8 Points to Improve the Accuracy of 3D Data Creation for Cultural Heritage

Creating 3D data for cultural heritage is not merely a task of preserving a three-dimensional appearance. It is work to prepare foundational information that supports a wide range of purposes such as preservation, restoration, research, public presentation, education, and disaster preparedness. Therefore, producing a visually pleasing model alone is insufficient; what matters is how much accuracy and reproducibility useful for downstream processes can be ensured.

In practical settings, the size and material of the object, installation environment, access conditions, available working time, and the intended use of deliverables vary greatly from project to project. Moreover, cultural heritage often cannot be redone easily, and omissions in site capture, inconsistencies in coordinates, and ambiguous handling of missing parts can directly degrade the overall quality of the deliverable. That is why 3D data creation for cultural heritage requires not only reliance on equipment performance but also a perspective that designs the entire workflow—from planning and on-site response to processing and verification—as an integrated sequence.

This article clearly organizes and explains eight points you should keep in mind to improve accuracy in practical 3D data creation for cultural heritage, arranged by process. It is useful both for those considering introducing such workflows and for those already operating them who want to raise quality to the next level.

Table of Contents

• Why 3D Data Creation for Cultural Heritage Is Emphasized

• Point 1 Clarify Purpose and Deliverables at the Start

• Point 2 Understand the Object Condition and Site Conditions in Advance

• Point 3 Design Required Accuracy and Density by Use Case

• Point 4 Establish a Concept for Reference Points and Alignment

• Point 5 Avoid Missing Data and Blind Spots during On-site Acquisition

• Point 6 Improve Shape Reproducibility by Addressing Light and Materials

• Point 7 Standardize Processing Rules to Stabilize Quality

• Point 8 Prepare Data with Post-delivery Use in Mind

• How to Continuously Improve the Accuracy of 3D Data Creation for Cultural Heritage

Why 3D Data Creation for Cultural Heritage Is Emphasized

The attention given to 3D data creation for cultural heritage stems from challenges that conventional photos and drawings cannot fully address. Cultural heritage often combines complex shapes, aging changes, progressing damage, and harsh site conditions, and there are many situations where two-dimensional information alone is insufficient. If the three-dimensional shape is preserved with high reproducibility, it has great value for future comparative verification, monitoring of changes, and consideration of restoration policies.

Moreover, preservation is not complete with recording alone. The time and effort invested become meaningful only when the acquired data connects to tasks such as restoration planning, maintenance management, exhibition, education, and remote sharing. What becomes important here is whether the 3D data are organized in a way that anyone can understand and whether the necessary accuracy is guaranteed. A model that only looks good is difficult to use for dimensional checks or comparative verification and will ultimately remain only a reference resource.

Additionally, work conditions at cultural heritage sites often have many constraints. It is not uncommon to face difficulties such as limited physical contact, restricted scaffolding, exposure to weather outdoors, inability to occupy a site for long periods while on display, and many dark or confined spaces. Therefore, pre-planning and the quality of on-site decision-making influence results more than in general 3D data creation. To improve accuracy, preparing high-performance equipment is less important than devising an acquisition strategy tailored to the object.

What matters for practitioners of 3D data creation for cultural heritage is to regard accuracy not merely as a numeric value but as quality that is usable in downstream processes. How much detail is necessary, is positional information integration required, is the data intended for long-term comparison, should public-facing and archival versions be separated—only by organizing such perspectives does consistency emerge in decisions from on-site acquisition through data processing.

Point 1 Clarify Purpose and Deliverables at the Start

The first step to improving accuracy in 3D data creation for cultural heritage is to avoid ambiguity about the purpose. If this is unclear, you may capture overly detailed data on site unnecessarily, or conversely, miss necessary information, leading to rework. High-quality work does not mean simply increasing the amount of information but achieving acquisition and organization that are neither excessive nor deficient relative to the purpose.

For example, if the primary objective is archival recording, it is important to determine to what level the condition of damaged or deformed parts should be followed, in addition to overall shape consistency. If the data will be used for restoration design, dimensional reliability and local shape reproducibility become priorities. If the focus is on education or exhibition, visual naturalness, lightweighting, and ease of viewing may be more important. Even within 3D data creation for cultural heritage, priorities can vary greatly depending on the purpose.

It is also useful to think of deliverables from the start in multiple layers. In practice, high-density data for on-site archival preservation, intermediate data for internal review, and lightweight models for public use may be required simultaneously. If you try to separate these ad hoc later, you are likely to leave insufficient information and complicate original data management. By organizing at the initial stage who will use the data, for what, and in what format, the scope of information to be acquired becomes clear.

Aligning recognition among stakeholders is also important. Managers of cultural heritage, investigators, conservators, and public relations personnel have different expectations for the same 3D data. Proceeding without reconciling these differences often leads to problems after delivery such as missing areas of interest, insufficient scale references, or weak linkage to spatial information. Improving accuracy, therefore, also means increasing the degree to which the data meet the requirements.

Accordingly, in the early stages of a project, it is effective to clearly document the target range, required level of reproduction, whether dimensions will be used, the need for positional information, deliverable formats, and anticipated use cases. This clarity makes it easier to judge what to prioritize on site and where to invest time, ultimately leading to higher-quality 3D data creation for cultural heritage.

Point 2 Understand the Object Condition and Site Conditions in Advance

One major factor that affects accuracy is prior understanding of the object's condition and site conditions. Each cultural heritage item is highly individual, and the same acquisition method does not always apply. Entering a site without understanding material, surface color, reflectivity, the fineness of relief, installation height, surrounding obstacles, lighting conditions, and movement restrictions can result in failing to capture the necessary data.

For example, objects with smooth surfaces and few distinctive points tend to be unstable in alignment. Highly reflective materials or wet surfaces tend to produce variability in shape acquisition. Dark locations increase blur and noise, and outdoors, direct sunlight and moving shadows have an impact. Even fences, glass, display cases, or vegetation around an object can cause blind spots or false captures. Because such conditions are often only discovered upon arrival, the thoroughness of pre-checks is important.

A useful approach is to identify in advance the faces or parts likely to present problems, rather than viewing the object as a single volume. The front may be easy to capture, but top, back, base, undersides of projections, and deep carved areas are prone to missing data. In cultural heritage, important traces or workmanship features may remain precisely in these areas. Even if the whole object is captured, missing data in the critical evaluation area greatly reduces usability. To improve accuracy, you need a perspective not simply of uniformly capturing the whole but of reliably recording key parts.

Additionally, confirming working time and accessible areas is essential. In public facilities, consideration for visitors is necessary; outdoors, you must anticipate sudden weather changes. On sites where work must be finished quickly, decisions about what to prioritize are crucial. The clearer your prior understanding, the more straightforward these priorities become.

Collecting information such as site photos, floor plans, existing drawings, and interviews with managers during pre-checks reduces wasted on-site actions. Improving accuracy in 3D data creation for cultural heritage does not mean reducing on-site decision-making but raising the quality of necessary decisions through preparation. Knowing conditions in advance allows you to design acquisition methods and supplementary records consistently.

Point 3 Design Required Accuracy and Density by Use Case

In 3D data creation for cultural heritage, the term "accuracy" can easily be overemphasized; in practice, you must separate accuracy and density. Accuracy relates to correctness of position and shape, while density relates to how finely something is represented. High point or surface density does not necessarily make data practical. Conversely, if necessary density is ensured where required, you can produce high-quality deliverables without making the entire dataset excessively heavy.

For example, if the purpose is to understand overall placement and external form, data design prioritizing overall consistency is appropriate. If investigation of surface damage, chipping, wear, or fine decorations is needed, local high-density capture is important. If this differentiation is not made on site, you may end up with high density in unnecessary areas and insufficient detail where it is needed—one of the most common practical mistakes.

It is also important not to manage accuracy goals solely by numbers. You need evaluation axes suited to use cases, such as allowable error ranges for overall dimensions, the reproduction level required to read fine details, and the stability of alignment for future re-measurement comparisons. Because 3D data for cultural heritage are often read, compared, and used for judgment by people, relying on a single numerical metric alone risks overlooking practical deficiencies.

Moreover, consider downstream processing load at the acquisition stage. Increasing density may seem reassuring, but overly heavy data makes editing and sharing difficult. Especially when multiple people within an organization handle data, performance issues and differences in viewing environments reduce operational efficiency. Achieving accuracy must be balanced with producing data that are operable in practice.

Therefore, for each project it is effective to assume layers such as overall, focused areas, and public versions, and design which level of information to retain where. Designing required accuracy and density by purpose clarifies on-site decisions and stabilizes post-processing policy. Improving accuracy in cultural heritage 3D data creation is not about making everything as fine as possible but about providing an amount of information rationally designed for the purpose.

Point 4 Establish a Concept for Reference Points and Alignment

Often overlooked but greatly influencing accuracy in cultural heritage 3D data creation is the concept of reference points and alignment. No matter how clean the acquired data may look, unstable positional relationships hinder dimensional checks, comparative verification, and overlays with other data. When acquisition is done in multiple sessions or from different viewpoints or dates, alignment design becomes the foundation of the entire outcome.

What is important here is maintaining consistency in how reference points are set for each site. Ideal placements may not be possible due to the surrounding environment or access constraints, but the role of the reference must still be clarified. Whether the reference is intended to stabilize the overall shape, to verify dimensions of a specific part, or to enable future remeasurement comparisons changes how you should place and retain it.

Also, alignment is not only a processing-stage task. On site, the order, directions, and amount of overlap in acquisition already affect alignment quality. Insufficient overlap can lead processing software to force alignment, resulting in local distortions or shifts. Even if the model looks fine outwardly, section or dimensional checks may reveal differences that become major issues later.

In many cases, it is useful to use reference points or known points to stabilize positional information. Especially for wide areas or outdoor cultural heritage, defining overall position makes it easier to connect with drawings, other survey results, and future additional acquisitions. For preservation management and recording changes over time, there is more value in organizing models as assets with positional information rather than as standalone 3D models.

However, what matters is not that placing reference points automatically increases accuracy, but that their intended use is clearly organized. If the concept of alignment is vague, decisions made on site and during processing may conflict, resulting in a deliverable with low consistency. To improve accuracy, decide before acquisition how to maintain positional reference and connect on-site capture with post-processing under the same concept.

Point 5 Avoid Missing Data and Blind Spots during On-site Acquisition

Missing data and blind spots are major factors that degrade the quality of 3D data creation for cultural heritage. These issues are often hard to notice on site and are discovered only after returning to the office during processing. At sites where revisits are difficult, this can be fatal, so you need a capture plan that avoids creating missing data during acquisition.

First, be aware that directions that are easy for humans to view are not necessarily those least prone to data loss. Even if the overall view looks good from the front, undersides of overhangs, deep grooves, the base, recessed surfaces, etc., are harder to capture. In cultural heritage, important traces or workmanship features often remain in precisely these places. Therefore, do not be satisfied with aesthetically pleasing whole views; make sure to approach and capture areas prone to loss.

Next, consider the acquisition order. If you first capture the whole and then reinforce important parts, you are more likely to secure a minimum deliverable even if time runs out. Conversely, spending too much time on local areas can weaken overall consistency and make integration difficult later. 3D data creation for cultural heritage requires a sense of balancing whole and part, proactively filling likely gaps.

Also, develop the habit of checking on site. Even a simple check right after acquisition to confirm continuity of shape and presence of missing areas, and conducting additional capture if anything looks suspicious, greatly affects final quality. Omitting on-site verification leads to reliance on post-processing to fill gaps, which risks infilling shapes that do not actually exist. Excessive inferential filling is to be avoided for cultural heritage; the principle is to obtain information from the object as much as possible.

Furthermore, do not overlook the influence of the surrounding environment. Vegetation, display fixtures, reflective surfaces, passersby, and changing shadows not only increase effort to remove or shape them in post-processing but can also mislead recognition of the target shape. Often small adjustments to your position or capture sequence on site can greatly reduce the impact of unwanted elements. Preventing missing data is not simply about capturing a lot but about planning captures with necessary angles and overlaps.

Point 6 Improve Shape Reproducibility by Addressing Light and Materials

You cannot ignore the effects of light and material in 3D data creation for cultural heritage. No matter how well procedures are organized, unstable surface appearance causes variability in acquisition results. Outdoor cultural heritage is particularly affected by direct sunlight, strong shadows, backlighting, and changes in brightness over time; indoors, uneven lighting and reflections also cause accuracy degradation.

For example, glossy or wet surfaces change appearance with viewpoint and destabilize shape recognition. Nearby glass or metallic reflective surfaces can cause unwanted reflections or misrecognition. Since you often cannot alter the cultural property itself, you must focus on adjusting acquisition conditions rather than changing the object. Simple measures such as adjusting the time of day, avoiding strong direct sunlight, or choosing timing when uniform brightness is available are low-profile but highly effective.

Also, different materials call for attention to different points. Objects with fine textures like stone or wood are easier to capture in shape but require care to distinguish wear and dirt. Relatively uniform surfaces like painted finishes or plaster have few features and may produce unstable alignment. In practical 3D data creation for cultural heritage, understanding material differences and predicting in advance which faces will be easy or unstable to capture contributes to improved accuracy.

Light conditions affect not only appearance but also processing stability. If shooting times and acquisition timing vary, even the same surface may show significant differences in color and brightness, resulting in an unnatural appearance after integration. Even when shape is the primary archival focus, surface information often aids reading, so capturing under as consistent conditions as possible is desirable.

In cultural heritage work, attention to light and material is not an aesthetic consideration but a basic measure to protect recording quality. To improve shape reproducibility, aim to secure a stable appearance at acquisition rather than relying on post-processing. If you want higher accuracy, observe how the object looks and adjust acquisition methods to match that state, not just tweak equipment settings.

Point 7 Standardize Processing Rules to Stabilize Quality

Even if on-site acquisition goes well, ambiguous rules in processing can destabilize quality. If each person has different judgment criteria, variations appear in noise removal, allowable alignment tolerances, treatment of hole filling, and approaches to lightweighting, making it hard to compare deliverables even within the same organization. Improving accuracy means not only enhancing individual project outcomes but also ensuring reproducibility across continuous operations.

First, be clear about what to keep as raw data and what to treat as edited data. In cultural heritage projects, you may later wish to reexamine data from a different perspective. Information judged unnecessary in initial processing can later prove useful. Therefore, preserve pre-processed information and maintain traceability of what was changed at each processing step.

Next, unify policies for dealing with noise and missing data. For example, decide how far to remove surrounding unwanted elements, how much to smooth surface irregularities, and under what conditions to perform hole filling. These choices affect both appearance and measurement fidelity. Excessive shaping may look good but produce data that diverge from the real object. Conversely, leaving data unclean reduces usability and legibility. Rather than relying on intuition for each case, formalize these balances into internal or project-specific standards according to intended uses.

Also, do not neglect verification steps. Beyond an overall visual impression, checking sections, dimensions, joints, shapes of focused areas, and consistency with surroundings from multiple perspectives reveals shifts and distortions that are not visually obvious. In cultural heritage, even slight differences can influence later judgments, so standardizing verification items stabilizes quality.

Standardizing processing rules may seem restrictive, but in practice it reduces time spent deliberating and minimizes quality variation across projects. If you intend to cultivate 3D data creation for cultural heritage as an organizational record asset rather than a one-off deliverable, standardizing processing workflows is essential. Improving accuracy requires not only strong on-site capability but also mechanisms that support consistent processing.

Point 8 Prepare Data with Post-delivery Use in Mind

The ultimate value of 3D data creation for cultural heritage is not in producing the data but in its subsequent use. No matter how accurate the data are, if file organization is confusing, coordinate and scale information is unclear, or records of capture date and target scope are insufficient, the asset becomes hard to use later. From an accuracy perspective, it is important to prepare not only the data itself but also its operability.

First, remember that future users may not be the original creators. Someone months or years later may refer to the data, so it should be clear what was captured, how, and how processing was applied. Since cultural heritage is a long-term preservation subject, 3D data should be organized for long-term operation, not only short-term use.

To do this, leave accompanying information such as the object name, capture date, capture range, positional references, types of deliverables, differences between lightweight and archival versions, and notes. This makes reproducing conditions for future measurements or comparative verification easier. Linking drawings, photos, positional information, and reports to the 3D data broadens potential uses beyond a standalone model.

Also, separate public and archival versions with different requirements in mind. For public-facing purposes, lightweight, easily viewable deliverables are suitable, while high-density data for internal verification have different needs. Confusing these often yields mediocre results for both. Organizing by use from the start reduces post-delivery rework and improves practical convenience.

If you plan for future additional acquisitions or change monitoring, organizing positional information is especially effective. Preservation management of cultural heritage is not only about preserving the shape at one point in time but also about tracking changes over time. If positional references are clear, comparing datasets from different times becomes easier. High-accuracy 3D data creation for cultural heritage is not about creating a beautiful model that ends at that moment but about leaving a record asset that can withstand future decision-making.

How to Continuously Improve the Accuracy of 3D Data Creation for Cultural Heritage

What the eight points above have in common is that the accuracy of 3D data creation for cultural heritage is not determined solely on site. Only when setting objectives, pre-understanding conditions, designing acquisition, aligning positions, conducting on-site verification, considering light and materials, standardizing processing rules, and planning for post-delivery operation are connected does a practically useful high-quality deliverable emerge.

In real-world practice, not everything proceeds ideally. There are constraints on time and personnel, and access to objects can be difficult. That is why it is important to clarify what to prioritize within limited conditions. Improving accuracy means not striving for perfection every time but anticipating and eliminating points prone to failure according to the conditions of each project.

For practitioners, it is particularly important to build a system that standardizes overall flow while flexibly responding to individual project conditions. By accumulating checklists and decision criteria rather than starting from scratch each time, the quality of on-site response steadily improves. As a result, reproducibility of data increases and the overall quality of work—including preservation, restoration, and public utilization—rises.

Also, in 3D data creation for cultural heritage, connections with positional information are increasingly important. For understanding the whole site, organizing relationships with the surrounding environment, and planning future remeasurements for comparison, being able to handle 3D data linked to spatial information is a major advantage. If you want to operate this efficiently, adopting methods that make it easy to obtain high-precision positional information on site is effective.

For instance, when you want to streamline recording of the surroundings, simple positioning, or recording of related equipment and survey points, using an iPhone-mounted GNSS high-precision positioning device such as LRTK can help organize on-site information more easily. Combining such means with 3D data creation itself is a practical option for practitioners who want to strengthen positional referencing and connection with surrounding information, thereby increasing the usability of cultural heritage records.