• Blogs
• Forums
• Tradeshows
• Partners
• White Papers
• Events

Working With Legacy Video Formats: Part 1

Jun 12, 2006 12:51 PM, Steve Mullen

In an ideal world, your HD productions will be composed entirely of HD material, or widescreen film that has been transferred to HD. Since the world is not ideal, eventually you will encounter a situation where you will need to incorporate NTSC video. (This includes 4:3 film telecined to SD.)

If you are lucky, the source material will have originated on Digital Beta or Beta SP. These are “component color” formats, so your only task will be incorporating 4:3 material within a 16:9 production. There are two obvious ways to accomplish this. First, place one or more 4:3 images within the widescreen frame. This option has the advantage that video need not be scaled up.

When you do need to upscale NTSC to HD, a linear scale will be required. You should confirm the quality of scaling performed by your NLE. For more information on hardware scaling, see the March 27 issue of this newsletter. Part 2 of this story will describe software that can perform high-quality scaling.

Of course, if you need to place NTSC (or 1080i60) video into a 24p or 60p Timeline, you will need to de-interlace your interlaced material. Perfect de-interlacing is also critical when you freeze an interlaced frame. Here again you need to confirm that your NLE’s de-interlace function is sophisticated enough to perform a high-quality de-interlace. Again, Part 2 of this story describes software that can perform high-quality de-interlacing.

If you are not so lucky, the source material will have been recorded on a “color-under” video format. Three-quarter-inch (U-matic), Video8, Hi8, VHS, and S-VHS use this recording system. Before recording, chroma information is bandwidth-limited. The chroma signal is then Amplitude Modulated by a "carrier" signal and recorded. The luminance signal is Frequency Modulated by a much higher frequency carrier and recorded. Because the chroma signal's carrier is lower in frequency than the luminance signal's carrier, this system is called color-under recording.

The Challenge of Legacy Tapes
The key to obtaining maximum image quality from legacy tapes is to use a VTR that has a correct set of playback features. These features include: 58-micron heads, TBC, DOC, CRI, a Motion Sensing 3D Y/C separator, 3D DNR, and 3D CNR.

To provide special effects, a consumer VCR typically has one 31- and one 26-micron-wide head. An industrial VTR will have two 58-micron wide heads that provide a significantly higher signal-to-noise ratio.

A Time Base Corrector (TBC) captures the picture coming from a VCR and stores it in digital memory — from which it is output. This process removes line-by-line picture "jitter."

Memory also provides the basis for a dropout compensator (DOC). A DOC monitors the RF signal arriving from the active playback head. When the RF signal is interrupted, previous "good" information from memory is substituted for the duration of the interruption.

NTSC Nightmare
Whenever your legacy source has originated from NTSC broadcast, analog cable, or analog satellite, has been recorded on a color-under format — or at any time has been moved as a composite signal – you have a serious problem to solve. The solution is to digitize the video using equipment that has a sophisticated Y/C separator. The separator can be in a VTR (with the output via either analog component or SDI) or in your NLE’s digitizer.

A simple Y/C separator employs a high-pass filter to separate chroma information from the composite signal, while a 3.58MHz notch filter eliminates chroma information contained in the composite signal thereby yielding luma. Unfortunately, the high-pass filter allows fine luminance detail or converging non-parallel lines detail to slip through and appear as, respectively, a rainbow pattern (“cross-color” artifact) or a Moiré (“herringbone” artifacts) pattern. Unfortunately, the notch filter cuts luminance bandwidth to about 3MHz, thus yielding only 250 lines of horizontal resolution.

Comb filters are the heart of more sophisticated Y/C separators. Alternate lines of chroma information have opposing phase characteristics. When two lines are added, chroma information is canceled out while the luminance signal is augmented. By subtracting two lines, the in-phase luminance signal is canceled and chroma information is obtained.

While a comb filter preserves horizontal resolution, it is not effective in preventing either cross-color or Moiré effects. Nor does it prevent horizontal “hanging dots” and vertical “crawling ants” from appearing at the junctions of objects that have complementary (e.g., blue and yellow) colors. These are called “cross-luminance” artifacts.

Artifacts
When cross-color and cross-luminance artifacts are scaled up to HD, the result is a terrible picture. Watching HDTV I see both types of artifacts. HD broadcasts and, for some stations, live news are very clean. However, video that passes through the plant as composite video — microwave uplink, ENG tape, graphics (especially keyed weather maps), and far too many commercials — look horrible. This is especially true of stations that broadcast 720p because unless the station has bought a very expensive transcoder that has an ultra-sophisticated de-interlacer, interlace artifacts (edge combing) from SD sources plague the transmission.

Let this be a reminder that you are in trouble even when a video source comes to you recorded on an analog (Beta SP) or digital (DigiBeta, DV, DVCPRO, or even HD) component format — if it was ever carried via a system that mixes luma and chroma. Once mixed, cross-color artifacts are part of the chroma signal and cross-luminance artifacts are part of the luma signal. In these cases, you should go back to the original media.

In Part 2, we will look more at choosing the correct VTRs, as well as at software that can perform transcoding from SD to HD.

© 2008 Penton Media, Inc.