Key Modelling Concepts

This section briefly introduces the key concepts that are involved in generating models using Real2Virtual Modeller. Key words used throughout the document are in bold.

From what can I generate models ?

Real2Virtual modeller can generate fully textured 3D models from images and drawings from sources such as ...

Digital camera photographs.
Scanned Images.
Graphical images saved from graphics packages, such as PhotoShop or Corel Draw.
Drawings created with Modellers Drawing Tools

What do I have to do ?

Using Modeller you will create a Project which maintains all the information about your model. The project may be created using a Project Wizard which helps you through each stage of the process. The project always contains ...

A set of 2 or more Views. A View is a description of what the model looks like from a certain direction, called a Viewing Direction. For example the view may be a photograph or drawing of the object seen from the front.
Closed curves drawn in the Views, called Silhouettes. These tell Modeller the shape of the object seen in that view.

The first choice to make when creating a project is how to specify the Viewing Directions. Two methods are available ...

Provide Viewpoints : You provide the viewing directions. These may be simple standard viewing directions like 'Front', 'Top' etc. or may be from special directions of your choosing.
In this case Views may be photos or drawings of the object from any source. A project of this kind need not contain a description of a Camera.
Viewpoint Estimation : You can let Modeller calculate the viewing directions for you.
In this case the Views must be photographs of an object resting on a rectangular shape of known size (such as a piece of paper). The rectangular shape is called the Viewpoint Target, and is used to automatically calculate the Viewing Direction for that view. You tell Modeller where the rectangle is in the view, by supplying a Viewpoint Cue this involves pointing to two points on each side of the viewpoint target.
In addition to this the project must also contain a description of the camera used to take the photographs.

Once a project has views, viewing directions, silhouettes and a camera definition (if required), then the model may be created - this is called Building the model. A model may also be Exported to a format called VRML which allows the model to be published on a Web page.

How are models made ?

Modeller generates models by...

Taking the shapes of objects in the project views (the silhouettes defined by you) together with their viewing directions, and then inferring a 3D shape that is consistent with them.
Adding surface details called Texture to this 3D shape. The details come from one of the following sources.
- From the digital camera/graphical project images. Texture is mapped from the images onto the model, using the defined viewing directions. The process is called Texture Mapping
- From separate texture files. We call this Uniform Texture and is texture from a single image applied over the whole model surface, without using viewing directions. An example might be an image of wood grain being applied to a model of table, to give a wooden effect
- By you defining the colour of the surface.
- Possibly a mixture of these methods.

How do I tell Modeller about my camera ?

You only need to do this if you want modeller to calculate viewpoints for you from photographs you have taken of an object. Note that a Camera Wizard is available to lead you through this process. You can define a camera in one of the following ways,

Select the make and model of your camera from a Supported List of cameras that Modeller knows of already.
Tell modeller details of you camera that should be in the camera's manual. This will then add your camera to the Supported List of cameras.
Take a special Calibration Image and use this to Calibrate you camera.
You can supply more technical information to describe a camera, this is for advanced users only and is called a Custom Camera.

Does Modeller create just a single object from the project files ?

No, not always. It can create the following models

A single object
A group of distinct objects
A collection of parts which together make up a single object
A mixture of these.

How are object collections and sub-parts defined ?

The Silhouettes in the project may be assigned to a Group. A Group is a collection of silhouettes. Up to 100 groups may be defined.

Very often a single object is described having no sub-parts, and a single Group of silhouettes is sufficient to describe the model's shape. By default all silhouettes are in Group 0, and you need not use other groups
Sometimes better results are achieved if the model is made up of a collection of simpler parts. For example a chair may be split into parts such as the legs, back, seat and so on. Each part could then be described using a Group of silhouettes.
When the scene contains several objects you may put the silhouettes for each in a separate group. In some cases separate objects may be defined using a single group, but this is not true for all cases. If you are unsure, use separate groups.

How do I put a finished model on a Web Page?

Exporting to VRML generates a VRML (.wrl) file, and for textured models, a number of small image files. These should be uploaded with your web pages.
Modeller also generates a Web-Page which it stores with the model. This page can be used as-is to display the model on the Web, or you may modify it. People with VRML-enabled web-browsers will then be able to see and interact with the model, by viewing this page.
You can put a model on your own web page using the HTML 'EMBED' command
You should also consider using IBM's HotMedia to publish your model.
For more information see the section on Exporting To VRML