#z depth

Z-Depth Animation Test

Much better way to create depth passes in XSI rather than relying on the in-built depth pass preset volume shader.

Basically: (as I understand it, anyway. I could be wrong.)

1. Two null objects to control the distance of the z depth being calculated.

2. A scalar state node set to Ray Length in order to return the linear distances between the nulls and the camera. This range is already a 'gradient' if converted from scalar to colour but is currently un-normalized.

3. Create your pass for it.

4. Apply a Constant mat to all objects in the scene (use BG_obj_partition node in the render pass node for ease)

5. Change Range of the scalar state node, clamp it from 0 to 1 and set its old(input) ranges to be the distances of the far null and the near null using the ctr_dist_cam<elem> expression. This normalizes the values given from the scalar state node.

6. Add an intermediate Scalar to Colour node and pipe that to the surface node of the material being applied to all objects.

Once in 3ds Max 8 (sorry, haven't upgraded in years), I set up a camera at 100 units from center. I check off the camera's Orthographic Projection option. That way, we don't get the distorting effect of perspective; parts of the object that are further from the camera are not decreased in apparent size nor moved closer to a vanishing point.

Then I place the model in between, with the farthest part of the model placed at the origin (0,0,0).

As you can see from the grid, the distance from the camera to the closest part of the model (the bull's left horn) is about 40 units.

Now it's time to render in Z-Depth.

Render... > Render Elements > Add... > Z Depth

With a Z Depth rendering, the camera isn't concerned about light sources, colors or reflections. All the camera cares about is how far an object is from it. And the further something is, the darker it gets.

So how does it know what to assign black to? Infinity? No, that'd be too far. Instead of having the program guess, the user has to set up the parameter. "Z Max" is as far the camera will resolve; here, I put in 100. That's not some arbitrary number; as above, I put the camera 100 units away from the origin, and I put the farthest part of my model I wanted to resolve at 0.

"Z Min" is the closest distance the camera will resolve. In this case, I put in 40, because that's how far the bull's left horn is from the camera.

Anything with a distance from 0 to 40 will come out white. And anything in between Z Min and Z Max will be rendered in an appropriate shade of gray. It's those shades of gray that we want.

Let's render...

Once rendered, we have to save it. I chose to save in 8-bit Grayscale TIFF format, because the extra bits used for color would be extraneous and wasteful. 8 bits gets us 256 levels of gray. (Note: I wonder if newer versions of 3ds Max allow for 16-bit Grayscale. Also, I wonder if my my CNC set up would accommodate higher bit levels).