x265 chunk encoding usage

[kabelbrand]

I'm trying to figure out the proper usage of the --chunk-start and --chunk-end options introduced in version 2.9

I guess it works as described but what is the use case here? I thought this is intended for parallel processing but it seems ALL the frames before --chunk-start are fully encoded and then discarded.

So instead of reducing the turnaround time this is no faster than a single complete encode and the total amout of processing time is MUCH higher.

Any comments? Thanks.

[benwaggoner]

I'm trying to figure out the proper usage of the --chunk-start and --chunk-end options introduced in version 2.9

I guess it works as described but what is the use case here? I thought this is intended for parallel processing but it seems ALL the frames before --chunk-start are fully encoded and then discarded.

So instead of reducing the turnaround time this is no faster than a single complete encode and the total amout of processing time is MUCH higher.
It's used to allow rate control to have some state information about the frames before and after the segment being encoded. One would only encode a little bit before and after, not the entire file.

[kabelbrand]

Thanks Ben. Should this be used in conjunction with --seek and --frames then?

So if I wanted to encode frames 1000-2000 (24fps, 1s VBV Buffer, 1s RC Lookahead) I'd roughly use:
--seek 976 --chunk-start 24 --chunk-end 1024 --frames 1048?

[benwaggoner]

I'm not sure about the syntax, but I think that's the general idea.