The technical progress regarding the integration of GIS and BIM and HBIM models is based on 3D representation and interoperability. In what follows, aspects related to file formats, 3D model geometry and its semantics, data, and the internet of things are developed.

In GIS and in BIM and HBIM environments there are many formats for storing 3D geometry. Among others, the formats proposed by European Directive 2007/2/CE for the Infrastructure for Spatial Information in Europe (INSPIRE) are available [4,24], namely gbXML, Open Geospatial Consortium (OGC) [36,37], LandInfra, and IFC. Among these, the most recognized and widely used open standard in GIS is the one issued by the OGC, the “City Geography Markup Language (CityGML)”; in BIM, they are IFC formats [1].

The CityGML format is an open, standardized geometry model based on XML [4,37,38]. This format is still suitable for GIS and BIM integration because of its data interchangeability [39]. It is the most widely used international standard for storing and exchanging three-dimensional city models with semantics [23,40,41,42] in the geospatial domain [24]. The CityGML core module defines the basic concepts and components of the data model; therefore, it is unique and must be implemented by any system.

The IFC standard has been developed by building smart [4] as an open international standard for BIM [40]. It is a standard and interoperable format that is object oriented and capable of representing objects semantically [10]. It serves as an exchange format between different platforms, allowing BIM models to preserve all the details that are integrated in that model [1,41].

There are still many problems and technical barriers related to integration; the fundamental one is the recognition of the nature that characterizes a project when we try to link a BIM model (IFC) and a GIS model (cityGML), causing loss of information. The reality is that an IFC file, by itself, does not contain all the information of the model from which it was extracted, and, additionally, there is a difference in the nature of the BIM and GIS models; that is, a BIM model is structured with geometric figures whose representation depends directly on parameters (width, length, thickness, texture, etc.)—a quite light model; on the other hand, the GIS model is made up of meshes (junctions of points/triangulations) that, although quite flexible, have the disadvantage that a triangulation represents more than one element of the model, which makes the individualized treatment of the characteristics of an element impossible. Additionally, the file is weighty because of the amount of information that needs to be managed to generate the mesh. It is therefore necessary to continue working on an intermediate mechanism between the two types of models to achieve an integration that enables both models to interact under a nature common to both.

In the GIS and BIM and HBIM integration, the geometry of the model is directly defined with its semantics. The semantics refers to the levels with which the 3D model is represented in the different preforms. These levels are parameters to measure the degree of semantics of the objects. They are divided into LOD (levels of detail)—more often referred to as “LoD” with lowercase “o”—for a GIS system, LOD (levels of development) for a BIM element, and LOK (levels of knowledge) in HBIM. The latter arise from the fact that authors wish to define levels of detail applicable to the management and conservation of built heritage [16].

The LoDs (from GIS) are developed in five levels of detail, from LoD0 to LoD4 (Figure 9), having different precisions and minimum dimensions, which are used to represent objects in the model of a three-dimensional city (El-Mekawy et al., 2012; Fan et al., 2011). Thus, LoD0 represents a terrain region in 2.5D; LoD1 are simple volumetric model representations, that is, “boxes”; LoD2 add the roof structure (flat or sloping) to the previous one; LoD3 present the architectural details on the exterior of the model, such as openings and wall textures; and, finally, LoD4 includes the representation of details of the interior of the model, such as the partitions and the delimitation of different spaces [41]. LoD3 and LoD4 levels containing architectural details such as balconies, windows, and rooms rarely exist because, unlike LODs (from BIM), their modeling requires multiple datasets that must be acquired with different technologies, and, often, this requires a lot of manual work [42]; hence, today, most buildings on an urban scale are represented, at best, in LoD3 [43].

Inappropriately, BIM LODs are often interpreted as being associated with a level of detail rather than a level of development. As a project goes through different phases, its semantics increase at different levels of development [40] classified into five groups, from LOD100 to LOD500 [16] Popovic et al., 2017) (Figure 10). LOD100, LOD200, and LOD300 refer, respectively, to the conceptual, schematic, and detailed designs, while the LOD400 and LOD500 refer to a level of development associated with the complex documentation of the project, reaching the final character of an as-built [40].

LOK knowledge levels represent the semantics of heritage management [16], classifying from LOK100 to LOK500 (Figure 8). LOK100 is associated with the identification of the heritage asset and its basic characterization; LOK200 enables the graphic characterization and sufficient information for the development of actions related to the legal protection of the asset and its strategic planning; LOK300 provides greater detail about the characterization of graphic entities to the point of being able to show the results of specialized investigations carried out using archaeological methodology or other specific disciplinary follow-up and diagnosis studies; LOK400 includes specific conservation and intervention actions on the asset’s elements; and finally LOK500 deals with efficient management of HBIM models.

The collaboration between various stakeholders involved in a project consists of sharing data through interaction, communication, exchange, and coordination [9]. Feeding a model with existing data enables not only better visualization but also coordination between views and efficient construction management with considerable cost reduction, whether in the construction, rehabilitation, operation, or maintenance phase.

Today, the most widespread dimensions of a BIM model range from 3D to 7D. 3D represents the three-dimensional model of the project, 4D includes the information about its time sequence [44], 5D refers to the costs of the model elements, 6D contains information on sustainability, and, finally, 7D includes aspects of the management programs in the operation and maintenance phase [16].

As for the HBIM models, and with the objective of coordinating all existing information, another five dimensions are usually adopted, coinciding in name with those referred to for BIM models, 3D–7D, but with somewhat different concepts. Thus, the 3D HBIM model, in addition to being related to the three-dimensional model, considers the data collection performed on the building. 4D is related to historical evolution. 5D cannot be directly related to the actual construction costs as in BIM, since, obviously, the building is already constructed; therefore, the transfer of this dimension from BIM to HBIM is not direct, and a parallelism is usually established with the estimated cost of the associated intervention process [16]. 6D includes the cultural context, and, finally, 7D addresses preventive programs and conservation of the building. Figure 11 illustrates the relationship between the BIM and HBIM dimensions.

Smart city has been a well-adopted concept in urban development worldwide [45,46,47], being, by analogy, the “motherboard” where smart buildings should be inserted, generating a new public–private relationship [39]. It encompasses different definitions, but all of them share, as a basic pillar, the use of technology [27], constituting a facilitating element in the improvement of public services, sustainability, and efficiency [9]. Smart city 3D [37,48], part of the digital twin concept, which was introduced in 2003 within a manufacturing concept and life-cycle management [44,49]. Digital twins integrate IoT, machine learning, artificial intelligence, and big data analysis to create digital simulation and feedback models, which interact with their physical counterparts, updating themselves [50]. Figure 12 lists several application cases and technologies developed in the field of digital twins of cities, extracted from the bibliographic consultation carried out.