About Rendering

 
 
 

Whenever you render, Autodesk Softimage uses the options that you have set for the selected render pass. For more information about creating render passes, see Passes & Partitions. Rendering options set for the render pass are described in Pass Rendering Options.

Softimage gives you the option of rendering using any one of the following methods:

Multi-thread Rendering

Rendering on a computer with more than one processor is an excellent way to speed up a render process: each processor has access to the scene, and if you're working locally, there is no need to transmit scene files across a network. However, there are a few things to be aware of.

Even though there is only one mental ray process running during a render, the renderer is able to take advantage of all the computer's processors. However, additional licences may be required depending on the number of processors.

Despite being very efficient, this type of rendering has a few drawbacks. Because each processor is rendering at the same time, each one requires a certain amount of memory to perform its task; as a consequence, more memory is needed. This usually doesn't amount to a problem if you are working with a few processors (less than eight), but it can become very demanding if using several processors during rendering.

Important

On some large (8+) multi-processor systems, all available address space can be taken by the 32-bit mental ray executable.

If a 32-bit address leaves only 2GB of virtual memory to be shared by 16 or 64 processors, their stack sizes will be insufficient regardless of physical RAM.

Distributed Rendering

mental ray tile-based distributed rendering lets you share the job of rendering among a group of computers on a network. This reduces the time needed to render frames and it takes advantage of any unused processing power on the network.

The host that reads and translates the scene is called the master. This computer is responsible for connecting to all other hosts (called slaves).

When you start a render task, the master computer instructs the others in the group to start a mental ray rendering process, then it sends a portion of a frame (a tile) to each remote processor. It collects and assembles the rendered tiles when they are complete and then sends more jobs until the task is done.

Setting the Tile Size influences the speed of a distributed render.

For details on how to setup for distributed rendering, see Distributed Rendering.