What is Rate-Distortion Optimization?
Rate-distortion optimization evaluates encoding choices by considering both their data cost and their reconstruction error. An encoder uses that comparison to select an efficient result for its objective.
How Rate-Distortion Optimization works
RDO gives an encoder a common objective for decisions that otherwise trade smaller output against closer reconstruction. For each candidate—such as a prediction mode, motion vector, partition, or transform—the encoder estimates coded bits and distortion, then combines them using a weighting factor. This search sits inside compression rather than delivery, and deeper candidate evaluation can improve bit allocation while consuming substantially more encoding time.
Key facts
- A common decision cost is expressed as distortion plus a multiplier times rate. The multiplier controls how strongly the search penalizes extra coded bits relative to reconstruction error.
- RDO can compare modes that look similar but signal very different amounts of side information, including partition choices, motion data, and residual coefficients.
- The chosen distortion metric shapes the result. Simple sample error is efficient to evaluate but may rank texture loss or structured artifacts differently from human perception.
When Rate-Distortion Optimization matters
Enable RDO when better visual quality at a target bitrate justifies additional encoding computation. A faster mode may reduce processing time but make less efficient choices about bits and distortion.