A parametric rate-distortion model for video transcoding

Over the past two decades, the rise in video streaming has been driven by internet accessibility and the demand for high-quality video. To meet this demand across varying network speeds and devices, transcoding is essential. This paper introduces a parametric rate-distortion (R-D) transcoding model that predicts transcoding distortion at different bitrates without the need for re-encoding. Experimental results validate the model’s effectiveness in predicting rate-distortion behavior for diverse video content. Using our model, visual quality (measured by PSNR and VMAF) of transcoded video can be improved through trans-sizing. Moreover, our model can identify visually lossless bitrate ranges. This allows service providers to adjust target bitrates with minimal quality loss. Experimental results validate the model’s effectiveness in predicting rate-distortion behavior for diverse video content. By using the VMAF measure, our model achieves a quality improvement of up to 2.55 and bitrate savings of up to 79.10%.