Digitization workflow

How to proceed for creating a 3D model of a khachkar and its surrounding:

  1. Record position and orientation, measure the object around the edges together which right and oblique triangles – take dimensions of patterns – write down distances to adjacent objects – for more complicate contours take a photo and reconstruct them later on.
    Alternatively, if it possible to walk around the object, take a 360° photo or several pictures to cover the whole range and all surfaces or use a laser scanner together with a CCD camera to reconstruct the surface. Both approaches have pros and cons, the accuracy depends on the object’s size, the scanner technology and the lighting (shadows, wind and weather conditions) of the scene. Most problems come from varying photo shot parameters, changing light, obstacles, vegetation and overgrowth.
  2. To produce textures, photos are mostly processed in RAW format: Adjusting brightness, contrast, color balance, saturation, sharpening, noise reduction or special effects like rectifying or aligning, correcting parts and adding to, subtracting from, or intersecting with an existing selection (Lasso tool). Finally, textures or their parts are resized and exported to a convenient data format.
  3. The wireframe model of the object can be realized using a reconstruction software together with the laser scanner data or via a 3D modelling software like Blender. Based on predefined parametrizable geometric primitives, the recorded object data, corners and edges or reconstructed intersections of plane surfaces and starting with the fundament followed by the corpus and the top, the whole artwork can be rebuilt.
  4. Then, the textures should be adapted and placed on the model. If the contours are not straight and smooth, the textures must be clipped and trimmed to fit the surfaces and their borders by reconstructing them on the texture. Sometimes it is simpler to partition the surface into a triangle grid.
  5. To allocate the right texture to the surfaces, the classical uv-mapping is implemented and executed by hand. It is possible to replace missing textures by similar ones, e. g., perpendicular sides or the top side.
  6. As a final step, illumination, shading and smoothing allows for an advantageous presentation of the artwork before bringing it to the necessary export file format, in our case X3D.
  7. To create a natural background, X3D has several option, a background node, a skybox using six textures with heaven, clouds, soil, water, mountains, plants and trees. To replace a flat bottom, an X3DGeometryNode of type ElevationGrid can be used. The rescaled object is placed and oriented into the scene via the X3D nodes GeoViewpoint, GeoPositionInter­pola­tor and GeoTransform.

References

  • DAVID 3D Scanning. Available online: http://www.david-3d.com/
  • R. Popescu, A. Lungu: Real-Time 3D Reconstruction Using a Kinect Sensor. Computer Science and Information Technology 2(2): 95-99, 2014.
  • Straub and S. Kerlin: Development of a Large, Low-Cost, Instant 3D Scanner. Technologies 2 (2): 76-95, 2014.