Videographica: Experiences with the Datavideo TBC-1000
The Datavideo TBC-1000 is an inexpensive timebase corrector (TBC): at $480 it’s a small fraction of the cost of “professional” TBCs. I haven’t used a professional TBC, so I don’t know what I’m missing, but I have been using this little box for a few weeks now, and it’s proven useful.
So What Is a Timebase Corrector, Anyway?
At its most basic, a TBC is a device that continually digitizes frames of video and spits them back out in analog format. This accomplishes two things: it re-generates all of the non-picture portions of the video signal, and, if the video frame timing is off, puts the frames back onto a precise schedule.
TBCs were invented to correct timing drift in videotape systems; tapes stretch, heads don’t spin at exactly the same speed from machine to machine, etc. TBCs have other uses, too. For one thing, since they rebuild the non-picture portions of the video signal, they get rid of troublesome copy control signals like Macrovision. (I don’t advocate video piracy; rather, I sometimes find it necessary to bypass these copy control mechanisms to accomplish a legitimate goal.) If you add external synchronization, TBCs are also useful in analog effects generation, like doing a dissolve from one source to another. The external synchro keeps two TBCs in lockstep, so that when you run video through them, both outputs are exactly synchronized so the dissolve is smooth. Without time base synchronization, one source might be halfway through the vertical retrace while the second source is 3/4 the way through a frame, which would result in a very ugly-looking transition.
In addition to external synchronization, I’m told that professional TBCs add features like color control, and precise control over the re-generated signal. This article answers the question, “If you take all these features out, are you left with a useful device?”
Test 1: Picture Damage
The first test I did was to check to see whether the TBC-1000 changed the picture at all. The basis of the test is to split a video source’s signal into two identical signals (I used a non-matrix type NxM video switch for this), running one directly to a TV’s video input, and the other through the TBC and from there into a second video input on the TV. Here’s the setup I used:
|
To the simplest possible test, I added a PC with a video capture card (the Pinnacle DC1000). I did this for two reasons. First, it was already hooked into the switch and TV, so it wasn’t a matter of hooking something extra up, but just using something that was already in place. Second, a capture card with live outputs acts as a kind of TBC: it is constantly digitizing the signal and then sending a new version of that signal back to the outputs.
The video source was a JVC HR-S9800U S-VHS deck. (I used a camcorder shape in the diagram because Visio doesn’t have a VCR shape. :) ) I put Toy Story in, ran it up to an uncluttered picture and paused the tape. I picked Toy Story because the simple cartoon form with sharp shapes and simple areas of solid colors made picture damage easy to detect. I checked several scenes, to double-check my impressions.
With the three video sources hooked up to the TV, I toggled through them with the TV’s Video button, examining the differences on each change. It quickly became clear that the TBC-1000 didn’t pass the video exactly as it received it. I’m not sure how to describe it exactly. I first heard someone else describe it as brightening the picture, but that isn’t accurate since the dark areas don’t change color. Seeing it myself, I originally called it “cooling,” because very light yellows do cool off, becoming a pale, pale blue; the overall effect is also generally cooler. Again, this is not an overall color shift, it only changes the brightest areas. After some more examination, I wonder if a more accurate description is high-frequency roll-off, as happens in low-quality audio systems.
Whatever the correct technical description is, the effect is not strong or unpleasant: it only becomes apparent in a test like this one. Viewing a video through it, you probably wouldn’t notice that the picture was off unless you were very familiar with the source video. Nevertheless, the TBC-1000 is not “transparent:” you should only run your video through it when you absolutely need timebase correction.
The third video stream — that passing through the DC1000 capture card — proved educational. Originally, its picture matched the TBC-1000’s almost exactly. That’s right, they both cooled the picture an equal amount. However, the DC1000 has color controls. After calibrating the DC1000 with a SMPTE color bar pattern and a little by-hand tweaking beyond that, I was able to make the DC1000’s Toy Story picture come very close to the source picture. The TBC-1000 has no color controls, so you’d either have to put a color tweaking box (like the Sima Color Corrector) inline with it, or live with the cooling effect.
Having calibrated the DC1000 with my DVD player (my SMPTE pattern is on DVD), I decided to run a different cartoon through it: South Park volume 1. This proved a fortuitous choice, as it challenged the DC1000 even more than the TBC-1000. Even after calibrating, tweaking and re-tweaking the DC1000, I came to an inescapable conclusion: the DC1000 can sometimes damage the picture more than the TBC-1000. (This is a professional-level capture card, mind you.) In one scene, I noticed that the fine black wrinkle lines on one character’s face went from easily visible (from the DVD) to slightly filled-in (with the TBC-1000) to significantly-blurred (with the DC1000). In another scene, the average picture level (APL) was rather high: an outdoor snow scene, so that pure whites took up about 40% of the picture area. The TBC-1000 lost some of the bright detail, but the DC1000 lost even more.
I then went back to the VCR playing Toy Story and examined some more frames: The DC1000 was better off, overall, than the TBC-1000. I guess the moral of the story is, that anything you put in the way of the video is going to damage it to a greater or lesser extent, so it’s best to keep the signal path as clear as possible. No doubt a higher-quality TBC or video capture card would have damaged the picture less than what I saw. It’s just a question of how much quality you can afford.
With that in mind, I see no reason to warn someone off of buying the TBC-1000. It certainly isn’t perfect, but I can easily imagine situations where it’s better to have the TBC-1000 in the loop than not. It’s also hard to argue with the price. If you need a basic TBC and can’t afford the pro-level units, it’s a good buy. If you have a second PC on hand with a capture card that passes its input through the capture section and re-generates it “live” on the ouput, this TBC probably doesn’t buy you anything.
I also did a similar test with all different equipment, obtaining very similar results. The only thing I didn’t change is the TBC-1000, corroborating my results above. Below is a list of equipment I used for each version of this test. Test 1a is the test described above, and 1b is the second test.
| Test 1a Equipment | Test 1b Equipment | |
|---|---|---|
| Video Source(s) | JVC HR-S9800U VCR and Sony DVP-S360 DVD | JVC HR-S9800U |
| Video Tape | Toy Story (VHS) and South Park (DVD) | TV program recorded to S-VHS master-quality tape |
| Video Switch | JVC JX-S700 | Sima SVS-4 |
| Television | High-quality consumer Sony 27" (KV-27V42) | Mid-quality consumer Philips 27" |
| Capture card | Pinnacle DC1000 | BMK-Elektronik Kfir-based card |
Test 2: Frame Synchronization
I have an old VHS tape I made as a class project in the late 80s. It was bad video when I made it, and time hasn’t improved it. :) It plays back as well as can be expected on a TV, but when I tried to digitize it with the DC1000, it refused to lock onto the signal.
My JVC HR-S9800U S-VHS deck, which I was using to play back the tape, has timebase correction on the output. It’s meant to stabilize bad videotapes like this one, but it was unable to make the DC1000 happy. I decided to put the TBC-1000 in the loop; lo and behold, the DC1000 began capturing the video without complaint.
This is not a completely happy tale: the TBC-1000 didn’t miraculously restore the video to its original condition. As the frame grabs below show, the TBC-1000 sent through some obviously-flawed pictures. Yet, I wonder if a pro TBC would do any better. The black band isn’t the vertical retrace data, because you can clearly see picture data above the band, and the data below the band isn’t vertically shifted. Is it even possible for a dumb box to detect and remove a video flaw like this?
 Normal frame |
 “Torn” frame |
Further exonerating the TBC-1000, when you watch the video through the TBC, you can’t tell that there’s a box in the middle. Instead of actual blips going through the picture when the VCR is playing the tape directly, you see digitized and re-created blips, indistinguishable from the “real thing.” Even when you digitize the “corrected” video and play that back from the computer, it still looks like the original, sucky VHS tape playing directly from the VCR.
Is the TBC-1000 valuable, then? I say yes. Without it, I couldn’t have digitized this video at all. Alright, so it didn’t perform a miracle and completely de-blip this rotten video. It certainly did better than my S-VHS deck’s TBC, and for all I know a pro TBC wouldn’t do any better. Anyone want to lend me a pro TBC so I can test this? :)