For certain values of "most" and "highly", and depending on how much compression or latency you're willing to sacrifice for more parallelism.
The more data (from earlier in the bitstream/image) you can use for prediction, the higher a compression ratio you can achieve. But the more data you use, the more sequential you become, since you have to wait for it to become available before proceeding.
Modern codecs generally don't require a single frame to be parallelizable for encode or decode, and you must sacrifice either latency (encode/decode multiple frames at once) or compression (split a single frame up into parallelizable slices) to become more than slightly parallel.
The more data (from earlier in the bitstream/image) you can use for prediction, the higher a compression ratio you can achieve. But the more data you use, the more sequential you become, since you have to wait for it to become available before proceeding.
Modern codecs generally don't require a single frame to be parallelizable for encode or decode, and you must sacrifice either latency (encode/decode multiple frames at once) or compression (split a single frame up into parallelizable slices) to become more than slightly parallel.