Color models in the process of 3D digitization of an artwork for presentation in a VR environment of an art gallery †

Citation: Drofova, I.; Adamek, M. Color Models in the Process of 3D Digitization of an Artwork for Presentation in a VR Environment of an Art Gallery. Electronics 2024,13, 4431. https://doi.org/10.3390/ electronics13224431 Academic Editor: Silvia Liberata Ullo Received: 2 October 2024 Revised: 1 November 2024 Accepted: 5 November 2024 Published: 12 November 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Article Color Models in the Process of 3D Digitization of an Artwork for Presentation in a VR Environment of an Art Gallery † Irena Drofova * and Milan Adamek Faculty of Applied Informatics, Tomas Bata University in Zlín, 760 05 Zlín, Czech Republic *Correspondence: dr[email protected] †This paper is an extended version of our paper published in Analysis of Natural Lighting Condition for the Digitization if Artwork in an Art Gallery Interior. In Proceedings of the WSCG 2024 International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision, Computer Science Research Notes, Prague, Czech Republic, 3–6 June 2024; pp. 391–394. https://doi.org/10.24132/CSRN.3401.43. Abstract: This study deals with the color reproduction of a work of art to digitize it into a 3D realistic model. The experiment aims to digitize a work of art for application in a virtual reality environment concerning faithful color reproduction. Photogrammetry and scanning with a LiDAR sensor are used to compare the methods and work with colors during the reconstruction of the 3D model. An innovative tablet with a camera and LiDAR sensor is used for both methods. At the same time, current findings from the field of color vision and colorimetry are applied to 3D reconstruction. The experiment focuses on working with the RGB and L*a*b* color models and, simultaneously, on the sRGB, CIE XYZ, and Rec.2020(HDR) color spaces for transforming colors into a virtual environment. For this purpose, the color is defined in the Hex Color Value format. This experiment is a starting point for further research on color reproduction in the digital environment. This study represents a partial contribution to the much-discussed area of forgeries of works of art in current trends in forensics and forgery. Keywords: color model; image processing; virtual reality; photogrammetry; 3D model; 3D reconstruction; forensic science 1. Introduction Currently, digitization processes are reflected in all areas of human activity. Digital technologies are used across commercial, scientific, and artistic fields. Especially in the field of art, emphasis is often placed on the highly realistic quality of digital reproduction. Digital technologies are already a standard part of capturing and processing images. However, digital technology and image processing processes are projected and part of artistic creation [ 1 ]. Nevertheless, the digital reproduction of works of art is still excellent for applying new procedures, such as machine learning, especially in connection with new trends, such as 3D visualization and virtual presentations in the online environment [2,3]. 3D realistic digital reproduction of a work of art also brings many challenges and unsolved issues in image processing. This also depends on the digitization itself in 2D [ 4 ]. One of them is the high quality of the reproduction of the object in connection with its texture and color, which are often changeable due to the influence of light and weather conditions. This problem is noted especially in exterior exhibitions and architecture [ 5 ]. The technology and methodology chosen for the 3D digitization process also greatly influence the quality of the digitization of the object [ 6 ]. The method and process of digitization depend mainly on the final output of the digitized object. The team can be 2D and 3D printing, 3D online presentations, or using an object in a virtual and augmented reality (VR/AV) environment in interaction with the user [ 7 , 8 ]. For a realistic digital 3D reproduction of a work of art, great emphasis is placed on the quality of the digital image and its realism. Image reproduction aims to obtain as close as possible to the original Electronics 2024,13, 4431. https://doi.org/10.3390/electronics13224431 https://www.mdpi.com/journal/electronics Electronics 2024,13, 4431 2 of 15 image [ 9 ]. The same attribute for evaluating reproduction quality is color, which is directly related to light and human vision. This issue is dealt with by colorimetry, color, and human vision [10–12]. This text responds to current trends in the digitization of art and the issue of realistic digital reproduction using the photogrammetry method and significantly expands the findings presented in a short paper presented at an international conference in Pilsen, Czech Republic [ 13 , 14 ]. The work also considers using LiDAR (Light Detection And Ranging) sensors for rapid image capture using a mobile application in a mobile device [ 15 , 16 ]. The presented experiment aims to determine the extent to which ambient lighting conditions can influence the 3D digitization of a work of art in connection with the chosen modeling method. Previous research in the field of image digitization used sensing devices, such as compact and DSLR (Digital Single-Lens Reflex) cameras, mobile device cameras, laser scanners, and 360◦cameras. One sensing device is used in this experiment [17–20]. The following chapters describe the digitization of the art work using the SfM (Structure from Motion) photogrammetry method and scanning with a LiDAR sensor in the interior. At the same time, image capture is performed indoors in daylight. The influence of lighting conditions on the color reproduction of the 3D model is analyzed in 3D point cloud models for one precisely defined color #758605 (Hex Color Value), color model RGB (117, 134, 5), and CIELAB/L*a*b* (54.30, − 8.46, 3.83). The findings from this experiment will subsequently be used to analyze the color visualization of realistic 3D digital reproduction of artwork types in a VR environment. Current studies on color reproduction in artistic paintings do not focus on direct, realistic color reproduction based on a precisely defined color or a set of colors measured on an object [21–23]. Therefore, working with colors and color models is a suitable solution for the realistic reproduction of works of art in digital and virtual environments. Accurate color reproduction and working with the color light spectrum can be a suitable complementary process for determining the original color in a digital environment. The market for duplicates, not only in the field of art but also in the clothing and industrial sectors, represents a large area for the trade in counterfeits [ 24 , 25 ]. Forensic science is also used here, mainly in digital forensic art, the subject of which is also the 2D and 3D realistic reproduction of objects [ 26 , 27 ]. In the digital reproduction of works of art, the color and texture of the material play important roles. The type and nature of lighting also affect these two attributes. This experiment represents the first basic research in the field of realistic color reproduction in a virtual environment with the aim of application in criminology and forensic investigation, which will complement the current methods applied in forensic sciences. 2. Materials and Methods The development of digital technologies and sensing devices, which gradually replaced the analog method of image processing, image digitization processes, and graphic software for image processing according to the type and purpose of the final output, has also been developed and improved. This includes developing and digitizing print, digital, and 3D printing, as well as 2D and 3D online presentations [ 25 , 28 ]. Currently, virtual and augmented reality (VR/AR) technologies are also being perfected and are now available to the general public, especially in the gaming industry [ 16 , 29 – 31 ]. The following section describes the digitization of the artwork. Sections 2.3 and 2.4 describe image digitization using the SfM (Structure for Motion) photogrammetry method and LiDAR sensor scanning to create a realistic 3D digital model. 2.1. Digitization of Art: Art Painting with Acrylic Paints on Canvas The artistic object for 3D digital reproduction was painted with acrylic paints on the canvas. The artwork is dominated by green and brown acrylic paints, and the tones of these two colors are created by mixing these acrylic paints. As seen in Figure 1a. Also marked in red in this figure is the base color space. In this place, the color value was measured with a Colorcatch NANO colorimeter from the Swiss company Colorix SA Electronics 2024,13, 4431 3 of 15 (Neuchâtel, Switzerland), and the value of the direct green color RGB = 117, 134, 123, and L*a*b = 54.30 , − 8.46, 3.83 [ 32 ]. These values proved sufficient to convert the color value into a Hex = #758605 color model for subsequent color segmentation in a realistic 3D point cloud model of the object. The CIE Lab/L*a*b* color space model is applied for digital imaging across digital and display devices. Electronics 2024, 13, x FOR PEER REVIEW 3 of 16 two colors are created by mixing these acrylic paints. As seen in Figure 1a. Also marked in red in this figure is the base color space. In this place, the color value was measured with a Colorcatch NANO colorimeter from the Swiss company Colorix SA (Neuchâtel, Switzerland), and the value of the direct green color RGB = 117, 134, 123, and L*a*b = 54.30, −8.46, 3.83 [32]. These values proved sufficient to convert the color value into a Hex = #758605 color model for subsequent color segmentation in a realistic 3D point cloud model of the object. The CIE Lab/L*a*b* color space model is applied for digital imaging across digital and display devices. (a) (b) (c) Figure 1. Digitization of an art object: (a) 2D digitized object and detail marked in red; (b) matrix of partial details the yellow range of the image; and (c) visualization of the detail of the structure and color of a partial part of the object. Figure 1a shows a photograph of a digitized object with a marked space in which the value of the direct green color RGB = 117, 134, 123 was measured by the Colorcatch NANO colorimeter. Figure 1b shows a detailed matrix of the color structure. The link matrix is marked in the yellow range of 5.563 µm × 4.171 µm. Figure 1b shows the details of the material structure of the object of the marked element of the matrix shown in Figure 1c. This essential visual inspection of the color of the object in detail of the material structure of the object (acrylic paint on the canvas) aimed to determine the possible influence of the material structure on the generation of unwanted points during the creation of a 3D realistic model using the photogrammetry method. Furthermore, the LiDAR sensor assumed a slight deformation in the structure of the 3D model. A 3D laser scanning microscope Keyence VK-X3000 (Keyence International NV/SA, Mechelen, Belgium) was used to inspect the material structure visually [33]. This device features scanning that provides adaptability to identify the minor surface features of a material. The object was scanned using scanning devices in daylight in the natural environment of the interior of an art gallery, aiming to capture the object’s color in such a way that it is perceived by human vision in this environment. The lighting conditions met the standard D65 (Standard Illuminant) in the interior. Furthermore, the art object was photographed in the dark to evaluate whether it was significantly affected by changes in lighting conditions in the art gallery at dusk. 2.2. Digital Image Capture Digital devices have gradually replaced analog sensing devices. Digital compact cameras and Digital Single-Lens Reflex (DSLR) cameras have gradually been supplemented by 360° cameras and scanners, RGBd cameras, Light Detection And Ranging (LiDAR) sensor technology, and other types of digital devices. Currently, SMART mobile devices, such as mobile phones and tablets, are already commonly used for these purposes, and emphasis is placed on low-cost methods and procedures [15,17,19,20]. An innovative mobile device with LiDAR technology was used in this experiment. An iPad 11″ Pro smart tablet from Apple was used to capture and digitize the artwork. This smart device has a high-quality camera with high resolution and a LiDAR sensor [34]. This smart device was used for the 3D Figure 1. Digitization of an art object: (a) 2D digitized object and detail marked in red; (b) matrix of partial details the yellow range of the image; and (c) visualization of the detail of the structure and color of a partial part of the object. Figure 1a shows a photograph of a digitized object with a marked space in which the value of the direct green color RGB = 117, 134, 123 was measured by the Colorcatch NANO colorimeter. Figure 1b shows a detailed matrix of the color structure. The link matrix is marked in the yellow range of 5.563 µ m × 4.171 µ m. Figure 1b shows the details of the material structure of the object of the marked element of the matrix shown in Figure 1c. This essential visual inspection of the color of the object in detail of the material structure of the object (acrylic paint on the canvas) aimed to determine the possible influence of the material structure on the generation of unwanted points during the creation of a 3D realistic model using the photogrammetry method. Furthermore, the LiDAR sensor assumed a slight deformation in the structure of the 3D model. A 3D laser scanning microscope Keyence VK-X3000 (Keyence International NV/SA, Mechelen, Belgium) was used to inspect the material structure visually [ 33 ]. This device features scanning that provides adaptability to identify the minor surface features of a material. The object was scanned using scanning devices in daylight in the natural environment of the interior of an art gallery, aiming to capture the object’s color in such a way that it is perceived by human vision in this environment. The lighting conditions met the standard D65 (Standard Illuminant) in the interior. Furthermore, the art object was photographed in the dark to evaluate whether it was significantly affected by changes in lighting conditions in the art gallery at dusk. 2.2. Digital Image Capture Digital devices have gradually replaced analog sensing devices. Digital compact cameras and Digital Single-Lens Reflex (DSLR) cameras have gradually been supplemented by 360 ◦ cameras and scanners, RGBd cameras, Light Detection And Ranging (LiDAR) sensor technology, and other types of digital devices. Currently, SMART mobile devices, such as mobile phones and tablets, are already commonly used for these purposes, and emphasis is placed on low-cost methods and procedures [ 15 , 17 , 19 , 20 ]. An innovative mobile device with LiDAR technology was used in this experiment. An iPad 11 ′′ Pro smart tablet from Apple was used to capture and digitize the artwork. This smart device has a high-quality camera with high resolution and a LiDAR sensor [ 34 ]. This smart device was used for the 3D reconstruction of a work of art using ground image photogrammetry. At the same time, the free application Scaniverse from Niantic Labs was used to compare the Electronics 2024,13, 4431 4 of 15 quality of the 3D model, which was intended directly for the 3D digitization of objects and spaces using the LiDAR sensor [ 35 ]. Both methods are described in the following section. The professional 3D modeling SW Agisoft Metashape Professional (online: agisoft.cz, 2021) from the company Agisoft (St. Petersburg, Russia) was used for the 3D reconstruction of the artwork and analysis of the color 3D reproduction [36]. 2.3. 3D Reconstruction by Photogrammetry Method The SfM (Structure from Motion) photogrammetry method calculates the location of an object in 3D space based on the description of information obtained from individual images taken from multiple angles. In the case of a specific object, the 3D reconstruction described below involves 250 photos, where the algorithm based on the principle of triangulation finds standard bodies in individual photos and calculates the individual position of the camera around the object. By subsequently calculating the Dense Cloud (cloud of points), each point obtains its own x, y, and z coordinates and thus defines basic information about the position, size, and geometry of the object located in space. Figure 2shows the principle of the photogrammetry method [14,37]. I’+l1X+l2Y+l3Z+l4 l9X+l10Y+l11Z+1=0∪J′+l5X+l6Y+l7Z+l8 l9X+l10Y+l11Z+1=0 (1) where I′=I−I0 ; J′=J−J0 ; l1−l11 is the Direct Linear Transformation Parameter (DLTP). The coefficients l1 to l11 are functions of exterior landmarks and interior landmarks. The initial values of the external and internal orientation elements are unnecessary in the calculation. The DLT equation can be used in the photogrammetry of consumer-class digital cameras [ 19 , 37 ]. The basic generated point cloud of the digitized object is shown in Figure 3. In this case, the basic 3D point cloud is created from 24 images, from which 13,828 points were generated. Electronics 2024, 13, x FOR PEER REVIEW 4 of 16 reconstruction of a work of art using ground image photogrammetry. At the same time, the free application Scaniverse from Niantic Labs was used to compare the quality of the 3D model, which was intended directly for the 3D digitization of objects and spaces using the LiDAR sensor [35]. Both methods are described in the following section. The professional 3D modeling SW Agisoft Metashape Professional (online: agisoft.cz, 2021) from the company Agisoft (St. Petersburg, Russia) was used for the 3D reconstruction of the artwork and analysis of the color 3D reproduction [36]. 2.3. 3D Reconstruction by Photogrammetry Method The SfM (Structure from Motion) photogrammetry method calculates the location of an object in 3D space based on the description of information obtained from individual images taken from multiple angles. In the case of a specific object, the 3D reconstruction described below involves 250 photos, where the algorithm based on the principle of triangulation finds standard bodies in individual photos and calculates the individual position of the camera around the object. By subsequently calculating the Dense Cloud (cloud of points), each point obtains its own x, y, and z coordinates and thus defines basic information about the position, size, and geometry of the object located in space. Figure 2 shows the principle of the photogrammetry method [14,37]. Figure 2. The basic principle of the Structure from Motion (SfM) method [37]. I’  lXl Yl Zl  lXl Yl Z10 ∪ J´ lXl Yl Zl  lXl Yl Z10 (1) where I´  I  I; J´JJ ; ll  is the Direct Linear Transformation Parameter (DLTP). The coefficients l to l are functions of exterior landmarks and interior landmarks. The initial values of the external and internal orientation elements are unnecessary in the calculation. The DLT equation can be used in the photogrammetry of consumer-class digital cameras [19,37]. The basic generated point cloud of the digitized object is shown in Figure 3. In this case, the basic 3D point cloud is created from 24 images, from which 13,828 points were generated. (a) (b) (c) (d) Figure 2. The basic principle of the Structure from Motion (SfM) method [37]. Electronics 2024, 13, x FOR PEER REVIEW 4 of 16 reconstruction of a work of art using ground image photogrammetry. At the same time, the free application Scaniverse from Niantic Labs was used to compare the quality of the 3D model, which was intended directly for the 3D digitization of objects and spaces using the LiDAR sensor [35]. Both methods are described in the following section. The professional 3D modeling SW Agisoft Metashape Professional (online: agisoft.cz, 2021) from the company Agisoft (St. Petersburg, Russia) was used for the 3D reconstruction of the artwork and analysis of the color 3D reproduction [36]. 2.3. 3D Reconstruction by Photogrammetry Method The SfM (Structure from Motion) photogrammetry method calculates the location of an object in 3D space based on the description of information obtained from individual images taken from multiple angles. In the case of a specific object, the 3D reconstruction described below involves 250 photos, where the algorithm based on the principle of triangulation finds standard bodies in individual photos and calculates the individual position of the camera around the object. By subsequently calculating the Dense Cloud (cloud of points), each point obtains its own x, y, and z coordinates and thus defines basic information about the position, size, and geometry of the object located in space. Figure 2 shows the principle of the photogrammetry method [14,37]. Figure 2. The basic principle of the Structure from Motion (SfM) method [37]. I’  lXl Yl Zl  lXl Yl Z10 ∪ J´ lXl Yl Zl  lXl Yl Z10 (1) where I´  I  I; J´JJ ; ll  is the Direct Linear Transformation Parameter (DLTP). The coefficients l to l are functions of exterior landmarks and interior landmarks. The initial values of the external and internal orientation elements are unnecessary in the calculation. The DLT equation can be used in the photogrammetry of consumer-class digital cameras [19,37]. The basic generated point cloud of the digitized object is shown in Figure 3. In this case, the basic 3D point cloud is created from 24 images, from which 13,828 points were generated. (a) (b) (c) (d) Figure 3. Creation of a 3D model using the SfM photogrammetry method: (a) Digitized object; (b) position of 24 photos from which the basic cloud of points is created; (c) Dense Cloud generation; (d) the resulting 3D texture model of the artwork. Electronics 2024,13, 4431 5 of 15 Figure 3a shows a 2D model of the artwork, and Figure 3b shows a 3D point cloud generated from 24 photographs that were used to apply the SfM method. The principle of this method is shown in Figure 2. The generated points provide information about the reconstructed object’s position, geometry, and color. From this basic information in the primary cloud of points, the points that form the Dense Cloud are added by further calculation. This high number of points will more specifically display the shape of the 3D object and its position in space, as shown in Figure 3c. This will create a complete point model, from which it can be determined in which places the calculation did not define points and where it is necessary to add points. The resulting 3D model corresponds to the shape and structure of the physical object in real space, as shown in Figure 3d. Let us add that the generated Dense Cloud 3D model contains 413,688 individual points. SW Agisoft Metashape Professional was used for 3D modeling. In the following section, the method of 3D reconstruction using a smart device with a LiDAR sensor is described and visualized in more detail. 2.4. 3D Reconstruction by a LiDAR Sensor In the case of using a LiDAR sensor and an image processing application, photographs are not used for the primary reconstruction of the point cloud. The finished 3D digital model of the artwork was scanned by the sensor directly in the free Scaniverse application and then converted to 3D graphic 3D SW Agisoft Metashape Professional. A point cloud was generated from this 3D model for subsequent image analysis. This method was chosen to compare the quality of the digital reproduction of a work of art captured by the same capture device, which is an iPad 11′′ Pro tablet. Figure 4shows the 3D reconstruction of the object using the LiDAR sensor. The Scaniverse application was used to create a 3D model. A 3D model of the object is shown in Figure 4b. This textured 3D model was transferred to SW Agisoft to generate a cloud of points. Figure 4c shows the 3D model exported to the Agisoft 3D SW. A Dense Cloud with a similar body 23 was generated in this case. Figure 4d shows the details of the generated 3D points. Figure 5shows the details of the Dense Cloud. Electronics 2024, 13, x FOR PEER REVIEW 5 of 16 Figure 3. Creation of a 3D model using the SfM photogrammetry method: (a) Digitized object; (b) position of 24 photos from which the basic cloud of points is created; (c) Dense Cloud generation; (d) the resulting 3D texture model of the artwork. Figure 3a shows a 2D model of the artwork, and Figure 3b shows a 3D point cloud generated from 24 photographs that were used to apply the SfM method. The principle of this method is shown in Figure 2. The generated points provide information about the reconstructed object’s position, geometry, and color. From this basic information in the primary cloud of points, the points that form the Dense Cloud are added by further calculation. This high number of points will more specifically display the shape of the 3D object and its position in space, as shown in Figure 3c. This will create a complete point model, from which it can be determined in which places the calculation did not define points and where it is necessary to add points. The resulting 3D model corresponds to the shape and structure of the physical object in real space, as shown in Figure 3d. Let us add that the generated Dense Cloud 3D model contains 413,688 individual points. SW Agisoft Metashape Professional was used for 3D modeling. In the following section, the method of 3D reconstruction using a smart device with a LiDAR sensor is described and visualized in more detail. 2.4. 3D Reconstruction by a LiDAR Sensor In the case of using a LiDAR sensor and an image processing application, photographs are not used for the primary reconstruction of the point cloud. The finished 3D digital model of the artwork was scanned by the sensor directly in the free Scaniverse application and then converted to 3D graphic 3D SW Agisoft Metashape Professional. A point cloud was generated from this 3D model for subsequent image analysis. This method was chosen to compare the quality of the digital reproduction of a work of art captured by the same capture device, which is an iPad 11″ Pro tablet. Figure 4 shows the 3D reconstruction of the object using the LiDAR sensor. The Scaniverse application was used to create a 3D model. A 3D model of the object is shown in Figure 4b. This textured 3D model was transferred to SW Agisoft to generate a cloud of points. Figure 4c shows the 3D model exported to the Agisoft 3D SW. A Dense Cloud with a similar body 23 was generated in this case. Figure 4d shows the details of the generated 3D points. Figure 5 shows the details of the Dense Cloud. (a) (b) (c) (d) Figure 4. Creating a 3D model using a LiDAR sensor: (a) digitized object; (b) 3D model generated by Scaniverse; (c) 3D texture model imported into Agisoft 3D SW; and (d) generated point cloud from the textured 3D model. (a) (b) Figure 4. Creating a 3D model using a LiDAR sensor: (a) digitized object; (b) 3D model generated by Scaniverse; (c) 3D texture model imported into Agisoft 3D SW; and (d) generated point cloud from the textured 3D model. Electronics 2024, 13, x FOR PEER REVIEW 5 of 16 Figure 3. Creation of a 3D model using the SfM photogrammetry method: (a) Digitized object; (b) position of 24 photos from which the basic cloud of points is created; (c) Dense Cloud generation; (d) the resulting 3D texture model of the artwork. Figure 3a shows a 2D model of the artwork, and Figure 3b shows a 3D point cloud generated from 24 photographs that were used to apply the SfM method. The principle of this method is shown in Figure 2. The generated points provide information about the reconstructed object’s position, geometry, and color. From this basic information in the primary cloud of points, the points that form the Dense Cloud are added by further calculation. This high number of points will more specifically display the shape of the 3D object and its position in space, as shown in Figure 3c. This will create a complete point model, from which it can be determined in which places the calculation did not define points and where it is necessary to add points. The resulting 3D model corresponds to the shape and structure of the physical object in real space, as shown in Figure 3d. Let us add that the generated Dense Cloud 3D model contains 413,688 individual points. SW Agisoft Metashape Professional was used for 3D modeling. In the following section, the method of 3D reconstruction using a smart device with a LiDAR sensor is described and visualized in more detail. 2.4. 3D Reconstruction by a LiDAR Sensor In the case of using a LiDAR sensor and an image processing application, photographs are not used for the primary reconstruction of the point cloud. The finished 3D digital model of the artwork was scanned by the sensor directly in the free Scaniverse application and then converted to 3D graphic 3D SW Agisoft Metashape Professional. A point cloud was generated from this 3D model for subsequent image analysis. This method was chosen to compare the quality of the digital reproduction of a work of art captured by the same capture device, which is an iPad 11″ Pro tablet. Figure 4 shows the 3D reconstruction of the object using the LiDAR sensor. The Scaniverse application was used to create a 3D model. A 3D model of the object is shown in Figure 4b. This textured 3D model was transferred to SW Agisoft to generate a cloud of points. Figure 4c shows the 3D model exported to the Agisoft 3D SW. A Dense Cloud with a similar body 23 was generated in this case. Figure 4d shows the details of the generated 3D points. Figure 5 shows the details of the Dense Cloud. (a) (b) (c) (d) Figure 4. Creating a 3D model using a LiDAR sensor: (a) digitized object; (b) 3D model generated by Scaniverse; (c) 3D texture model imported into Agisoft 3D SW; and (d) generated point cloud from the textured 3D model. (a) (b) Figure 5. Generated Dense Cloud: (a) 3D SfM photogrammetry method and (b) LiDAR sensor. Electronics 2024,13, 4431 6 of 15 Figure 5shows the details of the digitized artwork’s Dense Cloud structure. Figure 5a shows the details of the Dense Cloud and individual points generated from the photographs and the primary point cloud produced by the SfM photogrammetry method. In total, 413,688 individual points were generated in the 3D model. Figure 5b shows the details of the 23 points generated from the 3D texture model made by the sensor LiDAR. Individual points provide color information that ultimately defines individual points with a color value of #758605. 3. Colorimetry and Color Analysis Colorimetry is the science of color and light. This field deals with the color interpretation of objects and the environment, human vision, and color reproduction. Light is electromagnetic radiation. The spectrum includes visible light (400–700 nm), which is visible to the human eye. The visible color of light is based on the wavelength λ and subjective perception by the human eye. Color image processing is also related to this. For these purposes, color models and gamuts define the possibilities of the color display of individual tones and their maximum range. To unify color reproduction, the international standard CIE 1931 was adopted in 1931, which is based on current modern technology and procedures across manufacturing and scientific fields. 3.1. Color Model and Gamut In this experiment, in which a real object is transformed into a digital form, the RGB (red, green, blue) color model and the sRGB color space (gamut) are used. Figure 6shows their graphical representation. Electronics 2024, 13, x FOR PEER REVIEW 6 of 16 Figure 5. Generated Dense Cloud: (a) 3D SfM photogrammetry method and (b) LiDAR sensor. Figure 5 shows the details of the digitized artwork’s Dense Cloud structure. Figure 5a shows the details of the Dense Cloud and individual points generated from the photographs and the primary point cloud produced by the SfM photogrammetry method. In total, 413,688 individual points were generated in the 3D model. Figure 5b shows the details of the 23 points generated from the 3D texture model made by the sensor LiDAR. Individual points provide color information that ultimately defines individual points with a color value of #758605. 3. Colorimetry and Color Analysis Colorimetry is the science of color and light. This field deals with the color interpretation of objects and the environment, human vision, and color reproduction. Light is electromagnetic radiation. The spectrum includes visible light (400–700 nm), which is visible to the human eye. The visible color of light is based on the wavelength λ and subjective perception by the human eye. Color image processing is also related to this. For these purposes, color models and gamuts define the possibilities of the color display of individual tones and their maximum range. To unify color reproduction, the international standard CIE 1931 was adopted in 1931, which is based on current modern technology and procedures across manufacturing and scientific fields. 3.1. Color Model and Gamut In this experiment, in which a real object is transformed into a digital form, the RGB (red, green, blue) color model and the sRGB color space (gamut) are used. Figure 6 shows their graphical representation. (a) (b) Figure 6. Colorimetry: (a) RGB color model and (b) sRGB color space (gamut). Figure 6a shows the RGB (Red, Green, Blue) color model. This is the basic color model of RGB primary colors. The RGB color model operates using light components. This model primarily targets digital imaging and imaging devices (DSLR displays). The secondary colors created by mixing the primary RGB colors are CMY (Cyan, Magenta, Yellow). Mixing all three essential components of RGB creates white light (W). Figure 6b shows the CIE1931 trichromatic triangle, standardizing work with colors since 1931. The sRGB gamut (color space) is represented in the indicated trichromatic triangle in Figure 6. This space represents the maximum color range in which a digital sensing or imaging device can operate. It currently displays the color spectrum and range of most digital imaging and sensing devices in the sRGB gamut. Therefore, the RGB model was Figure 6. Colorimetry: (a) RGB color model and (b) sRGB color space (gamut). Figure 6a shows the RGB (Red, Green, Blue) color model. This is the basic color model of RGB primary colors. The RGB color model operates using light components. This model primarily targets digital imaging and imaging devices (DSLR displays). The secondary colors created by mixing the primary RGB colors are CMY (Cyan, Magenta, Yellow). Mixing all three essential components of RGB creates white light (W). Figure 6b shows the CIE1931 trichromatic triangle, standardizing work with colors since 1931. The sRGB gamut (color space) is represented in the indicated trichromatic triangle in Figure 6. This space represents the maximum color range in which a digital sensing or imaging device can operate. It currently displays the color spectrum and range of most digital imaging and sensing devices in the sRGB gamut. Therefore, the RGB model was chosen as the default color model for this experiment. The values of the individual color components of light of direct color #758605 (Hex Color Value) are as follows: R = 117; G = 134; B = 5. In connection with Section 3.1. subsequently, in Section 3.3, attention is Electronics 2024,13, 4431 7 of 15 paid to the color space and gamut L*a*b*, which is suitable for subsequent work with color reproduction and display in the VR environment. 3.2. Color Value #758605 Segmentation In Agisoft 3D modeling, the SW environment can work with information in the color of individual points or groups of points in the Dense Cloud in RGB and HSV color models. This can be performed with a precisely defined Hex Color Value, as shown in Figure 7. Electronics 2024, 13, x FOR PEER REVIEW 7 of 16 chosen as the default color model for this experiment. The values of the individual color components of light of direct color #758605 (Hex Color Value) are as follows: R = 117; G = 134; B = 5. In connection with Section 3.1. subsequently, in Section 3.3, attention is paid to the color space and gamut L*a*b*, which is suitable for subsequent work with color reproduction and display in the VR environment. 3.2. Color Value #758605 Segmentation In Agisoft 3D modeling, the SW environment can work with information in the color of individual points or groups of points in the Dense Cloud in RGB and HSV color models. This can be performed with a precisely defined Hex Color Value, as shown in Figure 7. (a) (b) (c) Figure 7. SfM—Points Segmentation #758605: (a) Dense Cloud 3D model using SfM photogrammetry; (b) segmentation points by color G#758605; (c) body #758605 in Dense Cloud. Figure 7 shows the segmentation of points carrying information about color value #758605. Figure 7a shows a 3D Dense Cloud model of individual points generated from photographs and created by the SfM photogrammetric method. Figure 7b visualizes the SW Agisoft environment for working with Dense Cloud and information about the color value carried by each point. The 3D SW works with RGB and HSV (Hue, Saturation, Value) color models in this case. Values can be directly numerically defined in Hex format. However, it should be noted that the RGB color model does not contain tonal or other values such as Hue, Saturation, and Value in the HSV color model. Figure 7c shows an example of the final segmentation of points with a color value of #758605 in the total number of clouding points in the 3D model. Figure 8 shows the process of segmenting the points using color information #758605. Figure 8a shows the 3D texture model captured by the LiDAR sensor in Agisoft 3D SW. A point cloud containing 17 points was generated from this texture model, as shown in Figure 8b. The segmentation process of the point with color information #758605 is identical to that of the SfM method, as shown in Figure 7b. As shown in Figure 8c, no single point with a color value of #758605 was defined out of the total number of points. For this reason, the segmentation of points in the 3D models by the LiDAR sensor is not shown in Section 4.1. (a) (b) (c) Figure 8. LiDAR—Segmentation of points #758605: (a) 3D model using LiDAR sensor; (b) Segmentation of points by color G#758605; (c) detail of the points generated in Dense Cloud. Figure 7. SfM—Points Segmentation #758605: (a) Dense Cloud 3D model using SfM photogrammetry; (b) segmentation points by color G#758605; (c) body #758605 in Dense Cloud. Figure 7shows the segmentation of points carrying information about color value #758605. Figure 7a shows a 3D Dense Cloud model of individual points generated from photographs and created by the SfM photogrammetric method. Figure 7b visualizes the SW Agisoft environment for working with Dense Cloud and information about the color value carried by each point. The 3D SW works with RGB and HSV (Hue, Saturation, Value) color models in this case. Values can be directly numerically defined in Hex format. However, it should be noted that the RGB color model does not contain tonal or other values such as Hue, Saturation, and Value in the HSV color model. Figure 7c shows an example of the final segmentation of points with a color value of #758605 in the total number of clouding points in the 3D model. Figure 8shows the process of segmenting the points using color information #758605. Figure 8a shows the 3D texture model captured by the LiDAR sensor in Agisoft 3D SW. A point cloud containing 17 points was generated from this texture model, as shown in Figure 8b. The segmentation process of the point with color information #758605 is identical to that of the SfM method, as shown in Figure 7b. As shown in Figure 8c, no single point with a color value of #758605 was defined out of the total number of points. For this reason, the segmentation of points in the 3D models by the LiDAR sensor is not shown in Section 4.1. Electronics 2024, 13, x FOR PEER REVIEW 7 of 16 chosen as the default color model for this experiment. The values of the individual color components of light of direct color #758605 (Hex Color Value) are as follows: R = 117; G = 134; B = 5. In connection with Section 3.1. subsequently, in Section 3.3, attention is paid to the color space and gamut L*a*b*, which is suitable for subsequent work with color reproduction and display in the VR environment. 3.2. Color Value #758605 Segmentation In Agisoft 3D modeling, the SW environment can work with information in the color of individual points or groups of points in the Dense Cloud in RGB and HSV color models. This can be performed with a precisely defined Hex Color Value, as shown in Figure 7. (a) (b) (c) Figure 7. SfM—Points Segmentation #758605: (a) Dense Cloud 3D model using SfM photogrammetry; (b) segmentation points by color G#758605; (c) body #758605 in Dense Cloud. Figure 7 shows the segmentation of points carrying information about color value #758605. Figure 7a shows a 3D Dense Cloud model of individual points generated from photographs and created by the SfM photogrammetric method. Figure 7b visualizes the SW Agisoft environment for working with Dense Cloud and information about the color value carried by each point. The 3D SW works with RGB and HSV (Hue, Saturation, Value) color models in this case. Values can be directly numerically defined in Hex format. However, it should be noted that the RGB color model does not contain tonal or other values such as Hue, Saturation, and Value in the HSV color model. Figure 7c shows an example of the final segmentation of points with a color value of #758605 in the total number of clouding points in the 3D model. Figure 8 shows the process of segmenting the points using color information #758605. Figure 8a shows the 3D texture model captured by the LiDAR sensor in Agisoft 3D SW. A point cloud containing 17 points was generated from this texture model, as shown in Figure 8b. The segmentation process of the point with color information #758605 is identical to that of the SfM method, as shown in Figure 7b. As shown in Figure 8c, no single point with a color value of #758605 was defined out of the total number of points. For this reason, the segmentation of points in the 3D models by the LiDAR sensor is not shown in Section 4.1. (a) (b) (c) Figure 8. LiDAR—Segmentation of points #758605: (a) 3D model using LiDAR sensor; (b) Segmentation of points by color G#758605; (c) detail of the points generated in Dense Cloud. Figure 8. LiDAR—Segmentation of points #758605: (a) 3D model using LiDAR sensor; (b) Segmentation of points by color G#758605; (c) detail of the points generated in Dense Cloud. 3.3. Color Model and Gamut in Virtual Reality The previous sections present work with the RGB (Red, Green, Blue) color model and the sRGB color space, in which the most common sensing and display devices work. However, the type of VR headset and technical parameters must be considered when Electronics 2024,13, 4431 8 of 15 viewing a VR environment. In particular, the type and resolution of the VR headset fundamentally demonstrate the quality of displaying realistic scenes or models. It should be noted that the types of display devices for VR and their technical parameters differ significantly, and the color space or gamut needs to be precisely defined. These attributes have not yet been standardized for VR technology. When creating realistic 3D models and scenes, it is necessary to know in advance the type and parameters of the display VR headset and the final output presentation. Image processing needs to be adapted to these factors, which may be different. An Oculus Quest 2 VR headset was used in this study [ 38 ]. Almost all standard 2D and display devices operate in the sRGB color space [ 39 , 40 ]. However, in the case of VR imaging techniques, standard color models and gamuts are not used. Considering the difference between classic and VR head-mounted displays (HDM) is necessary. However, different HDMs use different color spaces and specifications. Therefore, the colors will be different from those of the standard display. However, different HDMs visually interpret the colors differently. Therefore, it is necessary to consider the type of VR headset during the virtual presentation at the beginning of the process. Older Oculus and Quest/Rift headsets interpreted the image in the Rec.709 color space. This is the most common color space characteristic of Internet content. However, this color space does not have HDR (High Dynamic Range) technology, which simultaneously enables the extended reproduction of details in dark and light partially captured scenes. Therefore, the Oculus Quest2 VR Headset was used in this study to visualize the artwork in a VR environment. This VR display device operates in the Rec.2020(HDR) default gamut and includes a standardized D65 white point corresponding to daylight. This is a different range of color simplicity for display devices. In connection with the different color spaces for the sensing and display devices in this experiment, it is appropriate to use the L*a*b* gamut (CIE 1976). The L*a*b* color space is derived from the first standardized CIE XYZ color space (CIE 1931) and is independent of a particular sensing or display device. This color space contains the full color range of the trichromatic triangle, as shown in Figures 6and 9. Also related to the CIE XYZ color space is the basic standardized color scale of ColorChecker Classic (X-Rite), used in color sensing and for color calibration of display devices according to international standards [ 41 ]. It is evident from the above that it is necessary to transform individual color models and gamuts among themselves [42]. As shown in Figure 9, it is necessary to consider the choice of color models and color spaces depending on the desired final output for realistic color reproduction. Figure 9a shows the standard color ColorChecker (X-Rite) used in the image capture and calibration device. The individual color positions in the chromatic diagram are shown in Figure 9b. Figure 9c shows the color model L*a*b*. The individual color ranges of the Rec.2020(HDR), sRGB, and L*a*b* gamuts are shown in the CIE 1931 chromatic diagram in Figure 9d. From the above, it follows that realistic 3D reconstruction of works of art and their visualization in a VR environment must pay attention to the issue of colors, color models, and games in each of the individual steps of the entire procedure. Electronics 2024,13, 4431 9 of 15 Electronics 2024, 13, x FOR PEER REVIEW 9 of 16 (a) (b) (c) (d) Figure 9. CIE XYZ 1931 standardized color space: (a) Basic ColorChecker standardized color gamut; (b) position of individual standardized colors in the CIE 1931 chromatic diagram; (c) color model L*a*b*; (d) CIE 1931 chromaticity diagram with Rec.2020 gamuts; sRGB and L*a*b. 4. Results As mentioned in the previous sections, the artwork was scanned using an iPad 11″ Pro smart device with a LiDAR sensor from Apple. The artwork was shot indoors in natural daylight and then in natural twilight using a tablet camera. For the 3D reconstruction of the object, two different methods were chosen using one scanning device. The 3D reconstruction of the object using the SfM photogrammetric method uses the integrated camera of the sensing device. The second method of 3D reconstruction uses a LiDAR sensor and the direct scanning of the Scaniverse mobile application to create 3D models. The goal of using these methods is to compare the quality of the 3D reconstruction of an object using one scanning device of the image, as well as other attributes, such as the scanning speed or the process of processing the image into a 3D model. Table 1 shows the attributes of the reference photos that were part of the individual series from which the 3D models of the object were modeled using the SfM method. In the case of 3D reconstruction using LiDAR sensor scanning, individual photos were not the source of digitization. Therefore, the image attributes are not listed. Table 1 lists the attributes and characteristics of the photo series that were applied to create a realistic 3D model. While the resolution, bit depth, aperture, and focal length are identical, the exposure data size and ISO are different for the reference photos. These attributes affect the progress of 3D model creation and the number of generated points in the point cloud carrying color information. Table 2 shows the number of generated points. Table 2 also Figure 9. CIE XYZ 1931 standardized color space: (a) Basic ColorChecker standardized color gamut; (b) position of individual standardized colors in the CIE 1931 chromatic diagram; (c) color model L*a*b*; (d) CIE 1931 chromaticity diagram with Rec.2020 gamuts; sRGB and L*a*b. 4. Results As mentioned in the previous sections, the artwork was scanned using an iPad 11 ′′ Pro smart device with a LiDAR sensor from Apple. The artwork was shot indoors in natural daylight and then in natural twilight using a tablet camera. For the 3D reconstruction of the object, two different methods were chosen using one scanning device. The 3D reconstruction of the object using the SfM photogrammetric method uses the integrated camera of the sensing device. The second method of 3D reconstruction uses a LiDAR sensor and the direct scanning of the Scaniverse mobile application to create 3D models. The goal of using these methods is to compare the quality of the 3D reconstruction of an object using one scanning device of the image, as well as other attributes, such as the scanning speed or the process of processing the image into a 3D model. Table 1shows the attributes of the reference photos that were part of the individual series from which the 3D models of the object were modeled using the SfM method. In the case of 3D reconstruction using LiDAR sensor scanning, individual photos were not the source of digitization. Therefore, the image attributes are not listed. Table 1lists the attributes and characteristics of the photo series that were applied to create a realistic 3D model. While the resolution, bit depth, aperture, and focal length are identical, the exposure data size and ISO are different for the reference photos. These attributes affect the progress of 3D model creation and the number of generated points in the point cloud carrying color information. Table 2shows the number of generated points. Table 2also shows the