Good subject Freg, but a question with so many complicated and conditional answers that I could write a whole book on the subject...
Camera avoidance is a serious problem for those trying to continuously monitor their local deer population over long periods of time, or for those who are trying to pattern the movement of particular bucks. And there is no real answer for the problem. Every potential "fix" also has its downsides.
I agree that video mode on trail-cameras can often see deer behavior and catch individual deer that a single picture would not have caught. However, video mode also has its downsides. Video imagery is recorded at MUCH inferior quality (image clarity/pixel count) than still images can be. Many newer trail-cams use imagers of at least 5 megapixels, while video is captured at generally far below 1 megapixel. This can make identification of individual bucks by antler configuration, especially those in the background, very difficult.
In addition, visible flash trail-cameras produce FAR more camera spook and camera avoidance due to the long-duration illumination needed at night than the burst-flash of still imagery.
Video mode also uses up battery power and memory card space much faster than still images.
Personally, I like to set my cameras on still-magery and also set to take many burst-mode pictures with each trigger. Normally I want no less than 5 burst mode images for each trigger, and sometimes I'll set my cameras for 7 or 9 burst-mode images with each trigger. This carries the picture taking 10-15 seconds after the initial trigger. And for cameras pointed at scrapes or trails, I set my cameras for very short delay times (the amount of time the camera will wait before recording the next trigger). I prefer delay times for scrape or trail set-ups of 15, 20, or 30 seconds. With a burst mode of 5-9 pictures per trigger (which can take 5-20 seconds to complete), and a delay time of 15-30 seconds before the next trigger can be recorded, I'm missing little, and I'm recording the events at very high resolution (high image quality).
I do agree with the video in the link posted that setting cameras high and pointing them down greatly reduces camera avoidance. However, this also limits the trigger area greatly. Often, a highly valuable picture occurs because a deer in the background--not moving through the area I have targeted--triggers the camera. A high, downward set-up would not have captured that picture. In addition, unless you have just a few cameras to work with, who has the time and energy to carry a ladder with them to camera set-ups?
From my experiences, the best ways to reduce camera avoidance are:
1) Use true black-flash cameras. The difference in the number of camera events over time, and the number of repeat camera triggers by the same deer, with black-flash versus visible flash cameras is night and day. This is especially true when a camera must be left (or you want to leave it) in one location for long periods of time. With visible flash cameras, my trigger event counts decline rapidly obver time if a camera is left in the same location. However, with black-flash cameras, my trigger events actually increase dramatically over time.
Now without question, as the video suggested, deer ABSOLUTELY DO see the cameras, even in daylight, and this can cause negative reactions and camera avoidance. However, I have much less problem getting the same mature bucks to come back multiple times past a black-flash camera than a visible flash camera, even when I have pictures of them staring right at the cameras during daylight. The deer notice both types of cameras, but the black-flash produce much less long-term camera avoidance.
2) Do everything possible to reduce human scent when setting up and checking trail-cameras. I believe much of camera avoidance is due to deer smelling your walking trail to and from the camera. Either place cameras near easily accessible locations (near roads), or ride right to your cameras on an ATV. Using an ATV to ride right up to the camera really seems to help in scent reduction.
3) Move cameras to new locations fairly frequently. Even with black-flash cameras, I often move them weekly or biweekly. Not only does this reduce the effects of camera avoidance, it can help you monitor new locations that may be being used by deer that were not using the previous location, hence providing opportunities to catch "new" deer.
4) Use cameras to monitor small, out-of-the way attractant food plots. Not only do bucks prefer to use smaller food plots (they are less exposed), the constant human activity near food plots usually makes deer less wary of human scent in those locations. It isn't just human scent that scares deer, it is human scent in areas that normally don't contain human scent (out of the way locations back in the woods where humans only go during hunting season) that really spooks the crap out of deer. Deer will become aclimated to human scent where it occurs regularly, like around agricultural activity.