Thanks for that information, doublebishop! I’m glad to hear that- I’m trying to think of ways to guide the viewers eyes (even though they can still look anywhere), so for instance, if I’m pushing the camera in a certain direction, having it face the direction it’s moving would mean they’ll probably not be looking behind so much, and this will mean seams will be completely behind them. As quollism pointed out, it’s challenging to plan out shots when so many of the rules of composition no longer apply- but I’ll still try to make the experience cinematic if I can. And it’s worth mentioning, even if the viewer can look anywhere, I’m sure it’s still possible to guide their vision in the direction you want. Learning this new medium will probably involve a lot of that, if it becomes a mainstream art and not just a gimmick.
Here is one of the final shots (looped, of course). The others are still rendering; we’ve been having constant lighting storms here, which makes it difficult to keep the computer on for more than 6 hours at a time. This particular shot took 19 hours to render.
I’ve slowed down the camera, but does it still move too fast? This video has a bit rate of 9 Mb/s. Does it lag at all? Do you think I could push it higher? If I slow the camera movement down, that would probably improve the image quality a bit without having to increase the data rate.
The last shots of the video are rendering now. I’m hoping to have everything completed and uploaded on Tuesday. It’s taken between 18 and 25 hours to render each shot, but the results look promising. I’ve changed a number of the cameras, in the hopes of making the sequence more interesting. I’ll certainly be looking to improve my rendering capability in the near future- I’m sick of waiting for shots to finish!
I’ll post the video here as soon as it’s completed.
Had a look at your video, haven’t thrown it into our oculus / cardboards… but did look at the video.
Few tips.
– Having it bounce from going forward to backwards will feel disorientating…
– Not sure how you are rendering it out / compiling it, but our process has been to render out a jpg image sequence then compile with ffmpeg, gives more control over the encoding process (also makes interrupted renders nicer to work with)
– 24fps won’t be high enough for vr… you should be aiming to render off higher
– The resolution 1280x720 probably wont be good enough for vr either.
Not sure how much more i can say publicly… shoot me a PM if you want to talk more.
quollism - Thanks! I’m looking forward to hearing how it plays!
doublebishop - Thanks for your tips! I had to make compromises in order to get the render times down, the last 3 seconds of the film are rendering now. It’s taken about 110 hours of rendering to get to this point, and since I want to upload a new video every two weeks, I couldn’t push it much further. As I upgrade my hardware, I hope to render new VR videos at 1920x1080 and at least 30 fps.
I’ve uploaded the finished video to my channel just now, here it is:
I’d love to hear feedback on how it played. If anyone would like a link to the original file, I’ll put it up on my Google Drive as well! I don’t know if downloading the video from a site like KeepVid causes a noticeable drop in picture quality.
Hi I’m not sure what thread to start this conversation but thought this could be a good start.
Does any one know of a tool that allows me to track an object at the top of an equirectangular video and then replace it / patch it with another area of the video?