Be practical and reasonable; this isn't magic. We need feature rich images that are in focus with decent lighting. Another consideration is the comfort of the person being measured. Generally, normal athletic clothing is a good choice.
Probably the most common question I’m asked is about clothing recommendations; what should my athletes wear during collections? Although I believe I’ve been helpful in answering that question when asked, I’m not actually sure that I have ever answered it the same way. Perhaps it’s because it’s dependent on so many factors, or that my experience makes my opinion change. I’m not sure either way, but I am hoping that this post will help solidify some of this ambiguity, and provide the philosophical approach that I use when assessing the appropriateness of clothing choices during a markerless biomechanics collection.
In general, users typically think that the clothing should be similar to a marker based motion capture system: minimize the amount of clothing or make it skin tight. Unlike marker-based motion capture, we actually don’t necessarily recommend this. Instead, we generally prescribe body fitted clothing which can include jeans, joggers, shorts, or any piece of clothing that follows the body, without the requirements of being skin tight. More importantly is that the clothing is feature rich. What I mean by this is that, if you zoom into the image, are the body features really obvious (one example would be the kneecap) or is the clothing selection obscuring and making features less rich? As a more extreme example, a long flowy dress would obscure the entire lower body and the tracking would be less reliable (I bring this up because I think every lab collects this at one point or another). Here are some examples of good clothing. As they say, a picture is worth a thousand words.
A typical follow up question is whether tighter clothing will improve Theia’s tracking? As long as the clothing worn is body fitted, the pose estimation provided by Theia should not differ from tighter clothing. A study performed at the Queen’s School of Mechanical and Materials Engineering observed individuals wearing day to day clothing over three sessions (ie. different clothing, different days). The results indicated that there were no statistical differences between the clothing conditions, and that there was less variability introduced by clothing than introduced by marker placement. In other words, clothing didn’t have a big impact on the results.
A larger concern should be around camera focus, lighting and contrast in the video. It’s important that the participant is in focus by adjusting the video cameras to obtain a sharp image. The participant should also contrast with their background. If the background is white, the participant should not wear white clothing. This situation could be made worse with lighting that washes out the participant. So, the lighting needs to be bright enough to see the participant, but not too bright to wash them out. As you can see, it’s important to consider a few aspects simultaneously.
Not too crazy; basically, we need to be able to see the person very well, in a feature rich, focused, and well lit environment. Another thing I genuinely think of when asked this is how does the participant feel? If we force them to wear atypical clothing that isn’t theirs, are we introducing differences in their movement? If so, hopefully these aren’t significant enough to affect the outcome of the study.