The Single View Modeling project provides a tool to create a 3D-model from a single view image. This image can be a photography, a drawing, a painting or any computer-generated image. It is not a machine learning system, and the user has to provide all the information to make the computer generate the model.
It appears that when building a model from a single view image, it is useful to be able to fix the three coordinates of a precise point in the model. It is easier to set a local orientation of a surface : the user can intuitively warp the grid to make a 3D mesh. This is the basic idea of our project

The system starts with a simple grid, the size of which can be adjusted, matching the speed of the computer – the kernel is quite processor-consuming. Then, the user can add position or normal constraints to build the model. He can also add position and normal discontinuities, for example to get sharp geometric models, or make the separation between different layers.
The final tool is the fairing curve, which allows the user to propagate a constraint through a curve, via the control of curvature and torsion.

The final model can eventually be rendered with a 3D viewer, in a wireframe mode or textured, depending on the graphics options. Snapshots, vrml and animation are possible ways to export the model

The next links are a basic HOWTO for the last current version of the project. To get more precise and technical details, see the last link.

• Setting the constraints
• Adding discontinuities
• Fairing curve
• Adaptive resolution
• Interface Overview