MCF: Could big frames get too up standard cluster size?

andy

I have red main MCF specification doc, pretty interesting and well made (both format and documentation ).

I have red that cluster size should be closest as possible to 2-4 Kb (for best data check/recover...), but if a single cluster must contain at least one (key)frame, with high-bitrate video we could have very big frames...
So, Does cluster size irremediably grow?

Maybe the rule of clusters in MCF has too hard compromise between the target of data grouping/seeking and target of CRC-partitioning...  (clusters can get too big)

Interesting the Track Entries behaviour, will be useful at .MCF build time to add each new seek pointer every ~1-2Mb of cluster data , must say simple but smart, that index.There are speculations regarding when the format will be finalized/closed?
(despite library status)

unplugged,

thanks for your input, forwarding to MCF dev team for clarification ...

Thanks for your input - feedback is always a good thing

I don't know where from you picked that 2-4 Kbit from, but in our specs it says:
lt;Bgt;Too small Clusters result in some wasted space, but too large Clusters reduce data recovability and prevent error checking on fly. Sizes between 5 Ko and 2 Mo are okay, but using sizes outside of that boundary should be avoided.lt;/Bgt;

So, frames can be up to 2 Mo, and the default maximum cluster size is still not reached. Uncompressed frames can be bigger than that too, of course, but there is no real maximum size. In practice, the maximum size is only limited by the amount of free memory for MCF demuxer (whole Cluster is loaded to memory at once, to speed up things). In theory, the maximum size is 4 Gio.

Adler-32 can easily protect large chunks of data, without losing significant amount of reliability. Partial frames are useless, so that's not a reason to have frames split over multiple Clusters either..