Working with HDV in Final Cut Pro
COW Library : Apple Final Cut Pro Legacy Tutorials : Jerry Hofmann : Working with HDV in Final Cut Pro
When HDV appeared, the market was overjoyed at the thought of shooting and editing in HD with cameras that cost less than a new Mercedes Benz. Downconversion of HDV to DV is an available option with the cameras as well. HDV cameras can record DV25, and many of them can downconvert HDV recordings to DV during playback. It can be beautiful DV as well, easing the transition to HD with the format.
WHAT IS HDV?
HDV is an MPEG-2 recording format. Only a few of the frames that are in the recording actually contain all the information about each pixel in it. For example, if you shoot with Sony's HDV cameras in the 1080i format only 1 in 15 frames actually contain all of the data that builds an entire frame. In fact, Final Cut Pro can only deal with this particular flavor of HDV, called a "long GOP" (Group of Pictures) structure. This situation exists in all HDV recordings regardless of format, including formats that require transcoding. JVC and Canon cameras require Lumiere HD to work in their 24 frame formats, which use a shorter GOP structure, and FCP can't work with this media directly. This condition creates processor intensive overhead and possible transcoding procedures through a third-party application, thus rendering and set-up times might be longer, and extra disk space is required. But, the faster the Mac, the less painful the rendering time is and scratch disks are less expensive all the time.
Figure 1 - Easy setup for Native HDV 1080i60
There are two ways to handle HDV-acquired footage: one, in the native format; the other, by downconverting your footage to the DV format. For DV, all you do is set-up your computer to capture DV, and set your camera or deck to downconvert the HDV tape's playback on the fly to DV. Piece of cake.
To set-up your computer to capture and edit HDV natively, it's as easy as selecting the HDV easy set-up from the Final Cut Pro menu. See Figure 1.
When you want to capture, a specially designed Log and Capture window will open. See Figure 2.
Note the frame jog is gone, and while in the Clip Settings tab, you'll find that you cannot monitor audio using computer speakers during your log and capture process. When you need to do this, you'll need to monitor audio externally though self-powered or amplified speakers. Headphones also work by connecting them to the camera's headphone jack during playback and capture. The other difference between this Log & Capture (L&C) window and the standard one is that you cannot change the capture presets in the Capture settings window. This L&C window is exclusive for HDV native captures.
Rendering effects in the native format requires patience. There are more pixels that have to be read and the computer has to figure out just what each frame is supposed to contain, so rendering requires a lot of extra computing cycles and time adds up. How does it work? The computer has to regenerate its GOP (Group of Pictures) structure to render effects, record HDV back to tape, export a QuickTime movie, or burn a DVD. FCP calls it "Conforming". This conforming can take a very long time for a Mac to perform.
Figure 2 - HDV Native Log and Capture Window
In my tests with a 5-minute sequence containing one Motion clip, and 6 simple 10-second titles, a little color correction on each of the clips, a rendering of those elements took 34 minutes to "Conform." Then, when I selected Print to Tape, another 9 minutes of conforming took place. (Total 43 minutes for a render and record setup for a 5-minute simple program). I'll compare this same timing with identical sequences with other codecs later in this article. Edit to Tape isn't supported in HDV, you must use Print to Video.
DVD STUDIO PRO/COMPRESSOR
DVD Studio Pro 4 will not import a reference movie of a native HDV sequence to create an HD or SD DVD. However, Compressor 2 will import a reference HDV movie, and exporting directly from FCP to Compressor 2 is also supported. You can export it as a self-contained HDV QuickTime movie first, then import that larger file as an asset for your DVD in DVD Studio Pro 4. Of course, included in export time is the "conforming" process.
It's common to have to encode the HDV to MPEG-2 for the delivery of an SD DVD. On my test machine (dual 2Ghz G5), this took about 37 minutes for a five minute sequence exported directly to Compressor 2 from Final Cut Pro 5.0.4, using a one pass CBR encode preset. In DV, the job would have been done in about one third the time.
If you had a long form project, and were using a run-of-the-mill dual G5, you might as well take a mini-vacation while you wait for a dual-pass VBR process to complete, we're talking days here.
A 2.7 dual G5 machine running FCP 5.1.1 did considerably better with this. It took just about 4.5 minutes per-finished-minute of program to perform the encoding. On the same 2Ghz G5 it would have been a touch slower due to CPU performance differences. I can say that Apple's improvement with the Universal version of FCP 5 is definitely welcome. However it still renders slower than other more efficient codecs. But if you are considering working in native HDV, the "crossgrade" to the Universal version of FCP is a must. It's just about twice as fast "conforming" HDV.
An offline/online process is successful if you plan to finish in another format. The timecode on the tapes is preserved with the native format in the event of recapture.
A bright side to HDV is that the file sizes are small, they don't require an expensive disk array to playback the files, and there is no loss of picture quality involved editing natively. But in a world of "time is money", this situation simply may not work well for a professional. Tying up a computer for long rendering chores is a profit killer.
APPLE'S INTERMEDIATE CODEC
Apple's Intermediate Codec is an alternative codec used to turn the MPEG into a "friendlier" format. Setting up and using the AIC is as easy as HDV native, just select Easy Set-ups from the FCP menu and select the Intermediate codec setup which matches the frame rate you recorded. You may have to click on the "show all" button, which allows you to see the intermediate codec Easy Set-ups.
Once you select Log and Capture, no Log and capture window opens. The system simply starts capturing the video. There is no way to log material first with this codec, and timecode is reassigned to each clip.
RELINKING MEDIA, NOT!
Happily at each camera cut, the system creates a new master clip, but unhappily, they all start with zero hour timecode. Woe be to those who finish an edit, then somehow lose the source media. It would be unsuitable for an offline process because the timecode in the clips isn't what is on the source tape. You cannot recapture this material referencing the tapes timecode.
During capture, the AIC codec encodes the MPEG stream to an all "I" frame format, so you end up with an HD QuickTime movie that contains all the information contained in each frame of video. Gone are the long rendering and conforming times, however there are costs involved. It's scratch disk storage space, and the timecode assigned to these clips isn't what's on the tape.
In my tests with a side-by-side comparison between the native codec and the AIC, I could see a bit more noise in the Intermediate codec's playback especially if there was noise in the original HDV recording. Apple admits to this slight loss of picture quality.
Following, is a comparison of the storage space used for native HDV and Apple's Intermediate codec:
720p30 = 2.5 MB/sec. (9 GB/hr.)
1080i60/50 = 3.3 MB/sec. (12 GB/hr.)
Apple Intermediate Codec
HDV 720p30 = approximately 7 MB/sec. (25 GB/hr.)
Apple Intermediate Codec
HDV 1080i50 = approximately 12 MB/sec. (42 GB/hr.)
Apple Intermediate Codec
HDV 1080i60 = approximately 14 MB/sec. (49 GB/hr.)
AIC files are 3 to 4 times the size of the native media files. Yet they are not so large as to require faster drives other than an internal ATA, SATA, or even a fast FireWire connected drive on my dual 2gig machine. Render times are a lot faster, it only took 10 minutes to render the same five minute sequence.
When you're finished editing however, the AIC needs to transcode back to HDV native in order to print to video. On my dual 2Ghz machine, this took 38 minutes with my five minute test sequence. Edit to Tape isn't supported, but Print to Video is. Of course, you may not be delivering an HDV tape to anyone anyway, just depends on your workflow. You might want one for a back-up though.
HDV TO DVCPROHD
Converting HDV to Panasonic's DVCPROHD format is a valid alternative. The file sizes are roughly the same as Apple's Intermediate Codec and they are compatible with Panasonic's P2, and Varicam HD format, and so mixing the footage with these other sources should be a breeze if the need arises.
You can convert HDV to DVCPROHD with Compressor 2 or with a simple batch export. If you do this, you need to keep the same frame rate and size you shot your source footage with. In Compressor 2, you'll find the presets for DVCPROHD in the "Advanced Format Conversions" sub-menu. There are presets for 1080i as well as 720p.
The downside to this is the "double duty" spent in software compression after capture, as well as consuming more disk space. This, because you end up with the HDV files - which I'd not delete until I'm done - as well as the converted DVCPROHD files on your machine.
An added pleasure of working in DVCPROHD is the fact that it's a 4:2:2 color space. This is positive from the point of view that your graphics should look better, and the render times are a lot faster. With my 5-minute sequence, the render time (including the real time color correction) took 10 minutes (notably the same time as the Intermediate codec took). But this included the Full RT color correction, which in the AIC or native codec, had to be rendered because they weren't "Full" quality RT. This format handles a lot more real time effects than Native HDV. The file sizes are roughly the same as those for Apple's Intermediate codec. If it had been a real world job and the delivery was a tape, I wouldn't have had to render the color correction, and would have saved 5 minutes of the 10 used.
I really liked what I saw with the DVCPROHD codec. I sure didn't see any discernable loss converting the HDV to it, and it doesn't require those fast disk arrays to play. This could be a much better workflow to those without fast disk arrays and capture cards. In order to print back to HDV tape in the end, you'd need to conform your sequence to native HDV. (About 38 minutes on my machine for our 5- minute test sequence).
REALTIME CAPTURING & CONVERSIONS
By now you're guessing that I'd prefer another format than HDV for editing, and you'd be correct. I just can't abide the long renders involved in Native HDV. So here's my preferred method of working with HDV shot material. Capture cards: not only will they convert the HDV native to another HD or SD format, they allow you to monitor it externally as SD or HD and those of you who composite, understand the advantages of the 4:2:2 colorspace. I've long been an AJA user and ardent fan, so I'll talk about those cards, however Decklink and Convergent Design have solutions for this, as well.
AJA's LH series cards will capture the component analog HD signal from your camera or deck and transcode it to any flavor of HD as you capture it in real time. You could capture the HDV signal all the way up to Uncompressed 8- or 10- bit HD.
AJA's Kona 3 will do the same with an additional component analog to HD SDI converter (AJA's HD10A). The Kona 3 card is capable of cross conversions, so you can mix and match HD source material all you want capturing it to the same codec no matter what it was originally, as long as the frame rates are the same. What you do is run the signal from your camera or deck's component analog outputs to the HD10A, then from there via SDI to the card and convert this uncompressed HD signal to the HD format of your choice, as you capture it. The LH series cards already have component analog inputs, so they're ready to go. Both cards will downconvert to an SD signal and allow you to use a standard definition monitor externally, as well as send out a full uncompressed HD signal to an HD monitor.
You can preserve the original timecode using Kona and other capture cards. For feedback on the latest list, visit the Cow's FCP forum.
You'll use Apple's HDV device control setting for your Sony camera or deck for control and capture timecode through FireWire, or if you are using JVC's deck, you can use the 9 Pin remote control, and get the timecode that way.
Now we're getting places! Faster renders, no conforming, correct timecode, no 3rd party software needed, and better video formats. If you add to this mix a fast disk array, you could actually capture uncompressed HD from your HDV camera or deck. As a side note, Canon's new HD camera is capable of recording uncompressed HD during the shoot via its HD SDI port. Cool But with all uncompressed HD, you'll need some pretty beefy drives to play back the media files. Huge Systems' (now Ciprico) 4gig fibre channel arrays would be where I'd be looking. Speed to spare
My take on HDV is that it can shoot wonderful pictures but it leaves a bit to be desired in post, if you edit it natively. Of course if all you have is time, it's not an issue. Using the AIC codec saves time, but is dicey because it reassigns timecode. However, converting it to DVCPROHD or Uncompressed HD not only preserves the timecode and picture quality, it will speed your render times - and from a pro's point of view, time is money.
Jerry Hofmann would like to thank the Colorado Film School for their help in making this article possible. Check out this super film school out at www.coloradofilmschool.net
A MAC HDV PROBLEM & MATROX'S MXO™ SOLUTION
Find more great Creative COW Magazine articles by signing up for the complimentary Creative COW Magazine.