VrScene to renderfarm hints.

[Robert]

A few useful hints for everyone who needs to send vrscenes to a render farm:.

1. When V-Ray creates a vrscene it only includes the textures needed for the rendering, BUT NOT all the textures in the scene. So if a shader is not visible in the current view, or is in a hidden group etc. You won?t get that textures. This is different to how Modo LXO files work. Kind of my clever.
2. V-Ray will capture each frame in an animation, that caches all animated Objects and Mesh. The more is changed and the more frame the animation has, the bigger the file. Zipping only makes it 5% smaller, but 7z compression will get it down to 5-10%.
3. Modo uses replicators for particles too. So by default each prototype in a replicator item is seen as an individual animated item to make motion blur possible. Yet most people use replicators for static scattering of instances. This will produce a lot unnecessary data in a vrscene. The solution is simple the replicator item in Modo has an option to turn off motion blur of individual particle by deselecting animated replicator in the button properties of the item. Can save you 50% of scene file space.
4. GI caching is mostly unsupported by render farms with vrscenes. In order to do so do an multi frame incremental for IC and a flythrough for LC. You can manual type in see path for the files. The ?don?t render final image? option is not working, so simple set AA to a very low value for caching.

[GidPDX]

Thanks Robert- you've answered a few of my render farm questions.

[vladimir.nedev]

So if a shader is not visible in the current view,

That's not entirely true. If a texture is never hit by a ray, it's never loaded, but a reference to it is still included in the .vrscene file.

or is in a hidden group etc

Yes, in that case the texture won't be referenced at all in the .vrscene file.

So by default each prototype in a replicator item is seen as an individual animated item to make motion blur possible.

I just want to clarify that the geometry itself (vertices/faces/normals) is never repeated for the individual replicas.
Only the transformations. Still that could be a lot of data for a long animation with a lot of replicas.

Greetings,
Vladimir Nedev

[vladimir.nedev]

One more thing about number 3. Robert is talking about the option shown in the attached screen-shot.
It tells V-Ray for MODO that the replica's transformations are not changing from frame to frame.
Detecting this would require fetching the transformations each frame, as you can't ask MODO at the start of the animation to tell you whether they are changing or not, as far as I know.

Greetings,
Vladimir Nedev

[Robert]

Thanks for detailing that Vladimir, maybe you could add that to the Modo help file.. it took time to find out these things. Anyway, by time I hope render farms will also support rendering lxo files with V-Ray scenes.