How should uncertainty be shown in digital heritage reconstructions?
A three-dimensional image of an ancient town can look much more complete and convincing than its surviving ruins. Visual detail does not mean that every represented feature is known with equal confidence. A surveyed wall, a roof proposed through comparison and a hypothetical colour scheme may rely on different kinds of evidence. Showing uncertainty in digital heritage reconstruction is not merely an admission of incomplete research. It supplies the transparency viewers need to evaluate the representation. The image should communicate what it claims, what supports that claim and which alternatives remain possible.
Concept and evidence
1. Explain the purpose Recording the present condition, proposing an evidence-based restoration and hypothetically reconstructing a past appearance are different aims. The London Charter emphasises that users should understand what a visualisation represents. Confusion can arise if a pre-earthquake town model and a survey of present remains are presented without distinction. State the purpose, represented period and intended audience. Viewers can then evaluate whether the image is documentation, a research hypothesis or an educational narrative, rather than infer the status of the work from its visual finish alone. 2. Connect sources to decisions A general bibliography is helpful, but important modelling decisions should also be traceable to their evidence. Wall height may come from measurement, decoration from an excavated fragment and roof form from a comparative building. Notes, layers or explanation panels can communicate these relationships. How a source was translated into a model matters alongside the source’s reliability. An assumption in an earlier drawing can otherwise be repeated in digital form as if established, making uncertainty disappear through successive stages of representation. 3. Make visual codes understandable Colour, transparency, line styles and switchable layers can communicate different levels of knowledge. Explain the code and consider access needs such as colour-vision differences. Users should not have to guess which colour indicates stronger evidence. Dividing everything into only “certain” and “hypothetical” may also be insufficient for some projects. The scale should reflect meaningful distinctions within the research. It should not introduce artificial numerical precision merely to appear scientific. Clear categories allow viewers to understand rather than simply admire the model’s complexity.
Practice and interpretation
4. Keep alternatives visible Several plausible roof forms may fit the surviving evidence. If a model presents only one, explain why it was chosen and acknowledge other possibilities. Comparative views can show why the same evidence sometimes supports different interpretations. Presenting alternatives does not imply that every interpretation is equally strong. A solution supported by measurement differs from one based only on analogy. Retaining a decision history and version information also enables the representation to be revised transparently when further research becomes available. 5. Realism is not evidence Lighting, weather, crowds and sound can make a reconstruction compelling. Some of these details may be narrative choices rather than evidence-based historical claims. If a depicted crowd size or textile colour is not supported by data, production notes should make that clear. Realistic textures can encourage viewers to accept assumptions as facts. Educational design can preserve the relationship between aesthetic effect and evidential claim. When this is done openly, a persuasive image can support critical thinking instead of merely creating confidence in a single past appearance. 6. Questions users can ask Look for the represented date, source links, method statement and uncertainty key. Their absence does not prove that the image is false, but it makes evaluation harder. Institutional research publications or model documentation may provide further detail. A screenshot should not be detached from its explanation and circulated as a definite photograph of the past. Digital reconstruction is valuable not only for making a lost environment visible, but also for opening discussion about what is known, how it is known and what remains debatable.