A Location-Rich Travel Show Maps Out HD + HDV Conform
A location-rich travel show shot in HDCAM and HDV for the Fine Living Network led one production team to create a conform guidebook they now can use on all future projects. The show’s director and executive producer shows you how they got there.
Last March, my team at 24fps Productions (www.24fpsproductions.com) began a ten-month journey to create a travel show in high-def for the Fine Living TV Network called Any Given Latitude. The show uniquely combines three primary, often disparate travel priorities: local culture, adventure and breathtaking accommodations. For each of the series’ episodes this past year, we traveled to a different country, including Iceland, Belize, The Dominican Republic, St. Vincent and the Grenadines, Singapore, Vietnam, Thailand, Puerto Rico, Croatia, the Czech Republic, Austria, Portugal and Curaà§ao. The series premiered in October 2005 to some 30 million U.S. viewers and continues to air on Fine Living on Saturdays at 10:00 a.m. (EST). The final episode in this year’s series, shot in Curaà§ao, premiered on June 3.
The locations took us along the isolated ridges of glacial drifts, down into bustling city streets and restaurants and even under the water along lesser-known coral reefs off the coast of Vietnam. Most of these settings, featuring our host, Joanne Colan, and the people she meets along the way, were shot by our DPs Steve Miller, Jon Fordham and Jeff Fisher, with the Sony HDW-700A HDCAM camera at 60i. Although the Fine Living TV Network broadcasts in standard def, we chose to shoot in HDCAM for two reasons: its superior sharpness- a huge advantage for travel-themed projects- and the flexibility it would give us down the line for future use. We also used the Sony HVR-Z1U HDV camera for underwater sequences, in low-light conditions, and as a second camera for cooking segments.
The production schedule was tight as it was, but the delivery schedule was even tighter. With 13 episodes and very little time in between each one, we needed to develop a bulletproof post-production and conform process with constant and repetitive quality assurance to handle multiple half-hour episodes being edited simultaneously. As we wrapped and posted each episode, we’d tweak and refine this process. By the end of the series, we realized that we’d created the outline for an invaluable "conform bible" that details each step and will be our go-to guide for future projects. Our final strategy not only addressed the technical complexity of editing in full-resolution HD, but also gave us greater flexibility for small changes and reviews throughout the workflow. What follows is a general overview of this workflow and how we dealt with some of the technical issues we encountered along the way.
Pre-Conform: Editing in SD
Long-form editing of 1920 x 1080 high-definition material usually starts with downconverting the source tapes to a standard-def format (with matching time code), like DVCAM, then finishing the edit with an NLE like Apple Final Cut Pro or Avid. It’s far easier to complete the bulk of editing in standard def, then conform the edit back to HD after it’s gone through the review process.
The series was edited off-site and conformed at our studios in Manhattan using a Mac G5 with Apple’s Xserve 5.6 Terabyte RAID, an AJA KONA 2 card and the Sony HDW-F500 HDCAM VTR (which we rented from Liman Video Rental in New York for the final tape mastering).
The starting point for the conform is the near-completed SD episode edit in Final Cut Pro. This edit contains the music, SD-resolution template show graphics (show open, lower-thirds, name keys), photos, map animations and in some cases, a small amount of HDV material. You’ve got to make sure all these elements are incorporated into the edit. You need to create internal and network reviews that are as near complete as possible.
How you organize your sequences and tracks in the SD edit is critical. Any Given Latitude, like many half-hour formatted cable network programs, consists of four segments of varying lengths, so we created four individual sequences in FCP to hold each segment. It’s important to avoid nesting sequences as much as possible; it can make the results of media management (the next step) more complicated. With the exception of the lower-thirds and name keys, nothing in our sequences is nested. We also make sure to consistently assign certain types of media to specific tracks. This helps keep the conform and audio mix-down process efficient.
After we made any last internal or network requested changes to the edit, we finalized the timing of the show, which for our series is 19:20. This triggered the scheduling of the final voice-over recording, with our host, Joanne Colan. Only after we had reached this last stage could we start the actual conform.
The Conform: Step-by-Step
To schedule the conform process for standard half-hour shows can take anywhere between two-to-four days, depending how complex the material is. This is what we did for each episode of Any Given Latitude and how we’ll approach conforms on future episodes and future series in development.
STEP 1: Run media manager in FCP offline
The first thing we do for an episode is launch FCP’s ingenious Media Manager on the SD edit. Selecting the four base sequences, we run Media Manager, using the Create Offline function, and target it to our KONA 2 10-bit 1920 x 1080 29.97 set-up. This essentially creates a new project file, which contains the four new sequences (now at the HD resolution), and a media bin with the trimmed media files, represented as offline clips.
STEP 2: Move the offline edit into HD
Next, we pull our new media-managed project file onto the HD edit system along with episode-specific media files like the music, photos and the scratch-track VO. Before we launch the new project file, we disconnect the hard drive with the SD edit, which lets us easily identify media files we may have missed.
STEP 3: Ignore the offline warning but look for problems now
After we open the new media managed project file (dismissing the warning about media files that have gone offline), we analyze the sequences and media bin to look for any initial issues. Note that the few sequences that were nested in the SD edit (the lower thirds and name keys), end up as "virtual" sequences in the media managed HD file. They appear on the timeline, but not in the bins.
STEP 4: Revert any retimed clips from the offline
Clips in the SD edit that have been adjusted for creative effect (time remapping, reversing, speeding up, etc.) will show up incorrectly in the timeline as one frame in length. Watch out: Sometimes these clips can wreak havoc on the edit and even crash your program. Because of this, we usually revert these clips to normal timing in preparation for media manager, applying the creative time change in the conform.
STEP 5: Redigitize HDCAM clips
Once we complete this initial quality assurance (Q/A), we start to re-digitize the clips from the HDCAM tape masters using a Sony JH-3 HDCAM recorder. Undoubtedly, there are always a few clips that don’t capture correctly; we do these manually. For the offline HDV clips, we simply change the Easy Setup preset configuration in FCP to HDV 1080i60, and capture these clips using the Sony Z1U camera, connected via FireWire.
STEP 6: Re-link the other media elements
After all the offline clips have been recaptured in HD resolution, we typically re-link all the other media (music, photos, scratch track VO, etc.). The easiest way to do this is to quit FCP, relaunch it, and use the re-link function on start up. Our RAID contains HD versions of the show open, lower-thirds base graphics and name keys- these all have the same file name as the template used in the SD edit. This makes reconnecting these elements much easier.
STEP 7: Begin corrective passes and sound lock the conform
After all the media is linked, we start a series of corrective passes. In the first pass, we resize all the HDV footage from the native 1440 x 1080 HDV resolution to our working resolution of 1920 x 1080. To do this, we scale each HDV clip in the timeline by 133 percent, and adjust the aspect ratio (in the Motion tab) to -33.33, which stretches the 1440 x 1080 native HDV source material to properly fill our 1920 x 1080 frame. For some clips- especially those involving scenes shot underwater- we scale them by as much as 144 percent. In the second corrective pass, we replace the scratch track and load in the new final VO clips, adjusting their position as necessary. This step is sometimes done in the SD edit if there are delays in starting the conform.
STEP 8: Render and review
After all the sequences have been rendered, we normally initiate the first full review. It’s important to be consistent with the quality assurance, and have more than one person, besides the editor and producer, review it. What might be a glitch to one person might be an "edit effect" to someone else.
STEP 9: Time for color correction
Once any issues are corrected, we move on to the last corrective pass- color correction. Some producers and directors feel that every scene needs color correction. I don’t share this opinion, especially on projects with this kind of delivery schedule. I trust in my DPs to capture imagery in a way that doesn’t require it to be fixed in post. Still, there are always a few scenes that need to be tweaked. For these clips, we use FCP’s three-way color corrector. After these scenes are color corrected and rendered, we do another full review of the sequences, logging and correcting any new or outstanding issues.
STEO 10: Prep the delierables
The delivery requirements on this project to the network were HDCAM and DigiBeta masters as both texted (i.e., with lower thirds and name keys), and text-less versions. With similar deliverables, this means you’d need to make two versions of the edit, as we did. To do this, you simply duplicate all four sequences, then rename the copies as "text-less." Next, use the clip-enable function to disable the lower thirds and name key nested segments in the text-less sequences. Although these graphical elements are on the same video track, we never disable the track, since FCP uses a top-down compositing model and doing so would force a re-rendering of the entire sequence.
STEP 11: Final review before mastering
When the rendering is complete, we review both masters one more time. If we find an issue, it’s important to make the correction in both versions (assuming it doesn’t relate only to the texted version). Any video change from this point on will automatically require re-rendering. Occasionally, changes made to sub-sequences aren’t reflected in the master sequence. When this happens, we temporarily place a slug over the part of the master sequence that wasn’t updated, then immediately remove it, which forces a render. Be sure, after any changes, to re-check the edit. Things that you could easily correct can cause costly tape rejections by the network.
STEP 12: Master your final episode
Finally, we start the mastering process. Using the Edit To Tape function in FCP, we first blacken about one minute of the HDCAM tape starting at 00:58:30:00, then perform an assemble edit of the texted master sequence, making sure to watch and listen as it’s laid down to tape. Once this is done, we spot-check the tape before mastering the text-less version.
After both the HDCAM masters are made, we downconvert them to letterboxed DigiBeta and send them off to the network. Completed shows can take up to 700 GB on the RAID. We generally keep them intact as long as possible, deleting first the render files, then the video files. We then permanently archive project and related media files to DVD-ROM.
The Final Cut Pro conform for a wall-climbing scene in Portugal (below).
The show included underwater shots off Curacao that were taken with the Sony Z1U.
Both the Z1U and HDCAM were used during production in restaurants and in kitchens.
After the new VO has been inserted and adjusted, the conform is then considered sound locked. At this point, we create four OMF files and four mini "video reference" movies for each sequence, and FTP these to Mike Ryan, our on-location and mix-down sound engineer. As we’re continuing with the conform, Mike mixes down the show using ProTools.
The third corrective pass is to fix all the graphics (lower thirds, name keys, map animation, photos, etc.) and finalize the end-credits. For the lower thirds and name keys, it’s a simple matter of re-sizing the text, and adjusting the motion if necessary. The map animation, which lasts on the screen anywhere from 6 – 8 seconds, usually consists of about 350 separate TARGA frames. Importing all these frames into the conform edit tends to impede performance, especially when launching the project file. To get around this, we do the animation in a separate project file, export it as a full-res movie, then import it as a single clip into the conform. Photos usually require resizing, and minor move adjustments, which are relatively straightforward.
With a mastering template, we construct the two main output sequences. The template includes bars and tone, a slate, "holes" to drop in the four show sequences and 20-second slugs to place in-between. It also has a starting timecode of 00:59:00:00, which abides by the Network’s formatting requirements. Once the sequences are placed, and the holes are closed, we do a quick check to make sure that the total timing is frame accurate. If verified, we unlink the video and audio channels from each segment, and delete all the audio from the sequences.
About this time, we receive via FTP the final mix-down audio files from Mike. Fine Living requires four tracks of audio. Tracks 1 and 2 are full stereo mix, left and right. Track 3 is the narration, host stand-ups, and interviews and Track 4 includes the music and effects. Since we have four sequences, this equals sixteen files. Mike creates these files to be the same length as the sequences, so it’s just a simple matter of placing them into the timeline under each sequence. However, it’s very important to make sure the tracks are correctly set-up in FCP. Since tracks 1 and 2 are stereo mixes, placing them on the track turns them back into mono, as FCP defaults these to a zero pan. To correct this we go into the audio tab of each clip and adjust the pan slider. Track 1 is slid all the way to the left (-1), track 2 is slid to the right (+1), etc. Then, we make sure that the sequence is set up to output four channels, by verifying the audio output tab of the sequence settings. We have four dual-mono tracks, with 0db downmixing. One final thing we do is to make sure each of the four tracks in the timeline are correctly targeted to the four output tracks. We do this by control-clicking on the left part of the track and selecting the appropriate target track from the pop-up menu.
Once all the audio has been placed in and appropriately assigned, it’s time for another review of the program. In this review, we pay particular attention to the audio FCP, making sure the mixed tracks line-up with the video. We also monitor the audio being output to the HDW-500 HDCAM deck, selectively adjusting the output channels and making sure they match.
Next, we need to legalize both master sequences. We learned early on that even though you might not see any problem with the video, there may still be problems with excess luma or chroma beyond normal eyesight. Large edit houses use separate scopes and dedicated hardware "legalizers" to accomplish this. We use the Broadcast Safe filter in FCP, set to "extremely conservative," and apply it to each of the four sequences nested into both master output sequences. This requires every frame of both sequences to render, a process that takes around 6-8 hours at this resolution. We usually let this run overnight.
Throughout the whole post-production process, we developed and refined many systems to log, monitor and track the efforts of our team. The end result was a successful travel series that carried viewers to far away locations via the richness and clarity of HD. We’ve recently launched an edit facility and now offer our HD suite to others who can benefit from this experience.