In addition to what I already wrote, every track blends/composites with V1 - not V5 to V4, then V4 to V3, then V3 to V2, etc. If you want to know more you need to read the source code. I could make every video track blend with a hidden black background track, but then there will be less performance with a single video track even if there is no transparency. In some cases where images and filters result in RGBA, there is a check to see if there is any non-opaque pixel before taking the performance hit to run the blending operation. This situation may change in the future, but that is the way it is for now.