Advancements in 3D printing technologies, availability of online bureaus offering 3D printing services, and affordable high-resolution digital cameras (including those in smartphones) present opportunities for novel ways to visualize and interact with rocks and rock surface data. This paper documents and explores some of these opportunities with examples produced using the full-color binder jetting 3D printing technology. Opportunities include use by geo-educators, geotechnical investigators, museum curators, model railway hobbyists, and others who have a professional or informal interest in rocks and rock outcrops.
The significant advances and improvements in 3D printing (i.e., rapid prototyping) technologies (3DPrinting.com, 2015), 3D reconstruction technologies (De Paor, 2016), and digital cameras (including smartphones) offer the opportunity to generate highly detailed replicas of rock and rock faces for a variety of educational, technical, and general interest endeavors. Reasons for using 3D printed replicas include situations when a scale-model is required, access to the real object is not practical because of safety concerns, there are issues about handling the original item owing to its rarity or value, the item is physically inaccessible owing to its location, or it is inappropriate to remove the item from its original location.
This paper outlines the relevant rapid prototyping technologies and illustrates potential applications with some examples of rocks and rock faces that have been 3D printed. Replica rocks and rock faces that can be produced with some of the 3D printing systems are of sufficient resolution and color fidelity that these can be considered (even if scaled down to a manageable size) for technical studies and measurements, as a geology training or education tool before heading into the field, for a memento of a holiday or field trip, or as scenery items in model railway layouts or museum dioramas.
3D PRINTING TECHNOLOGIES AND 3D PRINTERS
The early rapid prototyping (RP) technologies of the 1980s that brought about the 3D printing capability include techniques such as: stereolithography (SL), selective laser sintering (SLS), and fused deposition modeling (FDM) (3DPrinting.com, 2015). The different techniques use different materials, and these, in turn, produce 3D models in a range of different materials and sizes, and with different characteristics (3D Hubs, 2016). However, it is the binder jetting (inkjet) (Additively.com, 2015) method with the “full-color sandstone” gypsum-based powder material that is most relevant to 3D printing of rocks and rock surfaces owing to the large color range and object size. Recent advances in the use of multiple colors in the SLS process have brought this technique into consideration, but this approach does not yet deliver the color fidelity and larger object size of the Binder Jetting printing systems.
The binder jetting technique (see Fig. 1) comprises a very fine powder that is progressively built up, one thin layer at a time. These individual powder layers are formed into a solid object by jets of adhesive. Each powder layer is scraped from the powder supply holding tank over to the powder bed and rolled level, after which the inkjet print head traverses above the powder layer, squirting jets of colored or colorless adhesive coincident with the walls of the 3D model geometry.
There are companies such as 3DSystems (www.3dsystems.com), Stratasys (www.stratasys.com), and Konica Minolta (www.konicaminolta.com.au/Products/3D-Printers) that specialize in the manufacture and/or marketing of 3D printers, and most have a printer range that includes at least a few of the different techniques listed above. Although there are affordable fused filament fabrication (FFF) printers such as the MakerBot (www.makerbot.com), the majority of printers and printing techniques are expensive. An alternative to purchasing a printer is, however, available in the form of online 3D printing service providers such as Shapeways (www.shapeways.com), Sculpteo (www.sculpteo.com), and i.materialise (i.materialise.com) that offer a relatively affordable way to print high-quality models, no matter if one’s interest stems from a hobby or professional activity. In addition, these service providers have a range of printers that cater to the different printing techniques, thus giving customers a wide choice of printed material without the cost of ownership overhead.
GENERATING REALISTIC AND ACCURATELY SCALED 3D MODELS OF ROCKS AND ROCK FACES
The process of 3D printing requires an appropriately constructed and sized 3D model, and the techniques for generating these have, fortunately, been advancing at a similar rate to that of the 3D rapid prototyping technology. In the case of rocks and rock faces, the goal is to generate realistic, scale models that portray as many of the real features, textures, and colors as possible, with sufficient resolution to the activity being undertaken. Products such as AutoDesk® 123D Catch (Autodesk, 2015), together with significant advances in the underlying techniques and algorithms of digital photogrammetry (which use digital images), provide the capability to generate the required models with relative ease and required fidelity. In other cases, the 3D models can be generated from laser-based or patterned-light scanning techniques; however, these may or may not include color texture images that are or can be directly mapped to the 3D mesh.
De Paor (2016) provides a complementary and detailed description and review of techniques applicable to the creation and virtual display of 3D digital models of rocks relevant to a range of geoscience subdisciplines. The techniques discussed by De Paor (2016) for creating virtual rocks are equally applicable to the generation of 3D models for 3D printing, with the important proviso that the models are made “watertight,” i.e., have closed surfaces and no holes. The Structure from Motion (SfM) 3D reconstruction technique for model generation, which makes use of smartphone or other digital camera images, is a particularly relevant and convenient method in the context of this paper for capturing and generating 3D models of rocks and rock faces, whether they are located indoors or in the field. The SfM technique is the basis of the commercial Agisoft PhotoScan and AutoDesk® ReMake products and the open-source VisualSFM application (De Paor, 2016). ReMake (which replaces the discontinued AutoDesk® 123D Catch application) and PhotoScan have attractive licensing options for educational institutions, which should be good news for those interested in applying 3D reconstruction to educational pursuits.
In most cases, the 3D models that are generated from 3D reconstruction methods are only surfaces (which have no thickness), and, unless these are inherently closed surfaces, they need to be thickened or solidified in order to have the correct and minimum thicknesses for 3D printing. In addition, some editing of the mesh or removal of artifacts may be required to fill holes, remove unwanted polygons and vertices, or generally improve the mesh quality for printing. The thickening process can prove somewhat challenging because the software tools to achieve this, on anything but simple meshes, are few and far between, and those that do exist provide limited capability and success. The readily available software tools that have provided the greatest success are: Blender (Blender Foundation, 2015), AccuTrans 3D (MicroMouse Productions, 2015), and MeshLab (MeshLab, 2014). AccuTrans 3D works best on surfaces that do not have faces that are parallel or subparallel to the direction of thickening; the relevant feature is “Extrude as pseudo 2D surface.” The AccuTrans 3D Help document provides details of the necessary steps to use this feature. Blender offers a more complete solution with its “3D Toolbox” and the “Solidify” modifier. MeshLab provides “Uniform Mesh Resampling,” and this filter with the “Absolute Distance” option can thicken mesh objects.
There are at least three elements to the accuracy of the final printed object—namely, the photo-reconstruction process, editing of the 3D model, and the 3D printing process. The accuracy of the photogrammetric process has been studied and assessed by authors such as Barratt (2013), Chandler and Fryer (2013), and Santagati and Inzerillo (2013). Accuracy of the 3D printing process depends on the chosen material and printing process. One study conducted at Curtin University (Islam et al., 2013) of the disparity between the manufactured (i.e., 3D printed) dimension and original designed specification provides some relevant evaluations. This study, which used a ZCorp Z450 printer, found that dimensional inaccuracies can occur in 3D printed models and that these were undersized in the xy (horizontal) plane and oversized in the z (vertical) direction. Another study of 3D printed vertebrae by Ogden et al. (2014) found dimensional measurement discrepancies between the input 3D STL model and the output 3D printed model in the case of a Makerbot Replicator 2 3D printer. Additional aspects of the printing process that should also be confirmed and assessed in terms of reliability and accuracy of the reproduction process include resolution and repeatability. For instance, the 3D Systems ProJet 660 Pro printer has a specified resolution of 600 × 540 dpi with a minimum feature size of 0.1 mm, and the ZCorp Z450 printer has a specified resolution of 300 × 450 dpi and a minimum feature size of 0.1 mm.
The larger example objects described later in this paper were not measured when the photographs were taken nor were their dimensions subsequently validated; but this would have been possible had suitable measurements been recorded at the time of data capture. The smaller objects were measured subsequent to the photography and the generated 3D models scaled to conform to the recorded measurements. In order to facilitate the correct scaling and measurement of objects, the instructions for 123D Catch recommend the inclusion of an item of known length within the initial photographs that can be used to specify a reference distance (Pouliquen, 2011).
3D PRINTS OF ROCKS AND ROCK FACES
The capability of available 3D reconstruction techniques, software applications, and digital cameras (including smartphones) to generate rock and rock-face models suitable for 3D printing is explored with a selection of four case studies. In addition, the current state of the binder jetting method using the “full-color sandstone” material is highlighted in the following four examples that were 3D printed by Shapeways on a 3D Systems ProJet 660 Pro printer (Shapeways, 2016).
Examples of 3D Printed Models of Rocks and Rock Faces
The first example of a 3D printed model is a ~0.8 m × 0.6 m portion of Permian sandstone outcrop from Cullercoats, UK, shown in Figure 2. The original 3D model (Fig. 2A) used to generate the printed model (Fig. 2B) was produced using SIROVision photogrammetry software (Poropat, 2001, and sirovision.dataminesoftware.com) and an in-house SIROVision TIFF to VRML converter. It was subsequently thickened for 3D printing using AccuTrans 3D. The same Cullercoats outcrop and model were the subjects of a surface roughness study previously published as an article in Geosphere (Baker et al., 2008). The original SIROVision model is available as a 3D PDF animation that can be downloaded in conjunction with the previous article (Baker et al., 2008, p. 423).
The second example (Fig. 3) is a 20-cm-long sample of Ventersdorp Contact Reef gold-bearing conglomerate rock from the Witwatersrand, South Africa, photographed 24 times with an Apple iPhone 4, reconstructed using 123D Catch, and prepared for 3D printing using a combination of MeshLab and Blender. Preparation included removing and filling several protruding faces (artifacts of the reconstruction process) and thickening the walls.
In the third example (Fig. 4), a replica of an ~80-cm-tall limestone pinnacle, one of the many that make up the Pinnacles in the Nambung National Park near Cervantes, Western Australia, was generated from 20 photographs taken using a Canon EOS 550D and 18 mm focal length lens. These photographs were processed into a 3D model using 123D Catch and prepared for printing using a combination of MeshLab and Blender software.
Next (Fig. 5) is a geotechnical example of a ~1.5 m × 1.5 m area of exposed friable rock face supported with rock bolts in an excavated cutting near Hobart, Tasmania. This rock face was captured with 16 photographs, using an Apple iPhone 4, reconstructed using 123D Catch, and prepared for 3D printing using AccuTrans 3D. Interactive Item 1 provides a 3D file of this model.
The final example (Fig. 6) is a ~75-cm-diameter rock photographed with an Apple iPhone 4 on Cable Beach, Broome, Western Australia. This model was reconstructed in 123D Catch using 29 photographs and prepared and thickened for 3D printing using Blender. Interactive Item 2 provides a 3D file of this model.
The dimensions and cost (in U.S. dollars) of the example models described above are summarized in Table 1. The model dimensions and material volumes given are the values reported respectively in cm and cm3 on the 3D printing service provider’s webpage for each model after uploading.
Level of Realism Depicted in Example Models
Figures 7–10 provide additional views of the printed model examples to illustrate the level of realism that is now achievable in the finished model. The printing resolution, texture definition, and color reproduction are able to depict fine details that include: thin ridges (Fig. 7); cracks, joints, and fissures (Fig. 8); indentations and footprints (Fig. 9); and complex convex and concave surfaces (Fig. 10).
This paper demonstrates that 3D printing technology has reached a sufficiently mature stage that it is possible to print 3D models of rocks and rock surfaces with adequate geometrical and visual realism. This technology, in combination with equally mature photogrammetry 3D reconstruction software, provides a compelling case to replicate 3D models of rocks and rock surfaces for a variety of purposes. In addition to the hand-held and ground-based camera (including smartphone) methods of image capture used for the examples included in this paper, the increasingly easy-to-use, readily available, and affordable drone-based camera systems open up even greater opportunities to acquire and generate very compelling and instructional 3D models of rock outcrops and other surface features.
The level of realism of 3D printed models of rock does not extend to the full geological character and physical property of the original rock material; for instance, the density, hardness, luster, chemical reaction, and, in some cases, object size do not represent the genuine specimen. Nonetheless, there are sufficient use cases that justify the application of these techniques to meaningful personal or professional endeavors. In some cases, such as the Pinnacle example, the powdery and chalky nature of the 3D print material somewhat resembles the feel and texture of the original rock. In other cases, when a glassy visual appearance is required, an epoxy resin can be applied to the surface of the “full-color sandstone” printed model, such as the “coated” option available from Shapeways (2017).
The additive 3D printing process that is used in the “full-color sandstone” and equivalent materials permits the successful reproduction of complex object shapes. This capability to cope with irregular and concave surface geometries is evidenced by the Cable Beach example (Fig. 10) and the numerous online examples of complex shapes provided by the 3D printer manufacturers and service providers (for example, Stratasys, 2017 and Sculpteo, 2017). Consequently, it is anticipated that even the most complex of 3D rock models can be printed successfully. However, complex rock geometries are more likely to give problems at the earlier stage of generating digital models using the available scanning and photogrammetry techniques, in particular, extreme concave areas of rock surfaces such as occur in scoriaceous basalt (De Paor, 2016) and geodes.
The accuracy and scale of the generated digital and 3D printed models are crucial, if the purpose of reproduction is for scientific and/or technical evaluation (or similar) purposes. These aspects can be quantified and validated by taking appropriate steps during the capture phase and subsequent processing of the 3D model geometries. Further work is required to assess and quantify these aspects of the models presented in this paper.
The availability of 3D printing (rapid prototyping) technology and robust photogrammetry 3D reconstruction software and the proliferation of digital cameras, especially those in smartphones, present some interesting possibilities for capturing the moment and objects encountered in our travels and professional endeavors. These three technologies combine to put a practical tool in the hands of those interested in rocks and their replicas for a variety of pursuits that include:
physical record of a vacation or field-trip memento;
reproduction of a rock specimen for non-technical demonstration purposes;
model of a rock outcrop or exposure for geological teaching purposes;
scale model of a rock cutting for a model railway layout or museum diorama; or
scale model of an inaccessible rock face for geotechnical assessment.
This work was supported by The Pawsey Supercomputing Centre through the use of advanced visualization resources located at The Australian Resources Research Centre with funding from the Australian Government and the Government of Western Australia.