AOM --use-16bit-internal

[macdavis]

Does libavif utilise AOM's API --use-16bit-internal during encoding? This could potentially preserve better quality especially for higher bits outputs.

[wantehchang]

--use-16bit-internal is a command-line option for the aomenc utility program. --use-16bit-internal only affects bit depth 8. For bit depths 10 and 12, --use-16bit-internal is forced to 1 and doesn't need to be specified.

For bit depth 8, I would expect that --use-16bit-internal shouldn't affect the encoded output. Any difference is likely to be not intended.

[wantehchang]

@jingninghan @jzern

Hi macdavis,

Correction: I asked Jingning Han (who leads the libaom team at Google) this question. He said that for 8-bit input, --use-16bit-internal needs to be used in conjunction with --bit-depth 10 to get better compression performance.