Does FCP handle HDV in 4:4:4?

[Alex Bowles]

Looking for some clarification on a statement made by Paul Saccone, Apple's product manager for FCP Studio (Film & Video Magazine, 10/05.)

With regard to editing HDV natively Mr. Saccone stated "If you take an HDV stream, whether you're doing color-correction or a 16-layer composite, we decompress all that video into a 4:4:4 color space, do our composites, and then do one single re-encode back down to the HDV format. So you're only ever incurring one generation of re-encoding."

My question is whether this is automatic, or is Mr. Saccone is glossing over a step? In other words, if I elect to set up a project using the HDV 720p setting (native, that is, to the 4:2:0 720p signal captured on my JVC HD-100) is the capture process automatically converting the footage into a version of the HDV codec that has a 4:4:4 color space? Or is Mr. Saccone describing a workflow in which I'd need to elect to capture the footage using another codec (e.g. the Apple Intermediate Codec) that includes 4:4:4 color sampling as part of its spec?

Assuming that the conversion is automatic, I could get into color-correction knowing that the material on my hard drive can support this sort of work. But if the latter is the case, it seems I'd need to make a specific choice when importing material if plan to do color-correction, and want to minimize signal degradation.

Can anyone out there offer any insight?

Many thanks,

Alex

[Tim Dashwood]

I would assume the 4:4:4 processing is automatic and set as the default when you choose the HDV 720P30 easy setup. You can check your timeline settings though and confirm if the default for video processing is RGB or High Precision YUV.

As a general rule, FCP always takes a "render once" approach, unlike Avid's old "render on top of previous renders" approach, so renders should always only be one generation.

My concern has always been re-encoding the HDV material back into HDV, even once. At a bitrate of 19.2Mbps, it seems like additional compression artifacts would be inevitable. That's just my assumption, and not based on tested results.

[Graeme Nattress]

FCP has to decompress everything to 4:4:4 (either Y'CbCr or RGB) to make things work for renders. It's automatic.

And yes, rendering back to HDV does make for nasty looking artifacts.

Graeme

[Damien Molineaux]

HDV is a beautiful capture option, but nobody (just about) is using HDV for distribution (projection or whatever). If you need an HD master, once you've finished editing, copy the elements in your timeline to either a timeline with uncompressed settings, if you're system can handle it, or a superior format, ideally your output format. DVCproHD is pretty good and only 100 Mbps (versus 25 or 19 for HDV). Personaly, my output is SD, so I copy everything to a 10bit uncompressed SD timeline which my system handles with no problem. If you plan on doing uncompressed HD, you will need a fast system with a at least a four disc SATA raid.

Cheers,
Damien Molineaux
earthling productions

[Alex Bowles]

Thanks all - this is v. helpful.

Regarding a return to HDV; I wouldn't, for exactly the reason Damien cited - it's a very weak mastering format. A relatively low bit rate is only the least of it. The tapes themselves are small and delicate.

Also, the format's native MPEG2 compression is interframe. By using it you're creating unnecessary artifacts by choosing a compression scheme that's been optimized for data transmission by minimizing information redundancy, whereas your task (mastering) is one where you're trying to optimize retention by maximizing information capture.


Instead of returning to HDV for your master, it's better to upconvert your material within FCP for output to a suitable master format (D5 or HDCAM-SR, and if necessary, DVCproHD.)

If publishing directly from FCP to DVD, going back to HDV (or any tape format) adds an extra step - far better to use Compressor to get your SD, MPEG 2 assets directly from your timeline, get into authoring without an extra set of conversions, and using your DLT's as archive masters.

Again, thanks for the insight on the 4:4:4 thing - one new question? How does FCP's 'render once' vs. Avid's work in practice? The name alone suggests an obvious advantage for Apple, but I'm curious about the actual differences in performance that result.

Cheers,

A

[Tim Dashwood]

Originally Posted by Alex Bowles

How does FCP's 'render once' vs. Avid's work in practice? The name alone suggests an obvious advantage for Apple, but I'm curious about the actual differences in performance that result.

Every time you add an effect that needs rendering in FCP, it will render everything through the chain for each frame. So if you have already added a few effects, rendered them, then decided to add one more, it throws out the previous work done on each frame and starts from the bottom up again. This can sometimes be very frustrating, especially if you are working with slow Magic Bullet filters.

In Avid, if you have previously rendered effects and add something new, it will apply the new effect to the old render. This obviously speeds up your workflow, but compounds compression artifacts on top of compression artifacts if you are working in any compression other than 1:1 uncompressed. I remember working on a few projects where the compression artifacts would get so bad on multiple layered sequences that it was just a screen of macroblocks. Mind you we're talking 20+ video tracks.

Both systems are very capable when it comes to real-time effects, so workflow speed isn't really affected if you have the appropriate processing power. FCP is a little more "dummy-proof" whereas an Avid editor would want to use media manager to delete all renders at project completion and then render everything from the top level to achieve the least compression artifacts.

Of course, in uncompressed codec Avid has the workflow speed advantage.

[Alex Bowles]

Exceptionally helpful stuff. Many thanks.
AB

[Tim Dashwood]

BTW, I found this pdf from Apple that clearly states their opinion on native vs. non-native workflows, and also confirms 4:4:4 processing of 4:2:0 before re-encoding back to 4:2:0.

http://images.apple.com/pro/pdf/L310570A_HDV_FAQ.pdf

[Evan Donn]

Now here's a related question I haven't seen a definitive answer to - if I capture native HDV, edit/composit/effect/correct etc, then export from the timeline to compressor for output to another format for mastering, does Final Cut render each frame and send the uncompressed 4:4:4 rendered frame to compressor for output? Or does it render and compress to the sequence format and then send that to compressor? It seems like the former would significantly improve the quality of rendered sections (which may be the entire video if color corrected) as well as speeding up the whole process by eliminating an unnecessary compression step.

[Graeme Nattress]

Sending direct to compressor does exactly that, but each frame ends up getting rendered multiple times and it's incredible slow.....

Graeme

[Alex Bowles]

[quote]BTW, I found this pdf from Apple that clearly states their opinion on native vs. non-native workflows, and also confirms 4:4:4 processing of 4:2:0 before re-encoding back to 4:2:0. http://images.apple.com/pro/pdf/L310570A_HDV_FAQ.pdf [quote]

Just checked it out - good stuff. But my takeaway is that you're resampling all your 4:4:4 post work (renders, effects, filters, etc.) when outputting - unless you're mastering to a format that supports full 4:4:4, such as HDCamSR. If you were to go out to HDCam or (I believe) D5, you'd resample to a 4:2:2 space instead, only going to 4:2:0 if you were taking the unorthodox step of mastering to HDV.

Thoughts?

A