Virtual World Heritage: More Than Three Dimensional Models

Cliff Ogleby
Department of Geomatics, The University of Melbourne
Victoria, 3010, Australia
Email: c.ogleby@eng.unimelb.edu.au


Abstract
Too often virtual reality is seem as the preparation of 3d models of objects or sites, although the more traditional definition of the term generally includes 3d viewing and a high level of interaction on the part of the user. When applied to 'world heritage', the visualisation of sites and monuments of significance presents several problems.

There are many methods now available for the generation of 3d CAD data sets, ranging from simple applications like Canoma, and Photomodeller though to coordinate measurement arms, laser scanning and sophisticated multi-image bundle adjustment. However the generation of realistic and believable virtual reality visualisations requires considerably more than a rendered 3d model.

Experiences with the Ayutthaya project (Ogleby 1999) and Virtual Olympia (Kenderdine et al, 2000) have shown that the success or otherwise of virtual world heritage is as much dependent on the material maps and lighting as the geometry, in many cases these elements are more important. The perception of the viewer that a model represents 'reality' is of paramount concern when producing a image or animation, and in the ultimate interpretation of the model. The more 'photorealistic' a synthetic image appears, the more easily it is interpreted to represent the 'truth'.

Several international organisations including UNESCO have expressed concern that with the availability and affordability of rendering systems there is the tendency for interpretations of world heritage monuments to be taken out of context. Without the ontology of the print publication and peer review process, many visualisations may be misleading and even incorrect. This paper will present ideas associated with the development of a 'quality assurance' approach to labelling virtual world heritage images and animations.

Introduction
This paper is intended to be a general discussion on the purpose of visualisation of world heritage, on the data acquisition technologies, the production of 'models' of monuments and precincts, on the role of these visualisations and on the interpretation of this imagery. Perhaps even a discussion on the differences between 'virtual reality' and reality.

The level of technology available for the visualisation of data sets, whether heritage or even say DNA strands, offer the facility to create images practically indistinguishable from 'photographs'. 'Photorealism' is possible on the desktop, but is a 'photoreal' dinosaur believable?

Visualisation

Figure 1a-g Interpretations of Stone Henge (Piggott, 1978)


The visualisation of cultural heritage is not a new phenomenon, the past has for centuries fascinated people in the present. The preceding images show that a monument such as Stone Henge in England has attracted many 'interpreters' who have applied their criteria to analyse the form and function of the site. What is being attempted now with computer graphics is merely an extension of this, what is different is that now images appear far more realistic (photo-realistic) than paintings and drawings, and they are more readily believable as a truer interpretation. It could be also argued that because computer systems have created the images, there are therefor also more technically accurate.

In many instances this is not the case, and without the established ontology of the published paper the supporting text and data is not available to assist the viewer in moderating or questioning the interpretation. A painting can be seen as an artist's impression, a rendered CAD file may well be seen as more 'scientific', and therefor better?.

Whilst there are good reasons for the production of photo-realistic interpretations of cultural heritage, there are calls for care in their presentation and interpretation.

3D Models and Cultural Heritage
The last 5 years or so has seen an increase in the availability and functionality of rapid 3d data acquisition technologies, many created for the entertainment and Internet industry. These include photogrammetric systems designed for use of non-photogrammetrists, (Photomodeller for example, even Canoma) automated photogrammetric point cloud and feature measurement systems, laser scanners (for both artefacts and large façades) and shape capturing devices (Minolta 3D 1500 for example). Some of these systems are optimised for geometry (laser scanners for example), others for texture mapped models (Photomodeller, Canoma). None, at this stage, create models that are ready for use as 'virtual reality' without substantial editing or manipulation, although this is not always undertaken..

What this technology has enabled is the creation of models of cultural heritage by people who are not 'expert' in the object or precinct itself. These visualisations range from the very crude and useless (Figures 4a and b) to very photo-realistic (Figure 5e). The level of detail is not only dependent on the requirements of the project, but also on the intended interpretation of the model, and even whether it is to be viewed as a still image or an animation sequence. If the model is a record of what exists today, then it should truly represent that object with actual materials on the surfaces, and be proportionally correct (if it is leaning or in ruins, so be it). If the model is a reconstruction, then the materials should be what they would be at that time with appropriate wear and so on, and geometrically the objects may be in a more pristine shape.
Figure 2 A material map made from an actual texture
 
Figure 3 A default library material


 

 
Figure 4a and b A material map derived from original photography in a simple 3d modelling package


Photorealism
A definition of photo-reality, when applied to computer visualisations, could quite simply be that the synthesised image should appear exactly as a photograph of the object it represents (or is intended to represent, a reconstructed cultural monument has no real world object to compare against). Expanding on this idea, Fleming (1998, p3) offers ten 'Principles of Photo-Realism' (of which he says an image should conform to at least eight in order to be considered photo-realistic):

  1. Clutter and chaos
  2. Personality and expectations
  3. Believability
  4. Surface texture
  5. Specularity 
  6. Dirt, dust and rust,
  7. Flaws scratches and dings,
  8. Bevelled edges
  9. Object material depth
  10. Radiosity
It could also be argued that in order to be photo-realistic, an image should also carry camera artefacts like depth of focus/field, lens flare and astigmatism. Modern computer graphics systems like 3D Studio Max offer the ability to create photo-realistic images both through complex methods of material mapping and through variables in the configuration of the render camera. Other software even facilitates the addition of camera artefacts, so it is entirely feasible to create a synthetic image indistinguishable from a photograph. However is this always what is necessary or beneficial?

Figure 5a-e. The Philippeion at Olympia, c200BC



Conclusion
The obvious question raised by the 4 images above is which better represents the Philippeion at Olympia? One could (should) also ask how was this reconstruction derived, were expert sources consulted, was it made from limestone, is the roof decoration generic or were remains found to verify their shape and colour? Is the purpose of the image to analyse the structure from an engineering perspective, or to show the grandeur of a lost civilisation? Why was it even created?

This paper has no answers to the questions, instead the intention has been to raise the questions in a forum specialising in the visualisation of cultural heritage. There is a need to eventually provide answers.

References

  • Fleming, W. 1998. 3D Photorealism Toolkit. John Wiley and Son. 328p
  • Kenderdine, S. C.L. Ogleby, J.A. Risetvski and K. Da Costa. 2000. 1000 years of the Ancient Olympics: Treasures of Ancient Greece. The digital reconstruction of Olympia, 3D Zeus and website. Proceedings of the 6th International Conference on Virtual systems and Multimedia. VSMM, Japan. 727p
  • Ogleby, C.L. 1999. From Rubble to Virtual Reality: Photogrammetry and the Virtual World of Ayutthaya, Thailand. The Photogrammetric Record. United Kingdon, 16:651-670
  • Piggott, S 1078. Antiquity Depicted: Aspects of Archeological Illustration. Thames and Hudson, 64p