Comment by schiffern

Nice. This reminds me how I felt watching the first pre-iPhone multitouch demos.

One small quality-of-life improvement that could be added: have the software timestamp when you say "here" and then (using a short low-FPS video buffer?) analyze the corresponding frame of video. This would eliminate the pause when you're pointing at something, waiting for the computer to respond.

Alternately a low-latency speech analyzer could look for the word "here" and quickly snap a screenshot. Perhaps project a temporary dot on the surface, so the user knows they can put their hand down now?

Quite the impressive demo! There's huge potential here.