Eyeon:Manual/Fusion 6/Position Pass Concept

Introduction

The Position tools in Fusion offer a completely new way of working with Masks and Volumetrics for footage containing XYZ Position channels.

ZtoWorld offers the option to create those channels out of a Z-Channel and 3D Camera information.

Just for the sake of this overview we will refer to the World Position Pass as WPP later on.

What is a WPP?

The WPP reflects each pixel's XYZ position in the original scene as RGB color values.

So if the face the pixel was derived from in the original scene sits at 0/0/0 the resulting pixel will have an RGB value of 0/0/0 and thus will be black.

If said face sits at 1/0/0 in the original scene the resulting pixel will be fully red.

Of course if the coordinates of our face are something like -60/75/123 the piexel will hold those values as well.

As you can see here, due to the huge extent 3D scenes can have the WPP channel should always be rendered in 32bit floating point to provide the accuracy needed.

This image shows a 3D Rendering of a scene with it's center sitting at 0/0/0 in 3D Space and the according WPP. For better visibility the WPP has been normalized.

Different coordinate spaces

WPPs can be rendered in different Coordinate Spaces.

These include World Space, Eye Space and Object Space.

The image below depicts how those different spaces look like, the tools in Fusion require the WPP to be rendered in World Space though.

The scene input

The tools offer a scene input, which can either be a 3D Camera or a 3D scene containing a camera.

While tha camera is vital for the ZtoWorld tool, Volume Mask and Volume Fog will still generate their output without any camera attached or with the camera position set to 0/0/0.

However, connecting such a camera which lines up with the original camera the WPP has been rendered from

or setting the camera's position manually greatly improves accuracy and look of the resulting Fog or Mask.

The "invisible sphere"

The example scene shown so far has an empty background, meaning there is nothing in the scene apart from the ground plane and the cubes.

If applying fog to a scene like that which is larger than said ground plane the result will look similar to the "w/o Sphere" example shown below

because whith no WPP information outside the ground plane the resulting value will be 0/0/0, so the fog fills that area as well.

To get around that you can add an invisible bounding sphere to your scene to create "dummy" WPP values to help the fog tool to create the correct volume as shown in "with Sphere" below.

Tips for Position Pass Concept (edit)

EyeonTips:Manual/Fusion 6/Position Pass Concept