How To: Use Displacement

This page shows you how to use displacement on an object in Cycles 4D.

Displacement is handled through materials in Cycles 4D. It is also currently regarded as an 'experimental' feature in the render engine,although it works well.

Enable experimental features

To turn these on, go to the render settings and change the Features drop-down menu to Experimental:

Set up the object material

To get displacement you need to input a value into the output node's Displacement setting. A Noise texture node works fine. You can also enhance it by manipulating the value. In this node tree, the value is multiplied by 2 to give extra displacement:

Displacement Method

This setting is a material-wide setting and therefore is found in the Material Settings tab in the right-hand pane of the node editor:

Change the Displacement Method drop-down menu to the desired option:

Bump

No displacement, use bump mapping only.

True

Use true polygon displacement.

Both

Use both true displacement and bump mapping.

These images show the difference between these settings, using the node tree shown above:

Micropolygon Displacement

So far, this is a conventional displacement system. But Cycles also has a new micropolygon displacement, which has a number of advantages. It offers:

  • faster rendering
  • much less memory use
  • greater detail

This is set up on a per-object rather than per-material basis. To use it, attach a cyObject tag to the object and in the Displacement tab, set Subdivision Type to Linear or Catmull-Clark. The Cycles engine implements adaptive subdivision (i.e. more subdivisions in polygons nearest the camera), which in Blender you have to turn on manually but in Cycles 4D setting Subdivision Type to anything other than None turns it on automatically.

Subdivision Rate

The amount of subdivision is set in two places as the 'subdivision rate'. For the final render, you can find this in the render settings, in the Subdivision section. This is expressed in pixels and by default is set to 1, which means that 1 pixel is represented by 1 polygon. This can be set to a different level in the real-time preview window, where by default it is set to 8.

You can see that the smaller the subdivision rate, the more detailed the surface will be. Don't make the mistake of increasing the rate thinking that this will increase the subdivision!

Dicing Scale

There is an additional way to change the subdivision on a per-object basis. In the cyObject tag, there is a setting named 'Dicing Scale'. This value is used to multiply the subdivision rate (in both the final render and the preview render) for that object only. So for example, if the subdivision rate in the real-time preview is 8, but for a particular object you would like this to be 2, set the Dicing Scale to 0.25 (8 x 02.5 = 2). The tag will tell you the actual subdivision for this object in the 'Final Dicing Rate' line:

With very high subdivisions, pixels close to the camera might have to be subdivided so many times it would cause a huge increase in memory use and render time. Therefore you can put a limit on the number of subdivisions: this is found in the render settings, Subdivision section, as the Max Subdivisions setting.