Pros and Cons of Cutting in HD

Hellboy 2 was a first for me in terms of cutting in HD. All of the other projects I’ve cut or worked on have edited in either NTSC or PAL, and so this was my first run through the process of organizing an editorial and post-production process involving HD. Overall, it worked very well. The editor enjoyed working with much higher-quality footage, the sound department loved mixing to higher-quality reels, and having that extra definition gave us assistant editors more flexibility across the board for how and at what quality we would turn materials over to the people who needed them.

That said, after a year of working in HD, there are some big limitations that got in the way, many of them from Day 1. I will go into them in detail, but I think I can sum them up by saying that the HD workflow was obviously not intended to be used for cutting film, and I hope that we soon move back to cutting with 4:3 material, hopefully in 1k or even 2k.

Aspect Ratios and the HD format

My number one biggest problem with cutting in HD is its inflexibility when it comes to aspect ratios. HD video of course has its 1.78 aspect ratio, which looks something like this:178

If you’re shooting film, there is a limitation inherent to the HD aspect ratio which forces you to make some sacrifices and some workarounds, especially if you’re shooting at 1.85 like we did on Hellboy 2. In the diagram below, I’ve taken my 1.78 box and put the boundaries for both 1.85 and 2.35 material inside. Take note of how little room there is between 1.85 and 1.78.aspect_ratio_compare

Now as everyone in post-production knows, we cut using a variety of extra information that we burn into the image, such as key numbers, tape timecode, audio timecode, source tape number, shoot day, etc. If you’re shooting for 2.35, then you might be okay, but for 1.85 this information simply doesn’t fit in the space between 1.85 and 1.78, and most likely what this will force you to do is to scale down your image in telecine. On HB2, we scaled our telecine down so that our 1.85 borders would be clear of our burn-ins, which gave us HD dailies where our 1.85 boundary (marked in blue) was something like this:aspect_ratio_compare_with_still_resized

This amount of scaling is a small change, but annoying enough to still require that you create a custom matte in Avid for screenings, QT exports, or anything where you’re showing images or parts of the movie to people outside of Editorial. If, for example, your VFX requirements require your camera department to clear 1.66 and you to telecine to that, then you’ll end up with an even smaller 1.85 boundary and a bit more pillar-boxing on the left and right than you see here.

Screening/Previewing with HD Dailies

One of the bigger distractions that assistant editors face is the task of preparing the movie for a preview screening. With HD, the promise was supposedly that by cutting in HD you would relieve yourself of the burden of having to upres, and that to make a publicly viewable copy of the movie you could just play out to tape. There are definitely ways that you can get yourself closer to this goal, but here are a few problems we ran into along the way:

  • DNxHD 36

    • We chose to cut in Avid’s DNxHD 36 format, and it has a lot going for it. It’s light on hard drive requirements, easy for your Avid to play either in Avid or in Quicktime, and can look pretty decent on your client monitor. But it’s the HD equivalent of cutting in 14:1 in that it’s a good offline quality but you wouldn’t want to screen it. It’s just not a projectable format, and I would have been embarrassed to take a tape with 36 material to our public previews. Some of the things we noticed included heavier pixelation in very saturated colors, especially in blues such as in the sky. Considering that our main character was a nice shade of red, when we projected DNxHD36 on a large screen we saw many instances of compression artifacts all along Hellboy’s body and tail. This compression also hinders your colorist substantially.
    • Solution: Screen in DNxHD 115
      • The next level up from DNxHD 36 is DNxHD 115. This is a projectable format, and we were very happy with our results. It does take time, especially the first time around, to up-res copies of your sequences to 115 quality. What we did was create a separate workspace on the Unity specifically to hold 115 material that we were redigitizing. Then, when more instances came along where we had to play out for screenings, all we had to do was take a copy of the current sequences, relink the 36 material to the 115 material we kept on that separate workspace, and then just redigitize a small portion of the clips that had been added or extended beyond their handles since the last time.
        I know that technically you don’t have to create a separate workspace since the Relink tool has a nice “Relink to highest quality” option, but we found it easier to manage the disk space on the Unity by keeping the 115 material separate, and we found it easier to see which clips we still needed to up-res by unmounting everything but the DNX115 workspace before starting the relink process.
      • You can also, of course, cut in DNxHD 115 from the very beginning. I’m a little wary of doing this right now, since we found that the more 115 material made it into the editor’s sequences, the more little glitches we noticed while the editor was cutting. The application’s response time would drop, it would hang momentarily, or occasionally freeze on a frame when it thought it was still playing. With the new version of Avid and the new hardware, it’s possible these problems have gone away. But since I haven’t used the new stuff yet I can’t say one way or the other.
        I can say that storage should not be the thing to hold you back, however. Storage is cheap, and if that’s all you need to make 115 work for you then by all means splurge on the storage.
  • Resizes and Burn-ins
    As noted above, our dailies were scaled down during telecine to allow our key numbers and other information to be present at the top and bottom of the frame without intruding into the 1.85 boundary. If your editor avoids using the resize effect entirely, then you don’t have to worry about where those burn-ins go. On the other hand, as soon as you start moving footage around by scaling it down or translating it vertically, your key numbers will intrude into your frame, and you will need to get rid of them before you go to preview.

    • Solution: Re-telecine, scan, or temporarily redo the resize
      • There just isn’t a great, easy solution to this problem. If you choose to re-telecine your material blank, then you’re taking the risk that your framing will be different from what it was when telecined the first time around. In our case, we did our telecine in Hungary to mixed results and inconsistent framing, but after shooting wrapped we moved to London, so we would’ve had to take the risk of moving all that negative and setting up a new telecine session in London without any kind of guarantee that they’d be able to replicate the framing or color of the first telecine in Hungary.
      • If you choose to have your DI facility scan the shots you need, then you have a texture problem in that the scans will look much better than the material you digitized from tape, and you have a color problem because the shots are ungraded and still won’t match the material around them in the cut. You can futz with it by sharpening your scans or applying neutral grades until it looks a bit more acceptable, but it will never truly match. The fact is that a tape to tape grade with a source tape that has a mixture of dailies, DI scans, and vfx shots from who knows how many vendors is a near-impossible task for your preview’s colorist. Our colorist was excellent, but there just isn’t enough flexibility to even everything out from an HDCAM tape that has such a wide variety of compressed sources.
      • The last and easiest option, although one that may make your editor cringe, is to just remove or at least reduce the resize so that they key numbers stay out of frame. We did this more on our 2nd preview since we were running out of time and our colorist had his hands full just with all the different color spaces of our vfx shots vs our dailies, and as long as you don’t cut out a resize effect that is important to the narrative of the film, this is definitely the fastest, cheapest, easiest way to solve the problem.

Rendering and Turnovers

Those were the biggest unforeseen problems with working in HD. One that we knew of in advance, and thankfully one that never quite crippled us, was the issue of increased rendering time for HD material. But this is a problem that is in a constant state of flux. As new processors emerge, render time decreases. As new formats with higher quality specifications are introduced, render time increases all over again. To give you an idea of how we handled our turnovers in order to minimize the amount of rendering/exporting we had to do in the Avid (since transcoding/downconverting/etc. outside of the Avid application is generally faster), see the article on Turnover Workflows.

2 Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  1. Hi, Great Site.

    Having never had the opportunity to work on a feature I’ve never encountered the need to track key info on screen. Why is it important for this information to be tracked in this way (ie burnt in). Does the Avid Timecode burn in Effect change this now that this info can be easily turned on and off when needed.

    Thanks

    Gavin

  2. Gavin-

    On a feature there are any number of people in different departments who access and manipulate the dailies, and since not all of them are on an Avid I would doubt that burnt-in tracking information will go away. It is frequently the case that a studio marketing department will cut a promo or trailer using dailies DVDs or an early DVD copy of the cut, and since the DVDs those departments use to cut have no embedded metadata, the only thing that we in Editorial do have that allows us to recreate the promo using the footage in the Avid is the burned-in TC and key code.

    Even in Avid, I use that tracking information daily, and although I can get the same information through Avid’s Timecode window or the new Timecode effect, neither is as fast or convenient as having it burned into the image. Additionally, the burn-in is usually the most accurate source of information available to tell you the keycode of the current frame. If there is a conflict between the flex file or bin I receive from telecine and the onscreen burn-in, I will usually assume the burn-in is correct until I can prove otherwise.

    For me, the new timecode effect in Avid will be most useful when I turn over to sound, music, and vfx. In this case, I usually need to cover up the burned-in TC and keycode and replace it with the timecode of whatever reel of the movie I’m turning over. Whereas before we would superimpose and line up a counter we digitized from a tape, we can now just put the plugin on our top video layer and set the render or playout off. I’m sure once I dig into that effect I’ll find other uses for it as well, but the most immediate one I can think of is for burning in a particular reel’s TC before handing it off to another department.

    –Evan