Quality-Based Bit Rate Ladder Constraints for Smartphones Based on the VMAF Metric

The use of video streaming services applying adaptive bit rate (ABR) streaming is growing continuously. In ABR streaming, digital video signals are encoded at various bit rates and associated qualities, which can be assessed using the objective metric Video Multi-Method Assessment Fusion (VMAF). These are provided to a variety of end-user devices via the Internet. In this paper, previously determined constraints for a 4K TV quality-based bit rate ladder design are examined for their applicability to smartphones. The aim is to maximize subjective quality while minimizing storage costs. Subjective tests were performed on a smartphone following international recommendations. A constrainted parameter is the lowest VMAF score at which a video signal is on average subjectively indistinguishable from the reference video signal. This VMAF score was determined to be 95 for 4K TV and the smartphone test shows 92. Another parameter is the maximum difference between two VMAF scores where the associated subjective quality is approximately the same on average. For 4K TV, this difference was set to 2, and the smartphone test reveals three differences depending on VMAF score: 3, 5, and 7. Overall, the tests show that the quality-based bit rate ladder constraints set for 4K TVs are applicable to smartphones.