I think you meant video decompression? Compression should theoretically be a lot more efficient, since the algorithm is a lot simpler and you can probably do a lot of SIMD to calculate the filters themselves.
In compression, "efficiency" usually means "compression level".
If by "efficiency" you mean "speed", then yes, I think OP's approach can be much, much faster than usual video compression approaches -- but this is true of all compression algorithms that don't compress very well. The fastest such algorithm -- leaving the input unchanged -- is infinitely fast, but achieves no compression at all.