FIG. 4A illustrates using a distance-based cost function to determine an optimal encoding ladder for a bitrate-quality data space of a single video segment. Specifically, three encoding profiles, profile 471, 472, and 473, are represented by the stars in FIG. 4A. Profile 471 is associated with the pruned trial encoding outputs for encoding profiles that include a value of 1950 for the maximum bit rate encoder parameter setting. Profile 472 similarly corresponds to the 5000 kbps maximum bit rate, and profile 473 similarly corresponds to the 9000 kbps maximum bit rate. In addition to optimizing the bitrate-quality trade-off within a particular maximum bit rate encoder parameter value, the distance-based cost function utilized in FIG. 4A may be configured to account for bit rate-quality trade-offs across the three maximum bit rate settings by minimizing the sum of the distances of the three star points illustrated by profile 471, 472, and 473 to performance boundary curve 465. The distance-based cost function jointly minimizes the aggregate distances (e.g., the sum of the three vectors 471a, 472a, and 473a that are perpendicular to performance boundary curve 465) and the per-maximum bit rate distances (e.g., the individual vectors 471a, 472a, and 473a that are perpendicular to performance boundary curve 465). For example, the cost function will select encoding profiles with distances for vectors 471a, 472a, and 473a that equal 1, 1, and 1, rather than encoding profiles with distances that equal 0.25, 0.25, and 2.5, even though both sum to the same value of 3. This is because, for example, the distances of 1, 1 and 1 correspond to a lower peak deviation from the performance boundary curve for each maximum bit rate setting. One of ordinary skill in the art would recognize that a wide variety of optimization algorithms can be used to determine the encoding profiles corresponding to profile 471, 472, and 473 used to optimize a distance-based cost function. For the example of FIG. 4A, algorithms including but not limited to least-squares, linear programming, and variants of heuristic-based search algorithms can be used to perform convex optimization. In another class of implementations, a cost function may be used that minimizes the per-maximum bit rate distances (e.g., the individual vectors 471a, 472a, and 473a that are perpendicular to performance boundary curve 465). In those implementations, the cost function may sequentially select the encoding profiles (e.g., profiles 471, 472, and 473). If desired and in those and other implementations, the cost function may also include one or more constraints to ensure that the selected encoding profiles are spaced apart from each other (e.g., to ensure that no two profiles are so similar as to be duplicative, nor so far apart as to omit valuable intermediate profiles). As an example, the cost function may include a constraint that each adjacent (along the bitrate axis) pair of encoding profiles should be separated by a VMAF differential of approximately 6 (e.g., since a change in VMAF of 6 may be considered the smallest change resulting in a user-observable different in video quality). As another example, the cost function may include a constraint that each adjacent pair of encoding profiles should be separated by a VMAF differential of no more than approximately 12. As another example, the cost function may include a constraint that each adjacent pair of encoding profiles should be separated by a bitrate differential of 20% (or some other factor such as 10%, 15%, 25%, 30%, etc.). If desired, the cost function may include a constraint that each adjacent pair of encoding profiles is separated by no more than a bitrate differential of 40% (or some other factor such as 20%, 25%, 30%, 35%, etc.).