Can we Canoma

Images courtesy MetaCreations.

MetaCreations those same wacky people that brought you Goo and Bryce are going after a new market the - "how do you create realistic looking virtual worlds if you're too lazy to model them" market. Canoma is a descendent of photogrammetry applications used by intelligence analysts and surveying types of applications that use photographs to build up 3D models.

The basic idea is simple you start with a single or several photographs of the same area and you add geometry to the image. Next you tell Canoma to map the image to the geometry and poof...you've go a virtual model. Well of course (surprise surprise) it's not quite that simple. In order for Canoma to map the geometry to the images, especially when the images are taken from multiple views it has to continuously satisfy mapping constraints.

The illustration at the top of this article shows the Canoma interface with one image. In this case an aerial view of some buildings. (A particularly amenable building due to it's simple geometry.)

Next we select some of the geometric building block from the bottom area of the interface and "pin" corners of those building blocks onto matching areas of the image. After performing this operation several times we wind up with a crude geometric approximation of the image.

Finally we tell the system to map the image onto the geometry. Voila! A 3D version of the previously 2D image. Note that the images are highly simplified and will a little more time (and skill) spent creating geometry the mappings would look substantially better.
The weird thing about using Canoma however is during the process of adding geometric primitives to the image. These are not static line drawings these are actively computed geometric mappings. While placing the 3D geometry into the scene, often the other geometry you previously carefully placed...moves! This is due to Canoma's attempt at reconciling the constraints you specified while placing the geometry. The "pinning" of object corners tells Canoma that that particular geometry goes with that spot of the image. You can also specify edge constraints via "beads". The long and short of it is, this tool isn't magic and takes practice.

Overall Canoma a the first fun to use 3D from images software package which with a little patience and persistence will help you produce compelling virtual environments. Canoma exports VRML also, creating separate texture files for each surface. There is a cool collection of resources also. It's not magic and don't be fooled by pictures of complex virtual worlds thinking they will simple to create. Like any other complex piece of software Canoma works much better the more experience you get with it's quirks. As always practice makes perfect and with enough practice Canoma can be a great time saver.

One Year Ago in Focus on Web3D ABC, Vee Are Emm Ell