TECH TIPS


Is It All Geek to You?
Deciphering the Alphabet Soup That Is Codecs
by Larry Jordan


Fig. 1: For those who love math, this is the DCT-II video transform compression equation.

Just when you thought it couldn’t get any worse, editors suddenly need a crash course in “geek.” You kept telling yourself, “All I need to do is edit; I don’t need to know all these different video formats.” That’s what a producer once thought when he rented a stage in Woodland Hills to shoot some green screen material. To save money, he rented DV cameras for the shoot. Ultimately, some poor editor had to tell that producer he couldn’t pull a key because it was shot in the wrong format––a $75,000 mistake!

Or what about the editor who is handed video using the latest-and-greatest video format—only to discover that it’s an unsupported format on all of his video editing systems. Or, how about the editor who’s been given a bunch of HDV tapes and told to “prep them for broadcast.” In all three cases, codecs are to blame for the resulting disaster.

What the Heck’s a Codec?
A codec (compressor/decompressor) is a small piece of software that lives inside a video camera (and your computer) and provides the instructions on how to convert the raw data received by your camera into images that can be edited. (If you want a more detailed explanation, you can simply contemplate the DCT Transform II formula illustrated in Figure 1. For this article, our simple definition will do just fine.)

It would be nice to ignore codecs entirely. Nice, but not possible. All video is compressed, and so we are forced to deal with codecs whether we want to or not. Here’s the problem: Every codec is good for some tasks (making small files), and not good for others (providing the highest possible quality). Consequently, just as we used to debate the virtues of various film stocks, cameras and lenses, we now need to factor in the quality of the codec that each device uses.

For instance, codecs that are great at making small files, such as DV and HDV, are not the best choice for chroma-keying or compositing. Files that are great at delivering extremely high-quality data, such as HD-CAM SR, create mind-bendingly huge files.

So, the purpose of this article is to provide a guide to choosing from some of the more popular codecs based upon your particular editorial requirements. It should be noted that at last count, there are well over 100 variations on a wide variety of codecs. Some can easily be edited; others can’t. It’s truly a situation in which what you don’t know can hurt you.

Popular Codecs
Figure 2 illustrates over a dozen popular QuickTime video codecs. If you are working in film, your film will be transferred into one of these formats during the telecine process. I’m ignoring codecs that are used for the web or DVD.


Fig. 2: A list of popular video codecs. Those marked with an asterisk (*) have multiple data rates due to supporting multiple frame rates; this table illustrates the largest. PAL and NTSC values are essentially the same.

As the accompanying table makes depressingly clear, there are a lot of codecs to choose from, so how do you decide?

Bit-depth determines how well an image displays color gradients, such as sunsets, color washes, or composited gradients; 8-bit video will often show banding, or stripes, while 10-bit does not.

Making the Right Choice
Following are some general suggestions on selecting a codec. Keep in mind that fistfights have occurred between otherwise rational people discussing some of these bullet points.
• Most cameras only use a very limited number of codecs; many only use one or two. So, your selection of camera is often dictating what happens to your video when it gets recorded to tape or disk.
• There’s a reasonable correlation between file size and quality. However, images that are not projected tend to have greater latitude than images that are projected.
• You can convert between codecs using cards plugged into your computer. However, while you are converting the format, any material lost when the image is first recorded into the camera-original codec remains lost. In other words, you can change for the codec, but you can’t restore the quality.
Note: High-quality conversion/capture cards are made by AJA (www.aja.com) and Blackmagic Design (www.blackmagic-design.com).
• Experience is teaching us that editing HD is not the same as editing SD. In an SD workflow, you pick the camera for the characteristics you need. In an HD workflow, you decide on your output format, which drives your editing format and that drives the selection of cameras and video capture settings.
• The small file sizes of 8-bit video works well for projects that are principally photographic with limited compositing and saturated colors––such as corporate videos, weddings, events, sports and independent films.
• The improved image quality of 10-bit video works best for projects requiring subtle color reproduction, extensive use of composites or gradients, large-screen projection or output back to film.
• GOP-based video formats such as MPEG-2, HDV and XDCAM often require an extra step on output called “conforming,” which can add time to the process. Conforming times vary based upon the speed of the computer. If you plan to work with one of these codecs, get the fastest computer you can afford.

Apple’s ProRes 422
The new kid on the block is Apple’s ProRes 422 codec for QuickTime that shipped with the recent release of Final Cut Studio 2. While it’s too early to know whether the codec lives up to its advanced billing, there are several very attractive features about it:
• It is 10-bit, meaning that it holds more color detail than 8-bit codecs.
• It is fast, allowing playback on slower computers.
• It does not require any conforming time, making it an ideal format to use for converting images acquired using GOP-based cameras to allow faster processing.
• It reduces the size of uncompressed HD files up to one-tenth of the original size.
• It has a lower data rate (18.1 MB/second), which makes it attractive for reducing the file sizes of high-quality SD media.
• This new codec is fully integrated into QuickTime and Apple’s Final Cut Studio 2.
Initial reports on this new codec are very positive, with Sony, Panasonic, and Red all announcing support.

It’s a Wrap
Codecs are a necessary evil when working with video. They determine image quality, file size and how well a particular format handles compositing. However, just because we have to live with them, doesn’t mean we can’t understand how they work and pick the best one for the job.

Which, when you think of it, is what our friends in production have been doing all along––whether this means selecting film stocks, camera lenses or tape formats. It’s all about how to get the best quality images for the stories we have to tell. Unlike the decisions that production made in the past, selecting the right codec has a direct bearing on the speed and quality of work you do as editors.

Larry Jordan is an Apple-Certified Trainer in Digital Media, and a member of both the Directors Guild of America and the Producers Guild of America. Based in Los Angeles, he has written two books on Final Cut Pro, and is the editor and publisher of Edit Well, the rich-media newsletter on Final Cut Studio. Visit his website at www.larryjordan.biz.

[ return to top ]