It seems like this isn’t an easy problem to solve without lots of money to pay for hosting… but I had the thought recently, what if a video heavy fediverse platform was designed from the ground up with the intention that people were going to remix video content and that those remixes/tweaks could be saved and transferred across the network as modifications of a pre-existing large video file instead of needing to copy the entire video file onto to some valuable and scarce storage again just to slightllllyyy tweak it?
For some content you might not even need to create a new copy of the video file at all, not even a clip, the “new video” could just be specific instructions for how to playback portions of the original video file with certain additions/tweaks.
I know this is abstract, but I think there might be something to treating the way certain videos and concepts become “meme-ified” as a logical and natural way to structure intelligent minimization of unnecessary data use. A “montage” from a livestream or long video could just be a series of playback instructions for a pre-existing video file or video files with some tweaks and additions like text overlays or different audio.
It would have lots of practical benefits too, say a professor uploaded a video and needed to make a correction to a small part of the video where she stated something incorrect in her lecture. If the edit to the video involved her creating a “new video” that just had instructions to this hypothetical video player to play the original lecture video and then insert a brief new video segment with the correction (or just several frames of a textbox with the correction), or to just simply cut that section of the video from the playback with a simple fade… might that be a much quicker process than creating an entire new video file and re-uploading it and better than just settling for a comment in the video overlay/in the video description?
shrugs I know it is just a vague idea but has anybody tried something like this? (in practice maybe it would look something like a really simple video editor built into a youtube-like interface?).
A very simple example of this for picture memes would be a meme creation and sharing fediverse software where each meme format was only ever represented by one picture on the server, and every meme created of that type was just vector graphics style instructions for how to overlay text on that image. You could store a massssive amount of memes with a tiny tiny tiny amount of storage space right?
I know it would be difficult to probably make a seamless playback experience without small hiccups at times, but fundamentally would the end user need to necessarily re-render a video if the instance of this “youtube-like” they were on had a copy of the original video file that was remixed?
in the end the user-agent still needs to re-render the video in some capacity otherwise there will be only static the user would see :)
So it comes down to: do you want to put the load on the network (pre-rendered video) or on the computer (re-render video according to instructions)
Yeah this is kind of an unavoidable consequence of the video format and the types of compression applied to make videos into a file size small enough for distribution… but I wonder if this is an inherent aspect of displaying video on your screen?
Isn’t this just considered a fundamental problem because of the ways people traditionally have approached video compression/storage/bandwith minimization?
I.E. you can’t reach into an .mp4 and just pluck out a certain part without re-rendering the whole thing right? Whereas if it was a film video you could physically cut and splice a section of video out without having to rerecord the whole video sequence to a new strip of film…
I mean yes, to some level I am suggesting more processing work done on the client side, but when talking about video are we fundamentally talking about a problem where anything other than a single continous video file has to be “re-rendered” in some processing and data heavy process on the client side?
At what point is a quick and fast video player that quickly plays sections of other video files (a sort of veryyyy streamlined video editor I guess) not “re-rendering” a video but simply nearly seamlessly starting and stopping a video at different points or starting and stopping videos from different sources?
I have a question about this. It’s hard for me to follow because I am not all familiar with it but from how you are describing all the work being done client side, wouldn’t that mean lots more energy use on every computer downloading it instead of at the source? Isn’t that already a problem we face because of bitcoin and AI?