ArrayFire

This app is amazing! It takes a 3D video of your face in just a few steps, and the rendering is spot on! RayBan Aviators are always a little big on me, and the app showed them on my face exactly as they fit in the store!-davidandrew

learn the innovative tech behind the scenes and the impressive results that followed! Aaron Taylor Jun 23, 2021 3 min read Many find the experience of purchasing new glasses stressful. Even simply trying on frames can be uncomfortable when done in a public setting. Glasses.com set out to improve the experience by providing customers a way of virtually trying on glasses frames in the comfort of their homes via smartphone or tablet.The Glasses.com app takes a picture of the user’s face, creates a 3D model of facial features, and virtually places selected frames on the user’s 3D image. The user is then free to view the 3D image from a variety of angles.The application relies on computer vision to recognize the features of a user’s face and then generate an accurate, high-fidelity, 3D model of that face. Virtual frames can then be positioned on the 3D model.This app is amazing! It takes a 3D video of your face in just a few steps, and the rendering is spot on! RayBan Aviators are always a little big on me, and the app showed them on my face exactly as they fit in the store!-davidandrewMissionGlasses.com brought our engineers an excellent, fully-functioning version of the application; they simply wanted it accelerated. The app took full advantage of the CPU, whichâ€

and in securing facilities and public events. Speed is critical in their technology, and they approached us seeking acceleration.Reveal Imaging was running its cone-beam reconstructor via CUDA code previously implemented by a different contractor, but the performance was sub-par. They tasked ArrayFire with improving that performance on their GTX690s.ActionOur engineers approached this by following standard CUDA practices. The project consisted mostly of cleaning up the existing code to allow it to run faster.First, we removed the dependence of the permitted slice thickness on the number of GPU cards in the system. Then we allowed for one of the GPUs in the system to be excluded from those used by the reconstruction code and dedicated to driving a display. Next, we assured that the code met specifications with the latest CUDA compilers. Finally, we implemented performance enhancements, bottleneck reductions, and bug fixes previously identified.ResultsThe conclusion of the project was followed by a three-week testing period, during which time Reveal Imaging determined the reliability and overall satisfaction of the code’s performance. The code’s original speed was measured at around 200 slices per secondâ€