Post-Production Now that we've looked at some production gear to help us get a better shoot, let’s look at exactly what problems the undersampling in DV, HDV, and HD creates for us which a designated keying program or plugin must overcome.
In its most basic form, any keying program creates a moving garbage matte based on chroma data. If you are using a green screen, then the keyer reads where all the pixels of green are and creates a garbage matte from that information. Obviously, if you don’t have enough chroma information, then your keying program has to guess where the matte is supposed to be based on incomplete data. If it uses only the chroma data, a keying program has a 75% chance of being incorrect when keying DV footage and a 50% chance of being incorrect with HD footage. Another way to put this is that DV footage can have up to a 3 pixel variance in its garbage matte in any direction from the actual footage. Three pixels may not seem like a lot but I have seen green-screened footage where 1/3 of a person’s head or face has been sheared off by this sort of variance, as DV pixels are much larger than HDV or HD pixels! This also means that you’re far more likely to have halo effects around your actors and/or blocky, artifacted edges, because of this much variance. While HD has the best possible chance, even it has only a 50/50 likelihood of success.
While one of these (Keylight) doesn’t attempt to fix issues due to DV, HDV, or HD color decimation, the rest of these programs attempt to compensate for them with everything from combining luma data with chroma data to get a more reliable matte to creating vector mattes that essentially connect the points of the larger DV blocks to originate scalable, smooth edges.
Our test setup for these tests was a PC outfitted with a 1.86 Ghz Core2 Duo processor, 3 Gigs of RAM, and an nVidia Quadro FX 1500 graphics card, which was provided for us by nVidia. While many of these plugins are available for a variety of programs, we used After Effects 7 with GridIron’s acceleration software, Nucleo Pro, for all the plugins. For comparison purposes, once we had keyed footage in the standalone Ultra CS3, we imported the alpha-channeled footage into After Effects 7.
dvMatte Pro 1.5 (8-bit processing) - $199
Specifically designed with DV footage in mind, dvMatte is tailored to compensate for the problems inherent in DV and HDV footage. It includes a simple tutorial to get started, but dvGarage, the creator, also have a large grouping of training footage and tutorials you can pick up as part of their Composite Toolkit. (Read more about it at the end of this article.) This package really has a nice built in ability for blending composited footage with the color space of the background plate. It does this by allowing you to define blacks, mids, and white for both your keyed footage and your background plate. (Obviously, it’s a lot easier if you record a frame of your background plate and a frame of your composited footage with a b/m/w card in the frame. I found that picking up color swatches from your local paint store with matte tiles of white, gray, and black and then gluing them to a piece of posterboard worked famously.)
In addition to the great color blending ability, dvMatte also has two types of light wrap, so that, if your actor or actress is in front of a lighted window, light seems to wrap around their body. Unfortunately, you can’t select partial lightwrap, so you may have to use multiple layers in After Effects with a moving garbage matte to create realistic light wrap for folks walking in front of a window.
Final Breakdown: Platforms: Win/Mac Program Type: After Effects/FCP Plugin Pros: Economical pricing, medium to low learning curve, decent speed, light wrap, nice color blending ability, extended customizability Cons: Composite tends to get a ragged edge at the bottom of the screen if your subject goes off-screen (even if edge repeat is on), refining the edge of the matte is a bit harder than I would like, lightwrap can’t be limited or directed Demo Download: Click Here
zMatte 2.0 (8- to 16-bit processing) - $395
Designed to give you a number of tools to key with, zMatte is a very popular tool with independent filmmakers, as it gives you a lot of power in a single grouping of tools. Starting by allowing you to select a single color (rather than a high and a low like in dvMatte or a grouping like in Ultra or AdvantEdge), it will screen out that color in its various permutations. You can adjust how much variance it will tune out until you get the matte that you want. In addition to the single matte, zMatte is unique in that it will let you create both a primary and secondary matte, which can allow you to create a core and an edge matte in a single location. This can be very useful if you can’t get a good edge with your main detail matte and you can’t get good detail with a clean edge matte. By combining the two, you usually can obtain superior results.
Once you’ve set up your composite, zMatte includes a nice lightwrap feature, but, like dvMatte Pro, you can’t limit where it wraps around in the program, which means you’ll again have to duplicate your footage layer and put a moving garbage matte on the top layer.
Final Breakdown: Platforms: Win/Mac Program Type: After Effects/FCP/Motion/Avid/Photoshop Plugin Pros: Good solid keying options, easy to medium learning curve, will allow you to do both a primary and secondary matte for hard to key footage, works well with DV/HDV footage, lightwrap Cons: Lightwrap can’t be limited or directed, shaves a little more off the edges than I would like normally Demo Downloads:Click Here
