Felix Fernandes, Samsung, noted that there are four means to modify power consumption via green metadata:
Modify power (i.e. computational complexity) in decoding algorithm
Modify display (or audio amplifier) power consumption
Media selection (i.e. encoding) to modify decoder power consumption
Modify presentation to mitigate aspects of power reduction
Reducing decoder complexity
Metadata could signal e.g. number of MOPS required to decode a fame. As a result, the clock of processor could be reduced, or it could go into sleep mode.
Audio could restrict bandwidth of decoding, but it is not clear if this significantly reduces power consumption.
In a point-to-point communications session, the decoder can request that the encoder send a lower-power (lower complexity) coded representation.
Reducing display power consumption
Decoder can reduce brightness of display and compensate by increasing dynamic range of pixels toward full scale. Might work similarly for audio amplifier and signal dynamic range.
Switching to a lower complexity bitstream
In a DASH context, metadata can indicate the power consumption associated with each representation
Compensation for possible quality degradation associated with low-power bitstream
For example, using prediction from high-quality frames to improve lower-quality frames
In video it is envisioned that only metadata is normative, and how to use it is out of scope.
In the CfP for green metadata for Video, it was required that responses demonstrate that metadata could in fact lead to reduced power consumption.
