| home | sideways | animated gallery |
| model | 52 photos were uploaded and combined in the cloud to reconstruct the surface. A piece of the carpet was left in place to give a sense of orientation, and to mask the hole in the bottom of the model - appropriate, as the actual model is hollow, too.
The video was produced inside 123D Catch, which has a basic keyframing system. Keyframing can lead to instabilities in the path. Since there is no clean way inside the program to define and rotate around a given location on the model itself, the next step was to export the model to a digital format and view it in three.js. The medium-quality model produced by 123D Catch takes awhile to load and display, as it consists of 433,725 triangles and uses a 4096x4096 texture. MeshLab's Quadric Edge Collapse Decimation (with texture) function was used to reduce this to 50,000 triangles, and Irfanview was used to reduce the texture to 1024x1024. The carpet accounts for a surprisingly high percentage (72%) of the triangles. Note that if you look at the model from below you'll see a place just under the top of the model where there are no polygons. Because the camera couldn't be placed low enough to look up at these areas, 123D Catch doesn't have information and so leaves them empty. The wireframe toggle for the model is a bit flaky, you may need to click above the box to toggle it. |
|
| 123D Catch | still gallery |