Practice Log 4: Matterport Scan of the Calderdale Industrial Museum

Date: 01/10/2023-01/09/2024

Matterport Scan Showcase

---

Reflective Diary

When I first considered creating a digital twin of the Calderdale Industrial Museum (CIM), I chose to use Matterport for three reasons: firstly, it is easy to use; secondly, it has widespread adoption across UK universities and thirdly, is the “most popular for virtual tours” (Sulaiman et al., 2020, pp.222-223) with extensive online support. However, this technical convenience soon revealed itself as a double-edged sword. While the workflow instruction was clear and did not require any knowledge/technique, it was heavily dependent on algorithmic decisions, which overrode human agency. This exposed a deeper tension between accessibility and authorial control in cultural digitisation with platforms.

Even though I had some experience with 3D scanning apps, this was my first attempt to scale a digital twin across four floors of an active heritage site, and the process was totally different from 3D scanning in two ways: firstly, I did not have much control over the whole data collection and generation process; secondly, I did not have the choice to edit the result (I meant to cut or adjust the finished model) but can only redo the whole scan of the camera points and ask the algorithm to regenerate the model. Therefore, to ensure a successful scan, I tested three different hardware setups: the iPhone 13 Pro’s built-in lens and LiDAR scanner, the Matterport motorised mount, and an external Ricoh Theta 360-degree 4K camera. By creating separate models with each of the three, I aimed to explore how different data capture modes affected the continuity and navigability of the final model. Similar to the previous practices, this multi-hardware method was not only a technical experiment but also a curatorial trial to explore if these tools, originally designed for real estate and architecture, could be repurposed to serve a museum’s narrative needs.

When scanning with the iPhone 13 Pro’s native camera and LiDAR scanner, I found that a handheld device could not record a perfect space. The blank spaces at the top and bottom of the 360-degree digital environment are relatively large, and recording in smaller environments filters out visual content from the overhead and floor spaces (although this content can be seen from the next scan point, the top and bottom of the new point remain blank). From a time-cost perspective, the manual scanning process was extremely time-consuming, as I needed to align with the camera angles prompted by the system (“complete scan mode” divides the 360-degree environment into upper and lower tiers, then requires rotating to take six photos for each tier, for a total of eighteen photos). Furthermore, the alignment of the scanned environment between different scan points was not accurate. When the misalignment happened, I tried to use the manual alignment option provided by Matterport, but the system rejected my decisions of realignment. This could be because the raw data from the scan was not high quality, but the system did not tell me the exact reason for the rejection. However, this situation happened with all the hardware I have tested, including Matterport Pro 2, one of the high-end proprietary devices. In the vast majority of cases, I had to rescan the unalignable environment, and it also lacked the capability to scan in open environments. Subsequently, I borrowed the motorised mount designed by Matterport for users scanning with phone cameras from the artist Dave Lynch, with whom I had previously collaborated. Although the addition of the gimbal sped up the process, the accuracy of the environment scanned via the camera could not be improved by it, and the frequency of alignment failures did not decrease.

To determine if this was a device issue or a Matterport system issue, I switched to a higher-precision external Ricoh Theta 360-degree 4K camera to replace the phone camera and conducted initial scans of all four floors of the museum. This camera reduced the time cost because its lens is designed for 360-degree photography, and does not require layered shooting. However, due to the complexity of the spatial layout of the museum,  the floors were misidentified, and scan points were misaligned. Meanwhile, the algorithm did not trust my decision to manually realign the 2D floor plan. This could be because the recorded environment was not detectable by the algorithm, and this issue also appeared when scanning open environments. However, that is not just a software issue; it could also be because this camera does not make use of LiDAR to detect depth for better alignment. Therefore, the scan results from this camera still could not achieve good alignment between different spaces. 

The alignment I refer to here means that, with the positions of two staircases established, I could only use one piece of equipment to scan from one staircase to the other, and when I got close to the other staircase, the 2D floor plan of areas I had already scanned moved away from where the second staircase is. The misalignment can be caused by the uneven wall thickness, which can be seen in the 2D floor plan. It is very visible, but, again, due to the system’s distrust of my manual adjustments, I could not correct the extremely obvious errors by myself. To address these issues, I spent over five hours communicating back and forth with Matterport’s support system, and the result was that I could only rescan to fix this issue. For a platform marketed as “user-friendly” for non-professionals, this is hardly a minor inconvenience.

These problems that emerged were not just technical; they reflected an epistemological hierarchy embedded in the software: the algorithm’s reading of space was prioritised over the practitioner’s interpretation. How is this not a form of hegemony? Moreover, I think the ideal state of a digital twin is one where when a physical environment/object is changed, the virtual model can be automatically updated. However, for now, the digital space is just a snapshot of a moment of the physical one, and every single change in the offline environment needs to be manually replicated in the model creation process, with the risk of error and failure. If this platform/tool/algorithm becomes too powerful and rigid, operating entirely according to preset programs while ignoring the intuition and judgment of an experienced curator (or any professional with judgment, like myself) about specific site conditions (such as correcting errors), then the original value and potential of the digital twin are greatly diminished, or even lost.

In July 2025, I had the opportunity to experience and test the Matterport Pro 2 camera. Compared to the more accessible equipment I had previously tested, this camera did offer some improvement in clarity and a significant improvement in model alignment, and it saved about two-thirds of the recording and post-adjustment time. However, it still lacked the capability to scan in open environments. Moreover, if the only goal is to fix the alignment and clarity issues, one needs to spend £3,165 to purchase this camera. This is nearly ten times the price of the equipment I previously used, but with no change in the exhibition features. So, do we really just want a guided tour that can be produced quickly, is slightly clearer than what standard equipment can record, but still lacks significant interactivity? (This point is explored in depth in Practice Log 10.) And again, how is this not a form of hegemony? 

The research of Hu et al. (2021) can corroborate my findings. They point out that “(1) The majority of existing DT models only involve one-way data transfer from physical entities to virtual models and (2) There is a lack of consideration of the environmental coupling, which results in the inaccurate representation of the virtual components in existing DT models” (p.1). In addition, they also mention, “Due to the variability, uncertainty and fuzziness of physical space, building models in virtual space to mirror entities with high fidelity is a fundamental issue […] The current DT lacks association with external environment. The mechanism explaining how physical object interacts with environment has not been fully embodied in the virtual models” (p.26).

There were more challenges that emerged during the post-scan integration process. Matterport was designed with real estate as the initial target consumers, which made the additional features inadequate for museum storytelling. Although “notes” and “guided tours” allow multimedia embedding, the lack of embedded environmental audio playback, limited clarity on mobile interfaces, and limited in-platform playback highlighted a low degree of compatibility, or even a fundamental mismatch between the current Matterport’s origins and curatorial needs. In that case, why has this software become so popular for presenting online museums? From the participant’s perspective, what is the impact of a Matterport online museum, or more specifically, a virtual tour online museum? I carried on exploring this in practice 10.

While Matterport’s platform is popular in the digital heritage landscape, I also explored Realsee. It is a Chinese-developed virtual exhibition platform. I learned about Realsee through a couple who came to the UK from China, dedicated to helping spread Eastern and Western culture through digital technology. During the final stage of my Matterport testing at CIM, they came to CIM and offered to create a virtual museum for the institution for free (this virtual tour is available now on the museum’s website, and was opened to the public in early 2025). Compared to Matterport, which I used, Realsee offered more interactivity and different functionalities, which sparked my desire to explore more. This comparison was not purely technical because it came from my desire to resist the subtle technological centralisation of Western virtuality. At the same time, I am not trying to sell Realsee to GLAM factors. I wanted to ask whether platforms built in different socio-economic systems offered distinct user logics. Although Realsee, like Matterport, started with real estate agents as its target clients and thus has fewer museum-focused customisations, it excelled in the reasonableness of its pricing and terms of use, the flexibility of its mobile interface, the interactivity offered by its features, and its trust in the model creator’s judgment. This raised a larger question: do our tools inherit the assumptions of the systems that birthed them? In heritage digitisation, even choices like “which platform to use” seemed neutral but actually encoded cultural politics.

I interviewed the owner of the company (interviewee A) mentioned in the previous paragraph, as interviewee A had used both Western (Matterport) and Eastern (Realsee) to produce virtual tour products, which could be beneficial to know more from a culture producer/professional perspective. The interview began with a discussion on technology choice, then dug into the commercial logic of Matterport and Realsee, virtual tour market observation from Eastern and Western perspectives in the context of cultural heritage, and the macro level of cultural exchange. Interviewee A mentioned that the initial reason (the decision made “many years ago”) of choosing Realsea was not because of a technical reason, but the communication with Realsee was more fluent and concerns regarding Matterport’s terms pertaining to data authorisation and copyright. Compared with Matterport, A stated that Realsee focused more on the iterative updates and software development of its own products, and its business model is for operating independently rather than building an extensive third-party ecosystem like Matterport. However, from my perspective, whilst this ecosystem may boost the development of other small companies, it could also lock them into the ecosystem, which benefits Matterport more than those companies. As a user but not a company owner, I think that when a technical platform provides its own equipment, which is expensive but effective for their own system, it also acts as a way to lock in users. However, at the end, as interviewee A commented, to choose which platform or tool to use is totally up to the user.

In summary, I found that Matterport’s scanning process was intuitive, the interface required no formal training, and the capture steps were clearly guided, but working within an overwhelmed, multi-level museum space revealed the platform’s spatial fragility. These moments exposed the embodied labour of scanning, often erased in polished virtual tours, and introduced expensive products/tools subtly. In other words, even though platforms offer users to use their own low-cost device to create the virtual tour, it takes a lot more effort to amend and still will not achieve the same result as the proprietary tools sold by the platform at higher prices. It is how business works. Additionally, I realised that while digital twins offer visual legibility, a routine was set: take a scan every two steps, hide from the lens, and repeat imperfect passes until it works. In fact, the platform’s efficiency was co-produced with my own bodily patience.

Reflective Methodological Note

From this practice, I learned that creating a virtual exhibition is not just about translating a space, but about negotiating between competing logics, like institutional visibility, visitor experience, and platform ideology. My role was hybrid too, as scanner, curator, critic, and translator of the space as well as the model.

Matterport presented an interesting paradox, which lowered the entry barrier, but also enforced a rigid visual logic that constrained museological flexibility. I realised that ease of use and depth of experience often stand in tension with each other. As one CIM volunteer put it, “It’s useful, but if I can’t guide the eyes and ears, I can’t tell the story I want to tell”.

From a sensory perspective, Matterport prioritised vision while underdeveloping the potential for sound and touch in both fidelity and interface design. This reinforced a broader issue in digital museology: despite claims of multimodality, visual centrism remains dominant (Classen, 1997; Marto et al., 2022). Although some virtual exhibitions may seem complete, they are often perceptually fragmented.

Could this be a form of monopoly and digital colonialism? For a fleeting moment, I felt as though I had almost become a sunflower chasing the sun.

When I reviewed this practice, I realised that Matterport/virtual tours do not merely digitise space; they shape the imagination of how the virtual museum can be. It reminded me that platforms are not neutral tools, but cultural interfaces (Manovich, 2002). Designing with these platforms demands ongoing negotiation between technological affordances, heritage requirements, and visitor needs.

---

Reference

Classen, C. 1997. Foundations for an anthropology of the senses. International social science journal. [Online]. 49(153), pp.401-412. [Accessed 12 February 2025]. Available from: https://doi.org/10.1111/j.1468-2451.1997.tb00032.x

Hu, W., Zhang, T., Deng, X., Liu, Z. and Tan, J. 2021. Digital twin: A state-of-the-art review of its enabling technologies, applications and challenges. Journal of Intelligent Manufacturing and Special Equipment. [Online]. 2(1), pp.1-34. [Accessed 28 April 2025]. Available from: https://doi.org/10.1108/JIMSE-12-2020-010

Lee, H.K., Park, S. and Lee, Y. 2022. A proposal of virtual museum metaverse content for the MZ generation. Digital creativity. [Online]. 33(2), pp.79-95. [Accessed 28 April 2025]. Available from: https://doi.org/10.1080/14626268.2022.2063903

Manovich, L. 2002. The Language of New Media. USA: MIT Press.

Marto, A., Gonçalves, A., Melo, M. and Bessa, M. 2022. A survey of multisensory VR and AR applications for cultural heritage. Computers & Graphics. [Online]. 102, pp.426-440. [Accessed 28 April 2025]. Available from: https://doi.org/10.1016/j.cag.2021.10.001

Sulaiman, M.Z., Aziz, M.N.A., Bakar, M.H.A., Halili, N.A. and Azuddin, M.A. 2020, December. Matterport: virtual tour as a new marketing approach in real estate business during pandemic COVID-19. Social Science, Education and Humanities Research. [Online]. 502, pp. 221-226. [Accessed 6 February 2025]. Available from: https://doi.org/10.2991/assehr.k.201202.079