Company: Creative Dimension Software
This 2D-to-3D software is an updated and upgraded version of Canon’s Software Object Modeller. It was originally developed in the UK at Canon’s Research Centre. Now, the developers have set up their own company and bought the software from Canon.
In the process, the price has risen substantially from the original £846 to £1,500, though this includes 12-month support and free upgrades. It’s a silhouette 3D modeller that allows you to create 3D objects with photorealistic textures, by taking a sequence of photographs of real-world objects.
The idea is that you place the original object on a mat, which guides you as to where to position the camera. You then shoot it from all angles, either by rotating the object on a turntable, or by moving the camera between shots. A contrasting backdrop is useful to help the auto-masking process.
Although used mainly for smallish objects (up to about 30-40cm), in principle you could photograph very large objects, by printing a large-format mat and using a heavy-duty turntable or moving the camera and backdrop between shots. No specialist equipment is needed beyond a camera and a calibration mat with special markings that’s printed from a file provided with the software. Full instructions and tips for photography are given in 3DSOM’s excellent manual.
There are some limitations in terms of subject matter. Nothing with transparency will work, and fuzzy surfaces such as stuffed toys may give problems. Very shiny surfaces may not work, though photographers’ matting spray might solve it.
About 20 shots are needed, preferably with most of the mat in view. However, tricky areas can be shot individually without the mat in view, and aligned manually later in the production process. There’s no particular need for accuracy when it comes to the position of the camera, as the markings on the calibration mat allow the 3DSOM software to calculate the alignment and angles.
Digital images can be imported individually or as complete sets into 3DSOM Pro. The original automatic masking has been extended so you can now vary the threshold for the auto-masker with a slider tool. A new ‘shrinkwrap’ feature lets you draw a rough outline around the edge of the object and alter the slider to ‘shrink’ the mask inwards.
Behind the mask
Once the first image is masked, the software can automatically work through all the others using the same parameters. Automasking may fail initially, but you can tweak the thresholds and keep trying. Masks can also be individually edited in 3DSOM, or exported to Photoshop for manual editing and re-import.
3DSOM calculates a 3D wireframe from the differences in the silhouette outline between each image and automatically optimizes it. At the same time, the photographic images are merged to form a continuous photorealistic texture. The realism of the texture will cover up many slight inaccuracies in the wireframe. Texture mask files are very small, and can be exported for editing in Photoshop, and then re-imported.
If the model has significant detail in parts that cannot be captured in the first sequence of shots, you can turn it on its side or upside down, shoot another sequence, and merge two or more together to form a single model. This works well for many objects, but still cannot cope with deep concavities that don’t create a silhouette from any angle.
You can export a partly completed wireframe mesh to 3DS Max (or other 3DS compatible programs) and edit it there, refining shapes and creating hollows by Boolean subtraction. The meshes can be re-imported into 3DSOM for texture application.
A model life
3DSOM Pro has sophisticated features, but it costs more than its two rivals, UZR iModeller (from about £70 to £350) and D-Vision Works D-Sculptor 2 (£500 to £900). The price difference probably isn’t that much to a high-spending 3D specialist studio, and 3DSOM’s ability to exchange meshes with a ‘real’ 3D modeller makes it worth the extra money.