option in distributed rendering is working?They are work in progress, so there is no specific build that I can point to. Using one of the latest builds will probably work.
Do you run the render servers as a Windows service? If yes, what user is the service running as? If it is the default system user, it may not have access to the network location with the textures.
This is true, but the log goes into the temp folder for the user that the service runs with. So if it runs with the default system user, the folder will be something like c:\windows\temp
We using DR on mixed renderfarm - Linux, Win and OSX machines. Problem with assets we solved such way:
1. Using centralized storage (Linux NAS server) for Maya projects. So all the resources located on it. Linux Maya stations uses autofs/NFS access to server
(“/hosts/server/share/project”). Windows Maya stations have SMB access to server projects folder (//server/share/project)
Linux and OSX render nodes have smb automount folder (like “/smb/server/share/project”) and it also linked to node’s root folder like (“/server/share/project” and “/SERVER/share/project”). so UNC path to project/resources which came from windows maya station looks same as those links (//server/share/project")
2. Using for addressing resources on Windows Maya UNC path (//server/share/project) instead of mapping drive letter (P:/project)
3. Using simple script for final check paths for all resources in Maya scene and copy/fix it to UNC path.
Thats all.
But IMHO, better to have some user accessible path mapping mechanism in Vray.
Vladimir Kolesnik
KPP
*) One is the VRAY_ASSETS_PATH environment variable which contains a list of paths to search for textures if they are missing;
*) You can use environment variables in texture file names, something like {$TEX_PATH}/to/texture/file.png - where the TEX_PATH variable may be defined differently on Windows/Linux.
Sorry to dig up an old thread, but rather than start a new one, here I am. Hopefully I didn’t overlook a more appropriate thread.
I’m using V-Ray 2.40.01 for Maya 2014 on Mac OS 10.6.8, and I get the same old “Cannot load texture file: Texture file doesn’t exist” on my slave nodes when using distributed rendering. This wasn’t a problem with Maya 2013 and V-Ray 2.25. I was under the impression that the “Transfer Missing Assets” option was removed because the logic to transfer missing assets automatically if necessary has been integrated. That doesn’t seem to be the case for me right now, because I’m getting the typical black buckets on buckets that get distributed. Am I missing something here?
Yes. Anything that registers assets with the 3ds Max asset tracker should work.
From my own tests it seems this is the case, but I have found that VRayScatter objects that use vrmesh do not transfer automatically.Do you see those .vrmesh files listed in the 3ds Max Asset Tracker?