AMD Performance APUs and GPUs are capable of enabling real-time facial animation captures at full camera frame rates with the new Mixamo Face Plus plug-in for the Unity development platform.  Whether it is a standard VGA webcam capable of 30 frames per second, or a top end camera capable of 60 frames per second, AMD has the performance to deliver where it counts.

Detecting and tracking facial features such as corners of the mouth, eyes, eyebrows, etc., with real time performance is a daunting task.  We had an opportunity to do some testing to measure the system’s ability to process video streams of facial expressions in real time to animate a 3D character with the Mixamo Face Plus technology.  According to Mixamo, computing facial expressions at 30-40 frames per second will give enough margin to provide good results with standard VGA webcams, while higher frame rates mean that you can take advantage of top-of-line camera capabilities.  Our testing shows that a solid range of AMD 2012 and 2013 A8 and A10 APUs are up to the task of tracking facial features with real time performance and high quality.       


The test project was Mixamo’s stress-harness benchmark consisting of a 480p “poor quality” video containing a full range of facial expressions recorded in an uncontrolled environment with uneven lighting and lots of compression - typical of what you can often expect from an inexpensive, standard PC webcam.  The video is loaded into system RAM to simulate the performance you would get when streaming live from the webcam, thus taking out any disk I/O load time.  Thereafter, all the frames are processed at full speed so as to push the system to its computing limit.  At completion, the average frame rate is shown which is based on the total number of frames processed and the total duration it took for the processing of the facial expressions to complete.    


Such a scenario is a stress test because it requires a great deal more processing to handle the poor quality and still achieve good animation results. According to Mixamo, for cameras capable of higher frame rates and larger resolutions, the increased amount of data needing to be processed is offset by the higher quality video such that the overall processing requirements compared to a poor quality 480p video are lower. Hence, using a poor quality, SD video to simulate something closer to a worse case scenarios was used, which enabled consistent testing to be applied across all parts tested. This benchmark project was correlated with actual results measured when streaming live video and found to be a good representation of what a developer can expect from the system in terms of Face Plus performance.

This video shows Mixamo Face Plus technology in action




Mixamo Face Plus Release Pack August 2013 (approved).png

As this chart shows, even a 2012 mobile part, the AMD A8-4500M APU which was introduced over a year ago, has the computational throughput to support real time captures with standard off the shelf webcams whereas even a 2013 desktop part from Intel, the i5-4670, falls well short of the mark. AMD is pleased to be able to enable great performance and productivity to developers across such a broad range of AMD products to Unity developers and we are excited to see the technology being used to enhance upcoming games and interactive movies developed with the Unity platform.




Clarice Simmons is a Senior Marketing Manager at AMD. Her postings are her own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites, and references to third party trademarks, are provided for convenience and illustrative purposes only. Unless explicitly stated, AMD is not responsible for the contents of such links, and no third party endorsement of AMD or any of its products is implied


* Tests performed by AMD. PC manufacturers may vary their configuration yielding different results. Benchmark and test files provided by Mixamo. Test project measures the average speed at which the system can apply the Mixamo Face Plus facial gesture recognition and animation algorithms using a 480p test video containing a full range of facial emotions and gestures that is loaded into system memory to simulate a live video feed from a standard webcam in an uncontrolled environment.  An HP ProBook 6475b notebook PC with AMD A10-4600M APU with AMD Radeon™ HD 7660D Graphics, 8GB DDR3-1600 RAM, video driver, 12/19/2012, Windows 7 Professional 64-bit (6.1, Build 7601) Service Pack 1 achieved 42 frames per second with  GPU acceleration on, 3 frames per second with GPU acceleration off.  TRN-201


With an AMD reference desktop PC configured with an AMD A10-6800K APU with AMD Radeon™ HD 8670D Graphics, 8GB DDR3-2133, video driver, 6/4/2013, Windows 7 Professional 64-bit (6.1, Build 7601) Service Pack 1 achieved 49.5 frames per second with GPU acceleration on. An ASUS All Series desktop PC configured with an Intel Core i5-4670 with Intel HD 4600 graphics, 8GB DDR3-1600, video driver, 5/7/2013, Windows 7 Professional 64-bit (6.1, Build 7601) Service Pack 1 achieved 22 frames per second with  GPU acceleration on. RID-30


An HP Pavilion dv6 notebook PC with AMD A8-4500M with AMD Radeon™ HD 7640G Graphics, 6G DDR3-1600, video driver, 3/28/2013, Windows 7 Home Premium 64-bit (6.1, Build 7601) Service Pack 1 achieved 30.9 frames per second with GPU acceleration on. TRN-202