The flexibility of game development engines like Unity and the exponential increases in GPU throughput (thank you cryptocurrency miners!) represent a revolution in 3D art and animation for manufacturing and other engineering-led industries. Mature product companies design, develop and manufacture their products using CAD/CAM engineering software. CAD/CAM, or Computer Aided Design/Computer Aided Manufacturing precisely renders product designs within the strict confines of engineering data sets.
When these data sets – representing a complete digital blueprint of a product – are combined with modern manufacturing techniques, it dramatically reduces the cycle time between design and manufacturing. Unfortunately, using those models for print, website, animation or even augmented reality/virtual reality via a traditional pre-rendering based production process is complicated, time consuming , and expensive. This process can be so arduous that many will create new, customer-facing 3D models from scratch – sacrificing accuracy and efficiency that should be derived from the original engineering data.
Due to their intricate detail, the 3D assets are typically very large files with massive polygon counts (oftentimes in the millions, which is impractical and not necessary for real time engine use). In standard practice, files are translated into 3D geometry models useful for print and digital via a lengthy rendering process. But given the size, it can take a significant time and manual labor to translate for their final intended purposes. Rather than investing in modification and rendering each time we need to create a new asset, we can lower the polygon count and texture models in order to allow a real-time environment like Unity to quickly generate specific views, animations, or even interactive 3D experiences.
To get the CAD models to a point in which they are optimized for correct display, you need to reduce the polygon count. This process at times can produce anomalies that cause issues with the integrity of surface quality (see below pictures as examples), but can be solved by creating normal maps using the original high poly model as reference. When you apply these normal maps to the low poly model, the surface quality of the models will render correctly – despite not being the same polygonal count. If downsampling is well managed and performed by a skilled artist, the visual fidelity can match and potentially exceed the traditional production quality at a tiny fraction of the time.
This drives the cost of high quality 3D work down, bringing the capability to deliver 3D content and immersive experiences to companies and product lines that couldn’t previously justify the investment. And leveraging the CAD data as a starting point for real-time rendering via game engines, we net an increased level of accuracy and fidelity that does not compromise the original intent of the object.
With translation complete the result is a versatile, accurate 3D geometric model. Surface materials are applied to the geometry that allow plastics to look like plastics and metals to behave as metals. Lighting, camera placement and an entire range of other artistic decisions provide complete visual control of imagery or animation – all in real-time. This allows artists and designers to experiment with immediate feedback – creating deliverables for multiple end applications with efficiency.