For supported video types, Horizon is optimized in the sense that it recognizes videos and allows viewers to stream it in chunks for seamless watching.
For supported video types, Horizon is optimized in the sense that it recognizes videos and allows viewers to stream it in chunks for seamless watching.
- What if the user's hardware is not performant?
- Transcoding is intensive, so their machine will become slow
- Even wasm based ffmpeg will not be as efficient and has its own issues
- Takes a very long time
2. Not if you do as proposed
3. Efficiency is not your problem, as you've stated with the previous 2 requirements. Your algorithm doesn't have to be maximally efficient if it takes a while to compute (eg you're using sleep in it)
4. Doesn't matter, most users of this service will only be taking a few photos or videos a day. Offer a paid service upgrade for heavy users who cant handle the upload times.
In the (possibly near) future, I do plan on scaling to the NA to give us better latency.