Quick question, I was wondering if Houdini can create stereoscopic titles? I need to produce some titles for a short trailer I am working on. Any help would be great.

Houdini 9.5 does not have any explicit support for stereoscopic 3D. But, you can create a simple stereo camera rig by putting together two standard cameras. After that, you can render the scene for each of them.

Then you can process the rendered images depending on your your final stereoscopic image format. For example, for anaglyph, you can use COPs to composite both images together by taking the red channel from the left image, and green-blue channels from your right image. Somewhat similar to this tutorial: http://captain3d.com/stereo/html/tutorial.html [captain3d.com]

What is the math between the 2 cameras? Do they just butt up against each other or is their a standard distance between left eye and right eye cameras in the scene?

Does the nodal point factor into this at all or is it as simple as playing with the distance between the 2 cameras, rendering out and then going back into Houdini and adjusting the distance?

Also this will be for Polorized Stereoscopic 3D.

BTW, thank you for the link, it is very helpful as I am just starting out with stereoscopic work.

http://en.wikipedia.org/wiki/Stereoscopy [en.wikipedia.org]
see the section called: “Taking the pictures”

I'm sure there are better links out there…but this gives the basics…

This is a good reference …

http://local.wasp.uwa.edu.au/~pbourke/miscellaneous/stereorender/ [local.wasp.uwa.edu.au]

-Drew

What is the math between the 2 cameras?

Usually the two cameras point in the same direction (ie, the camera view axes are parallel) and are separated by some distance. The distance is arbitrary and you usually “play” with it until it looks good. The larger the distance the more depth perception it invokes.

Does the nodal point factor into this at all or is it as simple as playing with the distance between the 2 cameras

A more complicated camera rig would allow for non-parallel viewing axes, but that may cause key-stoneing missalignment between left and right image. To avoid that you will have to play with the camera projection matrix. But having parallel camera axes is good enough most of the time, providing that you shift (and then crop) the left and right images appropriately, as described in the tutorial. So there are two distances you will deal with: inter-camera distance and the horizontal image shift distance.

Also this will be for Polorized Stereoscopic 3D.

The post-processing should be similar to the anaglyph, except you won't have to copy channels, etc. You will simply crop the images appropriately and supply them both (left and right) in to the above-below format (or whatever else).

If you are interested, you can find a really detailed description of stereoscopy in this document [whitepapers.silicon.com]

Rafal,

Thank you for all the good reference material, that link really made it click in my head as to how far apart each camera should be.

rafal

What is the math between the 2 cameras?

Usually the two cameras point in the same direction (ie, the camera view axes are parallel) and are separated by some distance. The distance is arbitrary and you usually “play” with it until it looks good. The larger the distance the more depth perception it invokes.

2.5“ is usually where you want to start, as that represents the standard human head interocular distance (which means that each camera's X distance is 1.25” from center). A lot ultimately depends on where and how you display the material. If it's on an IMAX screen, then there are more constraints to ensure that, for example, a 6' object in real life ends up as a 6' object on the screen. Other formats however are a bit more forgiving.