Do you know what kind of compression ratios you get out of curiosity? Presumably it would be lower than this format (because this format is meant to be lossy not lossless) but very curious as a baseline.
A gaussian gif. Coming to porn sites soon. The file format name works for them. But imagine whole movies shot in this format. The fly-on-the-wall fantasy of movies, without being locked in place for the whole shot. Narrative possibilities like being able to examine just how close the Tyrannosaurus Rex is to your rear view mirror. Clues in a mystery only visible on rotation. Hidden bonus scenes.
I hope this catches on just to be able to watch the evolution of cameras to capture it.
>imagine whole movies shot in this format. The fly-on-the-wall fantasy of movies, without being locked in place for the whole shot.
That would be terrible. Framing is the major expressive feature in cinematography, and any interactive format needs a lot more thought put into it than just having a free camera. Literally the worst of both worlds.
Light field video streaming is a thing, however it's pretty niche. OTOY pioneered holographic codecs and light field videos with some degree of freedom and a sense of depth more than a decade ago.
It wouldn't work for a movie, but it might be good for a concert video, stage play, dance, or sports, where you're capturing something happening in a defined space. You could pick a different seat in an auditorium.
Also consider connecting a virtual space with a real one, so it looks like something happening on a virtual stage that's connected to a room in your house.
A$AP Rocky did it for a music video. Granted, it was only used for the editing process and not the final video, but it still presents some interesting opportunities, though it doesn't really look real so the applications may be limited.
Ok but you're commenting on the general concept of animated gaussian splats. That's existed for a while and it's unrelated to what this actual post is about which is a new compression method.
I suppose it's the difficulty in recording them rather than the file size that has kept them experimental. But that's not a problem for AI generated splats. GenAI sites could have a toggle from image to video to splat4d.
That already exists though. I believe Braindance VR uses a rig with a couple dozen cameras to capture the same scene from multiple viewpoints then converts it to a gaussian splat that can be walked around.
Its great for porn for those videos when the camera seems to be focused on the actors balls when thats definitely not the part I want to be looking at. I can just look around the room instead.
Something’s way off with these numbers. The page says it encodes video at 640MB/s which is quite large even for 4D data and doesn’t match the filesize of the demo splat (7.4MB / 2sec, or ≈3.4MB/s).
In fact they say the raw file size of the demo splat was only 427MB, so maybe the 640MB/s was a statement about encode speed? Why write it that way instead of “this demo splat was encoded in 0.6sec” or even just “the time to produce the original splat took longer than the time to encode this video format”?
