There is no one size fits all, so ask a lot of questions and make sure the camera is the right one for you. Even though we don’t sell cameras, we will happily answer your questions and set you up with a great solution for your application.
There is no straightforward answer to this question, and, despite the fact that we don’t actually sell cameras, this is something our sales team spends a lot of time on. The reason it’s so important is that when the right camera is selected, you don’t notice it! This is exactly how it should be, right? The purpose is to measure humans, not stare at the cameras. However, when the wrong selection is made, it will plague data collections and likely render the system far less useful than its intended purpose. In this blog post, we will explore three key factors that may influence what camera you select when using Theia3D. Firstly, some definitions that may seem simple, but are really important: resolution, frame rate and integration/sync capabilities.
1. Resolution
Image resolution refers to the level of detail that an image holds. The level of detail is typically measured with pixels. A 5MP (megapixel) camera means the image captured is 50,000,000 pixels. Some camera manufacturers may also use the following notation to describe image resolution: 2432 x 2056P. This just refers to the amount of vertical and horizontal pixels in the video image. If you multiply these two numbers, you’ll reach the total resolution: 5MP.
Why does resolution matter? Well, we need granularity in the image to make accurate detections. Generally, we require the subject of interest to be at least 400 pixels tall in the video image. You can easily imagine a video where the subject is smaller, which would make accurate detections challenging (trust me, this is way more common than you think). Also, most cameras have the ability to collect video data at varying resolutions (which is often a mechanism used to achieve higher frame rates), so it’s important to look at the resolution at the frame rate you want to collect at. In summary, the greater the detail of the image, the better suited it will be for markerless motion capture, and this really does have an impact on results.
2. Frame Rate
Frame rate refers to the speed at which images are collected by the video camera. Frame rate is often referred to in camera manuals as fps (frames per second). Typical frame rates include 60, 85, 250, or 500 frames per second.
Assuming that the resolution requirements are met, frame rate is the next key consideration for markerless motion capture. For slower movements, such as walking, lower frame rates are probably okay.
However, using a lower frame rate for a high-speed, dynamic movement, such as a baseball pitch, will lead to many issues. Firstly, motion blur will likely occur, which will adversely affect tracking. Now even if this is corrected by reducing the shutter speed, if the capture frequency is too low the measured signals may not represent the actual movements (read up on Nyquist frequency for more information). In general, the higher the capture frequency (at the same resolution) the better.
3. Integration/Sync Capabilities
Finally, we strongly suggest considering peripheral device integration and synchronization. For many biomechanics applications, users will want to connect and synchronize peripheral devices like force plates. Synchronizing these after the fact is not best practice, and will be a giant pain, so if this is a requirement, it needs to be discussed early, because many camera companies do provide this type of integration capability. Our recommendation is that these peripheral devices are synchronized via GenLocked.
So yes, a few things to consider, but they are really important before buying that system. If you have further questions about how to select the best camera for your markerless motion capture application, please contact Theia Markerless here. We would be happy to hear from you and can also provide a demo of Theia3D. We can also share the different capabilities from each manufacturer and give a good idea of the best cameras for your use case (and likely provide many options for you to pick from).