I need to track a very wide 19:6 aspect ratio image, 13.5 metres wide and 4 metres tall. Users will be able to interact with various parts of the image, including on the edges. I can see that in this post Issue with flat image targets this was discussed with the response
“You can either adjust the dimensions of your image, or crop the tracking region to use the most feature-rich part of the image. Most of the time, you actually don’t need to upload the full image for tracking to work well.”
My use case may fall outside “most of the time”. I’ve ran some experiments and I’m able to display a 19:6 video with good tracking by cutting the image up into 4 separate image markers, then having 4 duplicates of the video and turning them on and off based on the currently detected marker. This works surprisingly well, but it won’t scale, I can’t keep duplicating everything because I’ll have complex animations, particle effects etc. Another approach I considered was reparenting content based on the currently detected “segment” of the image, however this doesn’t seem compatible with aframes assumptions regarding AR content. How can I resolve this?