sound frame lost to Compressor bug

I'm using FCP 5.1.2 with Compressor 2, so this bug would need to be verified with later versions.
Starting with a DV-PAL project in FCP, choose Export Using Compessor.
In Compressor, choose Advanced Format Conversions and then choose, for example, DV-PAL. Submit.
The output is a .mov file. Open it in QT and then open Movie Properties.
Movie Properties shows that the soundtrack is one frame shorter than the picture.
It's always exactly one sound frame lost, independent of the project length.
I haven't checked whether it's the first frame or the last frame.
Odd, isn't it?
Sound exported separately using Compressor does not lose a frame.
This happens when picture and sound are exported together in a .mov file.
I don't know if it happens with any but DV-PAL projects.
Any explanations?
(Please don't answer that there is never a reason to export a .mov file from FCP using Compressor. There are some.)

Dennis Couzin
Berlin, Germany

Serious, if mildly off-topic, question: Why would you ever choose "export using Compressor?" I've been using Final Cut for nearly a decade, and I've never once done that.

> Please don't answer that there is never a reason to export a .mov file from FCP using
> Compressor. There are some.

Nope, as with Jeff Harrell, not in my years of experience. I've used that feature early in Compressor's lifespan. It's clunky, prone to errors, and frankly has no upside. It forces FCP to sit idle while Compressor does its thing, a process which is basically a waiting job that could have been done as a background process while you continue to run FCP.

If Export - Using Compressor causes problems but Export - QuickTime Movie doesn't, why would you persist in using Export - Using Compressor?


www.derekmok.com

Oh wow, you guys are hurting me here.

I always use Export using Compressor, for just about everything I do other than laying off to tape, which i haven't done in years.

That said, i only use it for my master export. Simple checks and rough cuts i go straight through QT but for quality and flexability, Export through Compressor can't be beat.

First reason, it goes to the source media for every clip and encodes for your destination codec. Example. He's working in a DV timeline right? Let's say he's going to DVD. Let's say he has some simple name supers. He renders in his timeline to DV, then Compresses from DV to Mpeg-2, and the those names will look absolutely horrible. But let's say he went through compressor straight to Mpeg-2 or some other master codec. Then FCP re-renders those effects in 4:4:4 space for the final destination and you get crystal clear lines and filter effects.

Also, try resizing from HD to SD by going straight through FCP and QT. You won't get great results. But through Compressor you will.

You can change frame rates, crop the edges of sides for web delivery or center punching to 4x3, you can add a limiter on your audio to maintain consistency without going to Soundtrack, you can add watermarks, even animated ones, run custom scripts on the end of your exports, job chain exports together, etc.

Let alone the ability to export to multiple frame sizes and destinations the way batch export never could.

Man, I love Compressor, it's prob my 3rd fav app after FCP and Motion. Does it have problems? Of course but there's always work arounds. Does it take over FCP during export? Yep, but you only use it for the master export. Does it delivery the best kick ass quality? If you know what you're doing, absolutely it does. And i wouldn't be doing any of this if i wasn't striving for the best quality on my deliverables.

> But let's say he went through compressor straight to Mpeg-2 or some other master codec. Then
> FCP re-renders those effects in 4:4:4 space for the final destination

If his source media was DV -- which would be the reason he might be editing using a DV timeline codec -- then where would he get 4:4:4 chroma subsampling from?

> Also, try resizing from HD to SD by going straight through FCP and QT. You won't get great
> results.

He wouldn't be going through FCP for that operation, either. Which is our argument -- FCP's Export - Using Compressor is not useful. Bypass FCP, go straight into Compressor.


www.derekmok.com

Scott does bring up good points.

Apple just threw up a applecare doc last week about potential problems with export to compressor when using Virtual Cluster.

[support.apple.com]

Cure is to not use export to compressor.



Michael Horton
-------------------

FCP filters and generators all process at 4:4:4 initially, and then to the sequence codec. So if you go straight to QT, you're stuck with the sequence codec. Say if you make titles in Motion or have some cool intro clip from AE thats in Animation, once you render your sequence, you're stuck with that crappy DV color space.

But when you go out through Compressor, it starts off at that pristine level again and then goes into your final destination codec, say ProRes or MPEG-2 or H.264. You're not locked into your sequence codec is my point.

Again on resizing, you're working in a DVCProHD sequence, you export out of FCP, take that and compress for iPod or DVD. You're titles will look like crap. DVCProHD can't handle text the way ProRes or Animation can. But export out of FCP using Compressor, you've freed yourself from that crappy video codec for your end HQ mastering codec.

Anyway, I deliver 95% of the time to web or DVD and i want my work to look good, therefore I use Export Using Compressor. If it doesn't work with your workflow, I understand but don't knock it cuz just because it's slow. It's slow cuz its getting you quality. And outside of FCP, it can be wicked fast with virtual clusters which, as Mike pointed out, don't work with Export using Compressor, but if you have a proper cluster and shared storage, will work!

Since I started using virtual clusters, and they are blazing fast, I do not export to Compressor, but I export a QT movie and encode that.

However, I usually render out as Uncompressed/ProResHQ/DV50, because usually my source starts from a more compressed codec (DV/XDCAM EX/ DvcproHD) as Scott mentioned- because footage gets compressed back to whatever sequence codec you are on.



www.strypesinpost.com

Yeah, I guess most of that kinda makes sense, if you're editing on a seriously compressed timeline. Since I never do that ? my timelines are always uncompressed or ProRes, which is close enough to uncompressed that I've never been able to see the difference ? I guess I've never had a need.

The 4:4:4 thing, though, seems silly to me personally. Unless you're somehow trying to do a DI on your Final Cut system or whatever, you're always going to be going out to a 4:2:2 (or worse) format anyway. So jumping through a lot of inconvenient hoops to get Final Cut to render in 4:4:4 strikes me as a waste of time.

But obviously reasonable people can disagree.

>So jumping through a lot of inconvenient hoops to get Final Cut to render in 4:4:4 strikes me as
>a waste of time.

It does it automatically. On render, footage is decompressed to 4:4:4 YUV, rendered filter by filter, then recompressed back to your sequence codec. Going straight out to Compressor means that your footage is decompressed, rendered, then compressed straight to your delivery format.



www.strypesinpost.com

Right, I should have been more specific. The inconvenient hoops include having to tie up my edit system while the job runs, being unable to offload the job onto my Compressor cluster for faster processing offline, and still having to go back and export a Quicktime movie for my digital master anyway. It's just too much extra messin' around for the benefit, in my personal opinion.

Yea, I dropped it the moment I started setting renders to multiple cores. There's little or no difference if your intermediate codec is Uncompressed, even post CC. The trade off between time and quality is marginal, which means that exporting a SCQT is a much more feasible option.



www.strypesinpost.com

> and still having to go back and export a Quicktime movie for my digital master anyway.

The key is that most of us don't do exports solely to the web, and even less to just DVD. We'd usually have to go to a tape format first.

If graphics and chroma subsampling were such an important element, I would have questioned editing in a DV sequence to begin with.


www.derekmok.com

Glad I included that parenthetical at the end of the original post.
Thank you Scott Erickson.
My latest reason for exporting using Compressor was to take advantage of its motion compensated deinterlacing. I think it's best to deinterlace before other format-changing operations.
Concerning making 4:4:4 from 4:2:0 (or from 4:2:2 for that matter), it's interpolation and never perfect. But some interpolation is better than others. I think the the Y, U, and V data should be used together when interpolating each of U and V. The Y is known at every pixel and chrominance gradients generally correlate with luminance gradients. Upsampling to 4:4:4 is analogous to deinterlacing, and also desirable before other format-changing operations. I don't know if there's advantage to doing them together.
Concerning Derek Mok's questioning "editing in a DV sequence to begin with." One edits what needs editing, and makes the best of/from it.

Dennis Couzin
Berlin, Germany