Consider adding non-allocation converters/encoders

[paulomorgado]

Similary to Base64.EncodeToUtf8, IVideoEncoder methods that allocate a byte[] to return should provide methods that receive a buffer to write the conversion/encoding to.

[ha-ves]

I can't think of a use case for this library needing to utilize the design of that method.

Do you have cases it needs to be? Encoding efficiency tracing?
Because usually that kind of design is for when you want to compare the input and output data of an encoding.

I'm curious because depending on the IVideoEncoder implementor, the returned byte[] reference can be static and just be reallocated internally.

Or do you want an IVideoEncoder to write into "another object's memory"? (which seems to be just replacing the reference)

[paulomorgado]

This is for reducing GC pressure.

Explaining this, is way beyond a GitHub issue. If you want to learn about memory management in .NET, I recommend Pro .NET Memory Management (2nd edition). Reading .NET code and its usage is also very valuable: https://source.dot.net.

In my playground application with 3 feeds, I've reduced memory usage from 250MB to 100MB just by reimplementing the VideoTestPatternSource and VideoEncoderEndPoint. And I'm just starting.