Problems with the level of detail

In contrast to the render mode, the level of detail necessitates manipulating the geometry in order to display uncertainty. Hereby, the amount of detail proportionally increases the reliability of the sources and argumentation. However, there is no fixed way of describing the factor of change. Likewise, a change in detail is not always possible. There are cases, such as the rectangular ground plan of the podium, which does not offer any further simplifications. Furthermore, how can a classification or reliability index of sources be applied to the level of detail? Is it even possible to classify different sources over the level of detail? As a rule, segments can be modeled only when knowledge about them already exists. Does this not imply a certain degree of certainty?

117 One should also be aware of the fact that the level of detail also plays an important role in native 3D modeling. It was originally used to save computer resources and achieve optimal results between render time and image quality. Mixing the two applications could have fatal consequences. For example, the reality-based model of monument 434 had to be reduced in its level of detail too. The reason for this is not an increasing uncertainty but rather the possibility of continuing to work with the obtained data and limited computer performance. Similarly, a kind of paradox arises when comparing the uncertainty of the reality-based model to the evidence-based model (fig. 63). This means that the reality-based model has the lowest possible degree of uncertainty and the highest amount of detail, not considering the lod (=level of detail) of the performance- based reduction.

In contrast, two alternative versions of the evidence-based model are available; one with more detail, such as individual building layers, decoration and roof, and another one with less detail, only simplified boxes symbolizing the case and podium (fig. 63). Following the idea of a proportion between the level of detail and uncertainty, the version with more details should be less uncertain. However, it is not. In this case the behavior is anti-

proportional. To simplify the discussion, we will focus only on the podium’s case and the

two evidence-based models (lod1 and lod2).

Figure 63: Concept rendering of different levels of detail of monument 434. The rendering on the left shows the reality-based model (lod0), with the highest degree of certainty and possible details. The two models to the right are evidence-based. The middle one (lod1) shows more detail than the right one (lod2), but has less overall certainty (Brunke 2017).

Lod1 clearly shows individual building layers. Technically, they are realized by creating individual objects, with a slight bevel on each edge. Lod2 does not. It has only one block with the total dimensions of the podium and does not distinguish between individual construction phases. The remaining properties are all the same between lod1 and lod2. Comparing this with the actual remains of lod0, there is no clear evidence of the actual height of one building layer of the case. There is only the probability that it might be the same as in the core, or it might be similar to some other tumuli along the Via Appia. Both assumptions have a rather high degree of uncertainty but still imply the exact opposite at lod1. Lod2 indicates only the total height, which has a much lower degree of uncertainty,

118 since it can be fitted much better with the core of the podium. Yet the illustration with the bevel and individual building layers is regarded as the most representative. This is because the right balance has to be found. Likewise, the bevel was not chosen to illustrate uncertainty, because the case was built up from several layers. It displays a completely different kind of information. If one tries to assign the level of detail to one of the two concepts of uncertainty, it is not possible to achieve a classification after the class-based method. An ordering after the fuzzy-based method might be possible when using the uncertainty index as the factor that has to be multiplied with the level of detail. However, the level of detail does not offer such a unit yet.

Similarly to the render mode, the level of detail is best used as a tendency and not as a representation of actual values. As a definite reinforcement, it shows robustness against color blindness and it allows one to indicate parts with almost no evidences. This is one of the reasons for which it was used in the case study for the burial chamber, decoration, roof and filling. Taking the decoration as an example, almost no evidence is left. Some of the historical illustrations might show parts of the decoration. However, the literature describes specific kinds of ornamentation that occur on many monuments of this kind. In this case, there is most likely an object in the monument with no evidence of its actual appearance. Something similar is true for the roof. Nowadays, no roof is present, but several analogies and illustration suggest a cone, mound or plate as roof. Therefore, the rough form is known but not the exact exposition. In this case, it is better to use simplified geometric objects and only as much detail as necessary to recognize the object.

The significance can be enhanced by the wireframe representation of the render modes. The additional wireframe mode is important to make it clear that the simplicity of the segment originates from uncertainty and not the architecture.

Another example is the size of the facing bricks. It is known how high the building layers are and how high the average blocks along the Via Appia are. Therefore, there are two most likely possibilities for this part. Both can be modeled to indicate the range. However, since the length is highly speculative and due to the fact that there is no additional evidence, no length is added. It is basically one ring. Width can be derived from other blocks around the Via Appia. The previously named bevel is assigned only a height, but not a width or length.

The level of detail might be one of the most complex visualization methods of uncertainty. It is extremely difficult to draft firm and comprehensive rules. The main point is

119 supposedly the recursive uncertainty, considering the paradox between reality and the evidence-based model, and their relation of uncertainty and details. The best estimation is achieved through experience and simplification of uncertain segments to reduce their general unevenness. Similarly to render mode, it can be overlaid with the realty-based

model to show today’s extant surfaces. It is important to note that not all of the surfaces preserved today necessarily originated from the construction phase of the monument. It is precisely in this case study that medieval use and modern restorations are proven. Furthermore, it is not advisable to compare or connect the level of details of a reality- based and an evidence-based reconstruction. In conclusion, not all issues regarding the level of detail can be solved. The level of detail and especially the documentation are eminently suitable for future research. In particular, the graph database offers a unique approach.