I'm doing some research for a project I will probably jump into early next year. It's a feature-length doc that will be shot on the EX-3 in January/February. The producer is projecting about 175-250 hour of footage will be rolling in.

All the tapeless projects I've worked on up until now have involved less than 30-40 hours of footage, so it's been a no-brainer to convert everything to ProRes and never look back. 200 hours of footage is a different story, we might not have the resources to do a full RAID of 20TB of material. All the writeups I've seen have covered footage for a few days of shooting, not six weeks.

Is there a reliable workflow to create low-res proxies from the source files, then just pull full-res ProRes files for a finish, sort of like an old-school offline/online workflow? If anyone has done a large-scale gig like this, what did your workflow look like? Any advice, pitfalls, or nasty surprises to watch out for?

Also, has anyone heard of this & have any experience with it:

[pro.sony.com]


Thanks!

Jeff

>we might not have the resources to do a full RAID of 20TB of material.

I would probably transcode to ProRes LT in Compressor. Then batch capture the picture locked cut to XDCAM EX footage and transcode those to ProRes for finishing. You can also choose to finish in ProRes LT and skip the batch capture step. Personally, I will try not to cut in XDCAM EX as much as possible. With that amount of footage in HD, 12-15 TBs of editing storage is not unreasonable.

Preferably, you will want to break up your edit into parts, and not cut in one big FCP project file. This will keep the size of your project file down, so you get faster saves, better clip organization, and most importantly, you will avoid the dreaded "out of memory" error.

What I gather about the Cinemon plugin is that it allows you to edit the .mp4 files from the XDCAM EX footage natively. Not sure if I would want to do that, because XDCAM EX transfers pretty quickly due to the file size, and you will need a back up of the footage in most cases, so editing off your field drives on such a large project is a major hazard.



www.strypesinpost.com

> Is there a reliable workflow to create low-res proxies from the source files, then just pull full-res ProRes
> files for a finish, sort of like an old-school offline/online workflow? If anyone has done a large-scale gig l
> like this, what did your workflow look like? Any advice, pitfalls, or nasty surprises to watch out for?

Well, you're researching now, so that's already a good thing. The thing I'd do is to test the offline/online process. Grab a copy of some footage, transcode it to the intended offline format, cut it, add effects and speed changes. Organize the files the way you usually do. Then try to online it, see if anything goes wrong. Sometimes it's not even that there's something wrong with the intended workflow. An editor or assistant editor unfamiliar with the workflow can mess things up.

Remember to come up with a sensible backup system as well. Somebody in here said it well: RAID redundancy is not the same as a backup. You need both to cover yourself. And with a large volume of material, not only is the backup necessary, but proper administration of the backup system -- knowing where everything is, updating it, and documenting it.


www.derekmok.com

>An editor or assistant editor unfamiliar with the workflow can mess things up.

That's not an assistant editor. That's called a "villain" or an "antagonist". Always make sure all the staff on your show is well briefed on the workflow, especially if it is a big project. No one is supposed to assume best practices if there is an existing workflow.

Make sure you can manually match back the footage if eye matching is required at any point, so you may want to keep the name of the source clip in one of the log notes column. Also, you will always want to see how spanned clips behave in FCP. In FCP 6, spanned clips do not get assigned a reel name, so you may want to use a separate column to keep track of the reel names on spanned clips, if that is the case.



www.strypesinpost.com

Strypes, that's a good point about FCP project size. This is a documentary, so some thought will have to go into how to break the material up & not create an unwieldy project.

At the risk of sounding like an idiot, what are "spanned clips"?

Splitting up a project is good mainly because it preserves your sanity.

Spanned clips happen when a tapeless clip spans two cards or more. Usually it happens because one card is too full, so the data is "spanned" onto the next card, in some cases it is because the card runs on FAT32 (eg. Panasonic's P2), so long clips are split into 4 gigs per clip. FCP will then join the spanned clips together during Log and Transfer. In my limited tests with it, FCP was able to accurately batch capture even though the reel name is not accurately recorded. The problem is more to do with you figuring out which card the clip came from when you are trying to reingest. I'm not sure if this issue has been resolved in FCP 7.



www.strypesinpost.com