Multi-viewpoint robotic camera system creates real ‘bullet time’ slow motion replays #DigInfo


This multi-viewpoint robotic camera
system, under development by NHK, links the motion of eight sub-cameras
to that of an individual camera, so that all the cameras film the
same moving object. Each robot camera has two motors,
for pan and tilt. The cameras also share lens data,
so they can zoom in unison. This multi-viewpoint robotic camera
system can also be used as an image capture system for integral 3D TV,
under development by NHK. By generating integral 3D video from
multiple-viewpoint footage, 3D video of sports events will be
viewable on integral 3D TV. DigInfo News

69 thoughts on “Multi-viewpoint robotic camera system creates real ‘bullet time’ slow motion replays #DigInfo

  1. As exposure to 3D design increases it seems as though the effect that this camera system allows is simply a natural progression in the way we view media. Fantastic!

  2. the idea has been used many times for example with gopro cameras, the only new thing is the syncronized tripods.

  3. on the matrix movie this was computer generated and rendered. this time its happening in real time without image processing

  4. It doesn't create "real" bullet time effect, time still goes on at the normal rate. What it does do is allow for the bullet time effect to be done in real-time without needing to plan the shots beforehand.

  5. Nope, at least for the first one, they actually had a shitload of cameras (though it was on a greenscreen set, the background was added later). Somthing they did do though was run the footage thru a temporal smoothing algorithm to remove that "flickering" effect and create the illusion they were using an even higher number of cameras.

  6. The difference here is the cameras aren't on a fixed rig, allowing for the bullet time effect to be applied to things occurring in unpredictable positions.

  7. I actually found it even more interesting that we are getting closer to lightfield (AKA true holographic) screens.

  8. Think of this in terms of resolution. This demo has a resolution of 8 cameras. Imagine 100, or 360 covering every angle. Instant replays would be amazing.

  9. Remember when we where a little kid watching TV… somethimes we bend over to te left or right like it would change the view. Shit just became real.

  10. This is so awesome, multi point view slow mo haha, u could actually watch an olympic gymnast do a vault while reviewing the landing and mid air flight in like 200 degrees if not more.

  11. so with lightfield video cameras, shooting 3D with interpolation.. you could in theory get as close as being there. Obviously needs to be developed for sports, and used to avoid bad plays. But then the interpolated data would get disputed. Sigh.

  12. and ok.. so if you combine the images from any two adjacent cameras you get 3D essentially for free. I thought they had to begin a special set distance for it to be the most effective. interesting.

  13. This will be better than being at the event! there are hardly any replays or varied view points at events… except of course if the seats at events are also advanced to have individual screens like in airplanes, where you can control the replay and angle of view from your own seat! The future is here!

  14. Do they necessarily have to turn the cameras like that? Later on, when the processing is faster, they should just use wide angle lenses, that way, it will be easier to have a lot more than 8 different angles to render the 3d image.

  15. With something like a Football game why not have two controls one for side on view and one for top down so the camera's can be as far apart as they want while easily being able to focus on the same point or in a film have a marker that all the camera's follow with help from software. I think these 2 ideas could be used do develop this further and get less discrepancies. I do physics and we have to think up better methods of doing things that have issues in experiments.

  16. bullshit technology infrared based recording to get an acurate 3d using a 27 arrayed camera is the way of the future infrared pinpoints object cameras auto track objectsno need for human cameraman

  17. i think it is useless unless they will use any video taken by middle cameras. otherwise they can use 2 video cameras (for the two poles) and rest of the system can be -photograph- camera. it would be economical

  18. apparently it was the only way for a japanese guy to dunk. the regulation size is something like 3.05m (i REFUSE to calculate in feet,yards,inches and shit, there is only METERS and KILOGRAMS).

    so no living japanese can dunk in 3.05m rims/hoops.

  19. Sure there must be something new in this besides the mentioned use case. Nevertheless, for the comments of people wanting to see how this would work in a real transmission, search for EyeVision which was featured in a Super Bowl.

  20. I imagine sony has already made a 360 degree lcd screen. the flexible tv screens you could just wrap around a wide enough pole so when you walk around it you can see every angle from using these cameras. that would be amazing

Leave a Reply

Your email address will not be published. Required fields are marked *