5.4. Tutorial: Taking the right photos

Problem

In general 3DSOM Pro will produce good models if a large number of photos (between 30 and 50) are taken from a range of different angles. However you may wish to use fewer images, taken from carefully chosen angles.

Solution

By considering the type of object being modelled you can determine the right number of photographs to take and which viewing angles are most suitable. This tutorial gives some examples for different types of object.
Cubical objects

When photographing objects that have flat faces, such as the cube shown below, it is recommended that photographs be taken such that one face is just visible. In the case of a cube, this would require taking photographs with the front face directly at the camera and the side just visible.

The reason for this lies in how 3DSOM Pro generates wireframe models, and the problem faced by 3DSOM Pro can be best illustrated in the following diagram.

The diagram shows the actual object, a cube, with two photographs shown of the cube. The actual cube is shown, along with the 'shadow' that is hidden from view in the photographs. The region of space represented by the 'shadow' cannot be seen, and therefore 3DSOM Pro can make no judgement regarding whether this region is filled by the object, or not.

When building a wireframe, 3DSOM Pro works by, in effect, carving away parts of the object using silhouettes. By taking photographs that are orientated to glance down the edge of a face, 3DSOM Pro is able to carve away at the object most precisely down that face.

The following diagrams show this technique in practice. There are two examples of a cubical object. One was created using photographs taken at various angles around the object; the other had four additional photographs taken at glancing angles down each side of the object.

Whilst the model on the left hand side can be seen to be mostly cubical in shape, there is an edge visible on the face of each side. The faces of the cube are almost flat, but due to the angles that the photographs were taken, there remains a visible edge. By comparison, the model on the right hand side has very sharp edges: the photographs taken from an angle almost parallel to the faces have allowed 3DSOM Pro to carve very sharp faces.

Cylindrical objects

When photographing objects that have a cylindrical part it is essential to provide 3DSOM Pro with a number of photographs around the cylindrical shape.

Whilst 3DSOM Pro can carve away, it can only do so using what could be imagined as a virtual guillotine. This is very good for straight edges and cubes, but when presented with a cylindrical object can result in square sides rather than round. This problem is alleviated using the geometry optimisation tool that smoothes the surface where there is no evidence of sharp edges - see Optimising the surface geometry.

.

Even with the optimisation tool the user should provide 3DSOM Pro with a reasonable number of images, taken at various angles, around the cylindrical shape. Whilst 3DSOM Pro will continue to create the initial wireframe by slicing off straight edges, the number of straight edges may be increased to the point when the cylindrical shape is produced (especially after optimisation).

The following diagram shows the recommended camera views to take of a typical cylindrical object.

The following diagram demonstrates the difference between taking 6 photographs and 30 photographs of a cylindrical object without optimisation.

Objects with both flat and curved surfaces

When dealing with objects that have flat and curved surfaces, as would be the case for many objects, then care should be taken to ensure that a suitable set of photographs are taken.

If the camera shown above has photographs taken just from the side, then 3DSOM Pro cannot see a correct silhouette of the camera from above, and therefore would not be able to correctly model the lens.

The flat sides and cylindrical lens also need to be taken into account. The flat surfaces require photographs to be taken glancing down each of the flat surfaces to ensure a clean edge is obtained.

The cylindrical lens also needs a number of photographs taken around it, to ensure that 3DSOM Pro is able to correctly model the curved surface correctly.

TIP:

An alternative way of handling this example is to take two separate scans using side view photographs (one with the cylindrical part pointing vertically up and one with the cylindrical part pointing horizontally). The two scans can be merged using the alignment tool described in Merging multiple scans.

The following diagrams show the effects of not taking enough suitable photographs compared to a carefully chosen set of photographs.

< Previous topic
(Merging multiple scans) 		Contents
page		Next topic >
(Editing the geometry in an external application)