This report and its findings were produced by Tata Elxsi. The report was sponsored by Unity Technologies. Tata Elxsi benchmark report: comparing PiXYZ Studio and Unreal Datasmith A Tata Elxsi perspective Abstract In the automotive industry, prototyping costs have risen exponentially with the increasing complexity of vehicles and dependencies across physical, electronic and software systems. In order to provide an alternative to physical prototyping, we make use of new technologies like Augmented Reality (AR) and Virtual Reality (VR). Viewing design data in AR/VR gives a real picture of the final product, allowing us to make changes instantaneously and helps in quick decision making. It also provides the ability to incorporate interactions with software-driven elements such as in-vehicle HMI. The use of VR and AR involves real-time rendering of the CAD data, which consumes a lot of memory and reduces the performance of the application. So data preparation plays a vital role to boost the performance while retaining the topology and mesh quality. Data preparation being a time consuming process, we use software applications like PiXYZ Studio and Unreal Datasmith to reduce the time and efforts. This report benchmarks PiXYZ Studio compared to Datasmith for optimization. While 1
both products are effective in importing CAD data, PiXYZ s performance allows faster workflows and produces more efficient models; PiXYZ was able to complete some workflow steps 66% faster than Datasmith and produced a model that was 28% of the size of the Datasmith model. (These numbers are subject to our assumptions detailed below and the sample CAD data we have used for experimentation.) 2
Introduction Visualization of any CAD data plays an important role in any product based industry. Physical prototyping has always been the main solution to visualize any product s feasibility. This is a tedious process and adds to the time and cost of development. In order to reduce cost we use the latest available technologies like Virtual Reality (VR) and Augmented Reality (AR). These technologies reduce the cost to a great extent. For VR/AR, data optimization plays a crucial role, but, it s a time consuming process. In order to reduce the time for the process we use software like PiXYZ and Unreal Datasmith that can help us save time and in some cases partially automate the process altogether. Tata Elxsi supports global automotive OEMs by providing high quality design and engineering solutions and a superior experience for its customers. We have consistently and efficiently supported global automotive clients to develop vehicles conforming to international standards. We have Centers of excellence (CoE) in areas of styling / aesthetic design, Class A surfacing in BIW (Body in white) design and UI design. Tata Elxsi has in-depth understanding of global trends, consumer needs and technological developments. With an in-house state-of-the-art 3D prototyping facility and global vendor network, Tata Elxsi helps clients to build full scale mock-ups and prototypes. Tata Elxsi has a dedicated and experienced CAD/ CAM/ CAE team to support design validation. In this report, we aim to explore PiXYZ and Datasmith software that help to reduce the time consumed for the process of data preparation. These software applications also help in automating the process. (Note: Unreal Datasmith is a plug-in for Unreal engine, announced in July of 2017. Unreal more recently refers to Datasmith as part of Unreal Studio which includes the original Datasmith product and Unreal engine.) Benchmark Methodology This document describes the suite of test cases used to compare the workflow of optimizing CAD data using PiXYZ for use in Unity and Datasmith for use in Unreal. Functionalities Tested in the Benchmark i. Number of different software applications required for the process of optimization. ii. Total time for import processes. iii. Final properties of the model (e.g., polygon count). iv. Model quality (e.g., mesh detailing, flipped normal, duplicate faces, keeping topology as intact as possible). v. Degree to which the hierarchy of data is maintained. 3
vi. Preservation and maintenance of metadata as properties in Unity /Unreal. Benchmarking Strategy Figure 1 below illustrates the steps of the process followed for optimizing CAD data using PiXYZ for use in Unity compared to Datasmith for use in Unreal. The flowchart helps us in benchmarking. Fig. 1 Benchmarking Strategy The process is divided into three phases namely Input, Data Optimization and Output. The general steps for processing are: 1. Import CAD data 2. Repair the CAD mesh 3. Convert Z- up to Y-up 4. Tessellation (convert NURBS to polygons) 5. Repair the mesh to rectify orientation of surface normals. Then manual inversion of faces if necessary 6. Delete patches/lines 7. Apply materials with basic colors to meshes according to how they are manufactured and how the look (body, seats, plastics etc.) We apply transparent material first to check if layers of meshes are present and then change it to solid colors. 8. When transparent colored materials are applied, if any double surfaces are noticed, we manually delete it. (Note PiXYZ has tools to automatically identify duplicate faces.) 9. De-select all meshes, and then do merge meshes by materials 10. Decimate overall 11. If necessary, decimate each part separately as per requirement 12. Export model and import into Unity or Unreal Benchmarking Software Applications This section gives a list of software application and plugins that were used during the benchmarking. 1. PiXYZ Studio (ver 2018.1.0.13) 2. Unity 3D (ver 2017.3.0f3) 3. Datasmith Plugin for 3ds Max (ver 2018) 4. Datasmith Plugin for Unreal Engine (ver 4.19.1) 4
Model used for Optimization Figure 2, shows the CAD model used for benchmarking. This is generic CAD data of a premium sports SUV. Detailed Steps Comparison Fig. 2 CAD model used for Benchmarking Below are the steps that explain the process of optimization in detail using PiXYZ and Datasmith. The screenshots of workflow in PiXYZ (Top) and Datasmith (Bottom) are given below for each step. Numbers in the brief comparison table given below and screenshots are for the car CAD data we have considered in our experimentation. The system configuration we used is as follows Intel Core i7 CPU (4 GHz, 4 Cores), 64GB RAM, Nvidia 1080 GTX dual SLI, 8GB VRAM, 512 GB SSD. Note please zoom in to about 240% to clearly see all the information in screenshots. Note the numbers in the comparison table are based on assumption that 3DS max user's expertise is average. 1. Import STP data. In 3ds Max while importing we mention the mesh resolution value as -10 and Y-up Axis. 5
2. Repair the CAD mesh. This option is available for only PiXYZ. 3ds Max does this automatically while importing. PiXYZ also offer an automatic CAD reparation + tessellation feature while importing. 6
3. Convert Z- up to Y-up. We do this in PiXYZ as Unity is Y-up coordinates. 3ds Max gives us an option to choose the axis system at the time of importing. 4. Tessellation (convert NURBS to polygons). We tessellate the mesh in PiXYZ according to our requirement (Parameters- Max Sag = 0.1, Max Angle = 7, UV Mode = UniformUV, Create Tangents & Create Binormals are checked in the Advanced options.) Mesh tessellation in 3ds Max happens while importing and is based on Mesh resolution while importing. 7
5. Orientation of normals of a few surfaces are flipped as per requirement. PiXYZ has a built-in feature to do this automatically while we can write a script to do the same job in 3ds Max. Also, if needed, manual inversion can be done in both. 6. Delete patches/lines in PiXYZ is a separate (automated) option while 3ds Max does the job while importing. 7. Apply materials with basic colors to meshes according to how they are manufactured and how they look (body, seats, plastics etc.). PiXYZ Studio Editor offers the potential to automate the assignment of materials according to metadata (PiXYZ automation was not used for this benchmark). 8
8. We apply transparent material to check for double meshes. We manually delete any double/duplicate mesh present. PiXYZ Studio has built-in tools to assist with identifying double meshes. 9. We merge meshes by materials to reduce the batch count as it is crucial for real time rendering. Algorithms are available in PiXYZ to optimize the product structure, to make a model suitable for real-time context, based on materials or other metadata. In PiXYZ we can see the occurrences count to be 13 based on the number of materials count. And in 3ds Max we see that the alloys are combined together. 9
10. Decimate over all. We decimate it accordingly as per our requirement to reduce the polygon count. (Parameters chosen are Surfacic Tolerance = 1, (Advanced Settings) Normal Tolerance = custom, Degrees = 8). In 3ds Max decimation is done while exporting as *.UDATASMITH format. 11. Export model to FBX from PiXYZ and import into Unity. Similarly we import UDATASMITH into Unreal. Below are few images of the final output. PiXYZ in Unity (Top) and Datasmith in Unreal (Bottom). 10
PiXYZ in Unity Datasmith in Unreal Fig. 3.1 Final output images 11
PiXYZ in Unity Datasmith in Unreal Fig. 3.2 Front view images 12
PiXYZ in Unity Datasmith in Unreal Fig. 3.3 Interiors PiXYZ in Unity 13
Datasmith in Unreal Tabular Comparison Fig. 3.4 Infotainment # Functionality PiXYZ/Unity 3D Datasmith/Unreal Lead Leader advantage % 1 Number of Software Applications/Plug-ins used 2 (PiXYZ Studio, Unity 3D) 3 (3DS Max, Unreal, Datasmith plugin) PiXYZ 33% 2 Import STEP data (Import time + Tessellation) 227 seconds 390 seconds PiXYZ 42% 3 4 Applying Materials (Manual Process) Combining Meshes(Automatic Process) 5 hours* 6 hours PiXYZ 17%* 2 minutes 5 minutes PiXYZ 60% 5 Decimation (Automatic Process) 2 minutes 6 minutes PiXYZ 33% 6 Final Polygon Count (Initial - 17 million triangles) 519,000 triangles (3% of initial) 1.8 million triangles (10.5% of initial) 7 Mesh Detailing** 9 9 Tie PiXYZ 71% - 8 Hierarchy of Data Maintained 100% 100% Tie - 9 Preservation of Metadata and Reusability** 9 7 PiXYZ 22% * An automated material assignment process is also available in PiXYZ Studio; this benchmark compares only the manual process. Automating the assignment in PiXYZ Studio will significantly reduce the time required. ** Rated on a Scale of 1 to 10, 10 being the highest. 14
Conclusion In this benchmarking process we have compared PiXYZ using Unity Editor and Datasmith using 3ds Max and Unreal Engine. Both are effective in importing CAD data, although PiXYZ supports 14 more CAD formats than Datasmith (e.g., Alias, CSB Deltagen, IFC (BIM), Maya, PLM XML, etc.) and works with any rendering solution or game application whereas Datasmith works only with Unreal. Regarding data preparation, PiXYZ offers performance advantages of up to 71% and leads in 7 of 9 areas measured. PiXYZ also offered more data preparation features (e.g., smart hidden geometry removal, de-featuring / simplification) which allows for more efficient workflows and more efficient models to import into Unity or Unreal. There are more advantages for PiXYZ than Datasmith. From the tabulated results it can be inferred that PiXYZ is doing a better job as the triangle count value has been reduced by more than with Datasmith, while maintaining the topology and mesh quality. Many users who tried PiXYZ found it very easy on the UI interaction side. For Datasmith, users also have to be familiar with 3ds Max UI. The dependencies of software applications are lesser in PiXYZ as compared to Datasmith. Both the software applications do a great job in preserving the topology, mesh quality and hierarchy. The total process is easier to do in PiXYZ due to the automation of many features. PiXYZ also lets you automate the process for data directly from PLM which is an added feature compared to Datasmith. 15
About the Author Tata Elxsi s Digital Visualization team works on cutting edge technologies to help various industries in diversified domains like Engineering, Marketing, Training, Education etc., to keep them a notch higher than their competitors. Augmented Reality (AR) and Virtual Reality (VR) is the technology that is growing at a very high pace and is on demand. AR/VR helps in visualization of products, thereby cutting down the development time and cost that puts them ahead in the market. This team comprises of experienced designers, engineers and artists who work towards providing integrated niche solutions to clients. Contact us For more information, please contact at: info@tataelxsi.com About Tata Elxsi Digital Visualization (AR / VR & MR) service offerings Tata Elxsi s Augmented and Virtual Reality practice combines design, technology and visualization to help businesses to drive innovation, increase operational excellence, enhance productivity and create personalized customer experiences. The AR / VR team is working with diverse industries to propel digital innovations including - Automotive, Marine, Healthcare, Broadcast, Retail, Education, Entertainment, Manufacturing and Engineering. Some areas we are currently working on include: Field Force training Product visualization Manufacturing In - Store experiences Product marketing With industries converging, Tata Elxsi s expertise in delivering cross-industry services puts us in a leadership position to deliver better innovations, enhanced solutions and bring business transformation. About Tata Elxsi Tata Elxsi is a fully integrated global design and technology consultancy, part of the $ 103 billion Tata Group, creating innovative products, services and experiences to build brands and help businesses grow. By intersecting design and technology, Tata Elxsi enables clients globally to bring new ideas and products to market. Its services span across research and strategy, to design, creative digital content including augmented reality and virtual reality, technology implementation and systems integration. Tata Elxsi addresses the automotive, communication, consumer products, defense, healthcare, media & entertainment and transportation sectors. This is supported by a network of design studios, development centers and offices worldwide. Tata Elxsi is a listed company and headquartered in Bangalore, India. 16