[Feedback provided by David Ronca, Netflix:]
CEA-608 mandates a 32x15 grid in a 4:3 display area, CEA-708 allows either 32x15 (4:3) or 42:15 (16:9). If the input was CEA-708, how would a client know whether the display region should be 4:3 or 16:9.
Suggest adding a metadata header field AspectRatio=4:3 or AspectRatio=16:9 to capture this and then make the position calculation for 708 dependent on this.
Added some further description of how to convert the two different grids in CEA-708 to WebVTT and introduced a AspectRatio metadata header.
Surely we can say that VTT measurements are on uniformly-sampled axes, and have the author correct the non-uniform numbers to uniform ones? Do we need this complication?
David: This does not change WebVTT. It only explains better how to transcode a CEA-708 file to WebVTT. The AspectRatio metadata field is also just an hint for the video player in case it needs it. The conversion spec uses only percentage values for positioning, so it ends up being uniformly sampled.