Your photos form the foundation of the model. Consequently the quality and resolution (number of pixels) of your photos directly impacts on the quality of your model. A good quality photograph is well exposed, the whole of the subject is in focus and the entire subject is in bright, even lighting. Small automatic digital cameras will give good results, provided that these conditions are met.

To use the stereo effect, the Scanetica 3D Scanner automatically determines the angle between the two camera positions and each point in the subject (as shown in 'How it Works'). As with our own vision, this is effective only over a limited range of angels. Too close with too large an angle and it becomes impossible to match the images. Too far with too small an angle and the stereo effect is lost, and it is impossible to determine the distance to the subject.

The following diagrams illustrate both good and bad subjects.

Illustration A.

This is a good subject. The subject has a small 'depth' (Green arrow) compared to the distance to the camera (Blue Arrow). All parts have similar angles in both camera positions. An array of suitable subjects can be seen in the Gallery.

Illustration B:

This is a bad subject. The Subject has a large 'depth' (Green arrow) compared to the distance from the subject to the camera (Blue Arrow). This is because the parts of the subject that are close to the camera have too great an angle between the two camera positions, while the more distant parts are so far away that the angle between the two camera positions is almost zero no matter how far apart the photos are taken.

Only objects which are between about 5 and 20 times the camera separation away will form a good surface.

The Scanetica 3D Scanner works by matching up details of the subject in the two stereo photographs. If the subject has very few surface details then the resulting model will be a very poor representation of the subject. This is a particular problem with man-made objects made from materials such as polished metal or moulded plastic, and which are plain and undecorated.

Shiny subjects usually show reflections, which can cause a problem when making the model. Shiny objects act like a mirror. When you look in a mirror, the objects you see reflected have the appearance of being behind the mirror.

In the same way, shiny subjects can have reflections, and objects with curved surfaces are worse and can have many reflections. If the reflections overwhelm other surface details, then The Scanner matches the reflections, causing an 'indentation' to appear in your model.

The illustration (opposite) shows a 3D model created from images of a glazed pot. You can clearly see the indentation in the surface which corresponds to one of the reflections of the lights used to illuminate the pot. Note that the other reflection was in an area showing stronger detail and was not affected.

In general, very shiny objects make poor subjects, though it may be possible to turn the subject or adjust the lighting to reduce reflections.

Subjects which have very dark areas tend to make poor subjects since the dark areas often have insufficient detail for the program to match.

However even on completely black subjects, there is sometimes sufficient dust, lint or debris on the surface for The Scanner to work. If in doubt simply try making the model and then examine the result.

Subjects with very diffuse or 'open' shapes, such as wispy hair and most plants, cannot be modelled correctly as a single surface with sufficient resolution. However it is possible to produce a model of some subjects, such as dense bushes or trees, which is useful to provide the general impression of the subject, but which is not particularly accurate. It is simply a matter of making the model and then examining the results.

Simple 3D models are represented in the computer by a single surface. As a result it can be difficult to represent some subjects correctly.

One consequence is where a part of the subject stands out in front of other parts of the model; for example a statue of a person with an arm held in front of the body. We recognise the arm for what it is and imagine it, correctly, as being separated from the body along most of its length. The computer is not able to make that type of judgement and doesn't actually know what is behind the arm. As a result the model will have the arm joined to the body by the single surface of the model.

A second consequence of this happens when a few hairs droop over a subjects face. The Scanetica 3D Scanner matches the hairs in the stereo photographs and draws the surface through the hairs. This can give odd bumps which do not exist on the model. This is very obvious in the illustration on the right showing one of the photographs used to create the model with a few hairs over the subjects forehead. The effect on the resulting model is shown.

The texture (or 'colouring') of the model comes from the original photographs. Consequently, any uneveness in the lighting, such as shadows or reflections, will be visible on the model.

In addition, on areas which slope away from the camera, each pixel in the original images will represent a larger surface area compared to the regions which are facing the camera. This causes the texture on these areas to appear smeared. The steeper the slope away from the camera, the more the smearing.

This is illustrated in the 3D model opposite. The eye sockets of the replica skull slope very steeply away from the camera. Examination of the area inside the eye sockets shows strong smearing of the pixels.

In general it is not possible to make good models from moving subjects with simple equipment. The Scanetica 3D Scanner matches two stereo photographs and relies on the relative positions of all the points to determine the surface of the subject. Even slight movement can destroy this relationship and radically distort the model.

It is possible to minimise the problem of small movements of the subject by taking almost simultaneous photographs with two cameras. With practice these can be sufficiently good to make models of animals staying relatively still (For example, a pet such as the dog shown in the Gallery).

Unfortunately, without sophisticated, expensive equipment, it isn't possible to take good stereo photographs of subjects which are moving at any reasonable speed.