Hacker News Reader

Politics aside, is this potentially one of the most comprehensive multi-vantage-point training video datasets available today? I cannot imagine another that contains such a multitude of 1) videos shooters, 2) positions, 3) actual humans and 4) temporal start/end times.

View on HN

Now imagine letting lose SfM and tracking algorithms on this video dataset and reconstructing full temporal 3D model of the event ala Enemy of the State (1998). Add in some of [1] and we could reconstruct full Cyberpunk 2077 Braindance.

[1] https://arxiv.org/pdf/1810.04826.pdf VoiceFilter: Targeted Voice Separation by Speaker-Conditioned Spectrogram Masking

I would imagine so, great multiangle dataset w crowd sourced labels. Grim, but I do look forward to the meeting of all those videos and a "matterport" style reconstruction. I assume the Nytimes is paying some studio to do this right now.

Reply on news.ycombinator.com