How to Convert
VHS Tapes to DVD

My recommendation for the best way to do it, based
on many months of screwing around with the problem.

Single Frame from Video Capture of Killing Time (1998)
640x480 Resolution Using TBC and VirtualDub

There are many guides and websites dedicated to the process of converting VHS videotapes to DVD media. Most of them are very detailed and extremely helpful. I could recommend many of them to you. But, instead, I'm going to tell you about what I've discovered after reading a great deal and experimenting for many months. Let me qualify my opinion, though, by stating that I consider it a personal challenge to figure out how to do things the cheapest and simplest way possible. As it so happens, my experience has shown that the cheapest and simplest ways happen to produce the best results, even if they take a little longer to obtain.

Furthermore, I have no special hobbies, agenda, or conditions that make my needs unique or even unusual. I'm not trying to capture Russian cartoons or Japanese anime. I don't have a stack of 3/4" Sony U-Matic tapes from a weekly show produced at Peter Bent Brigham Hospital for the children's wards. I don't maintain hundreds of Beta format tapes simply because I KNOW deep in my heart that the format will soon be coming back into popularity. No...I just happen to have a collection of good movies on original standard VHS video tapes. It is easier and more convenient for me to have them on DVD, for personal viewing only. I want the best quality image and sound I can manage, maybe with title menu and scene selection.

Now, you might not be impressed with the single frame at the top of this page, but it's from a real-time capture of a rental videotape that's over seven years old. The video was saved at a full 640x480 pixels and exceeds the resolution of most standard TV sets and monitors. There are no noticeable compression artifacts or anomalies. Plus, you're looking at a compressed (JPEG) image...the original image is even clearer. In fact, the only problems with this video are (1) some slight reduction in color saturation, (2) the ever-present telecine interlacing, and (3) the huge size of the file, which means you are eventually going to have to encode/compress it to be able to get it written to some portable medium.

Single Frame at 640x480 Pixels
Demonstrating Absence of Line-Doubling

Enlarged 80x60 Pixel Detail
from Image at Left

Mercifully, I'm not going to waste your time telling you all the detailed settings and minutiae to get you through the process. This page is all about selecting your methods and saving yourself from going up blind alleys. So someone told you to capture your video at a vertical resolution of only 264 pixels? Well, guess what? The standard number of horizontal lines is 525 (i.e., it takes 525 horizontal lines to make a single frame). About 486 of these are actually visible. Since these include interlaced odd and even-numbered lines, it might make sense to capture at 262.5 pixels, or 243 pixels, or maybe 486 lines...but don't forget that NTSC pixels are not actually square, so you won't get a true 4:3 aspect ratio. (See this page for a much better explanation.) Just trust me. Capture at a standard 640x480 pixels and you'll be happier down the road.

Still don't believe you can capture a full 640x480 pixel image?? Take a look at the images above. The single frame at the left is the same video frame as before, converted to JPEG format so it isn't 900 KB in size. I then extracted an 80x60 pixel (4:3 aspect ratio) detail (1/6 of the frame dimensions) as a bitmap and enlarged it without extrapolation. Examine it and you'll see that there's no evidence of horizontal line doubling. Each horizontal scan line, as represented by each horizontal row of pixels, is completely independent of the one above and below. You can't do much better than that. (Well, actually, we can, as shown in this example of resizing by interpolation, but there's not much point in it. Each individual frame in the video would be over 3 MB.)

So let's get going!

Let's get this out of the way right up front. If you want to buy an "all-in-one" DVD maker suite, or use the one that came with your video capture card, great! If you want to buy a VCR-DVD recorder combo or DVD burner and dub your tapes directly, have a nice life! Good luck! But don't start crying when the DRM keeps you from converting commercial VHS tapes to digital format, or you find that older tapes turn out crappy. You'll also find that you're very limited in your options to improve the quality of the digital movie file(s). Trust me, I found out the hard way.

First, you need a VCR to play the tapes. You can get something fancy, like a JVC DigiPure series with built-in time base correction, but you're probably just going to use the same old VCR you have sitting around anyway. That's fine. You DO, however, need to locate the component (video, left channel audio, and right channel audio) outputs and have a high-quality patch cable to run these outputs to your computer. I find it convenient to route the RF (modulated to VHF channel 3 or 4) output to my TV to keep track of where I am on the tape. Don't try to use the RF output for actual video conversion! One of our goals for this whole project is that we want the final product (DVD) to look as good, OR BETTER, than the original videocassette!

Second, you need a computer. It is actually not too important to have a powerful CPU or even a lot of RAM, but you DO need a big hard drive. I'm serious. I wouldn't even try this unless you have a minimum of 120 GB of free disk space (not total disk space...FREE disk space). You can expect to have at least three or four files of 40 GB or more at any one time. (Yes, really. Check the image at the left. My original video capture of Killing Time was a single file over 74 GB in size.) A hard drive partition formatted to FAT32 may run a bit faster, but formatting it to NTFS will let you handle the large files better. More CPU power and RAM will reduce the amount of time it takes to process and encode the video after you've captured it, but the actual capture itself will not tax your computer's resources very much.

Third, you need a video capture device. You can buy a Dazzle unit, or other units that use hardware compression, or a nice Canopus unit, but give it some thought first. Hardware compression is nice...fast and efficient, especially for external USB units...but you will be locked into certain compression schemes and codecs. I've found that it is better to capture your video using lossless compression, which can't be done with many hardware compression units. In my opinion, you should just go on eBay and find a Leadtek capture card that uses a newer Conexant chipset. A good example is a Leadtek Winfast TV-2000XP Expert PCI Card, which sells for less than $50 on eBay and at online stores.

Finally, download VirtualDub, Avery Lee's outstanding video capture and editing tool, and install it. It's free and has an astounding array of filters, guides, forums, and other support. Next, download and install the Huffyuv codec. The Huffyuv codec allows us to do lossless and near-lossless compression of video. There are various other codecs and filters we may end up using, but these will get us started. The video these samples came from was resized slightly and cropped to eliminate some residual noise at the top and bottom of the frame. Then I adjusted the video levels and increased color saturation slightly to give it a little better definition. No noise reduction or deinterlacing was used.

If you find that you can't use VirtualDub, either because you don't have compatible drivers for your video capture device or because it won't operate correctly with your operating system, you can use any one of several other video capture applications, including Nero (NeroVision Express video capture mode shown at left.) I personally find that VirtualDub is faster and more reliable, as well as providing superior editing capabilities. If you use another application, make sure you capture with either NO compression or with the least compression possible. See below to understand why this is so important.

This is worth the price of admission alone! We are going to capture video at the highest resolution and with the least compression possible. Why? Because any reduction in resolution and any compression we apply will be irreversible. Once you compress the video, you can never get back to the original quality. Most compression methods select "keyframes" (called I-frames), which are compressed as much as they can be, but the only information that is stored for the few frames that follow (called B- and P-frames) is what the DIFFERENCES are between them and the keyframe. Naturally, you need a new keyframe each time a scene changes and when too much change occurs too rapidly...and this is where all the computing power goes when you compress video. The compressor is looking for scene changes, comparing frames, estimating image changes, and making decisions about the best way to handle the video. (For more information on intra-frame and interframe video compression, check out this website or this one for a great explanation.)

You can see, too, why video compression might create problems and overhead when we want to edit a video file. With film, you have separate images...24 of them per second. Each one is discrete and you can cut between any two single film frames. After you compress video, you can't cut or edit between just any two frames, because a frame may be dependent on the frame or frames that precede it. Frame 247 may depend on frame 238, which is a keyframe. The information for frame 247 may just be a little bit of information on the difference between it and frame 238...not nearly enough to create a single image. This is why video editing software and tools use "frameservers", software applications that reconstruct each film frame so it can be edited or processed as a separate image. In fact, VirtualDub is often used as a frameserver so other video tools can work with compressed video. We're going to try to avoid the whole problem by minimizing our compression.

So far, the expenditure has been minimal and fairly low-tech. But we have two problems looming on the horizon: Macrovision copy protection and NTSC video synch problems. Macrovision copy protection on a videotape causes the color or picture level to change periodically. If your captured video goes from B&W to color and back again every few seconds, it's Macrovision. A TV can use AGC circuitry to compensate for this, but a video capture device usually has problems and you end up with unwatchable video. Several manufacturers, including Sima, make video stabilizers that attempt to control the color/video level to eliminate the effect. I have one and it works fine, but some people say that their combination of tape, VCR, and video capture device won't work with certain video stabilizers. Video synch errors can result from misaligned playback heads and/or worn tapes. The helical scan process that reads (and records) the video signal is pretty complex and delicate. See my page on time base correction for more details on this second type of problem.

Oh, No! What are we going to do? Well, I've already done the hard work for you. You can solve both problems by using a time base corrector (TBC). You don't HAVE to have one, but for less than $100, it will save you loads of heartaches and wasted time. I go into more detail about TBCs here. In a nutshell, a TBC deconstructs the video signal digitally and then rebuilds it, inserting new synch pulses that are more accurate. It also adjusts the signal levels so they produce a much more stable and detailed picture. As a side-effect, this completely removes the Macrovision protection. I find I get the best results if I route the composite video output (the yellow RCA jack output on the VCR) to the video input on the TBC, and then route the S-Video output from the TBC to the video capture card. If you have a VideoLabs Video Scaler Pro and a Leadtek Winfast TV-2000XP Expert capture card, this works perfectly.

The TBC is simplicity itself. You have your choice of three inputs, and you can select which input you want to be active from the front panel. (If you are too confused or disinclined to do even this, the unit will auto-select the active video input and format.) There are three outputs, too. You can convert between any of them EXCEPT that you can't convert the VGA input to other formats. It just passes through to the VGA output.

Here's how I recommend interconnecting your equipment. The TV set is connected solely so you have an independent means of verifying where you are in playing back the VHS tape and for adjusting the tracking, if your unit does not already do this automatically. The video capture will run more smoothly and you can use real-time filtering more easily if you don't use the "preview" features in VirtualDub. ("Preview" is a CPU hog if you don't need it.) Having the TV simplifies knowing when to start the capture and when to end it. It also keeps you from wasting your time if it's not actually playing the tape you want. (If you have a SIMA CopyMaster, or similar video processor, simply substitute it in place of the TBC. If you have neither a TBC nor a video processor, just connect the composite video signal directly to the capture device.)

If you have problems with hum or noise in your audio, check out my page here. You can reduce or eliminate the hum using one or more of my techniques. If absolutely necessary, you can use a ground isolation transformer, which is available from Radio Shack for less than $20. In the method of interconnection shown in the diagram, I'm taking advantage of the internal audio connection that the Leadtek capture card provides, but you could also route the audio directly to the line level input on your own audio card. Make sure, though, that you choose the correct audio input source in Windows and in VirtualDub. Again, test it for a few minutes before you waste a lot of time converting a tape only to find that you have no audio track.

Once you've captured the video as a lossless AVI file, you are going to have to edit it, enhance it, and then compress it in MPEG-2 format so it is compatible with DVD authoring programs. VirtualDub will do all the editing and filtering you need. I then use TMPGenc Plus 2.5 to encode the AVI file to MPEG-2 format as separate video and audio streams (an M2V file and a WAV file). This takes a long time (usually 4 to 8 hours, so don't be in a hurry...you don't want to have to do it a second time.) TMPGenc will apply sufficient compression so the end product will fit on a standard single-layer 4.7 GB DVD-/+R. Hint: You can over-ride the compression and produce files that will not fit on a standard DVD (for example, the files may add up to be 5.4 GB or so). Then, you can use one of several freeware utilities to convert the WAV audio file to AC3, one of the standard DVD audio compression formats. The resulting audio file will be much smaller than the original audio file and you will then be able to allocate more space on the DVD to the video. The result is better-looking video at a lower compression rate. You can also enhance the audio track while it is still a WAV file using an application like GoldWave or Audacity. This allows you to reduce hum, normalize the volume, apply equalization, and so forth. Just remember not to edit or change the size of the file, since it must match the separate video stream exactly, or your audio will be out of synch.

If you don't just happen to have a copy of TMPGenc Plus 2.5, you're still not out of luck. Many CD/DVD burning suites (including Nero, with NeroVision shown at the left) are perfectly suited to create video DVDs from AVI files, provided that you have the correct codecs installed. I won't try to cover all the different steps and/or options for each DVD application...just check your manual or Help pages. Keep in mind that the creation of the files that will be burned to the DVD is an extremely critical step. Up to this point, we have tried to retain the maximum amount of video information by refusing to use a "lossy" compression stage. Now that we have an (essentially) uncompressed AVI file of more than 50 GB, we will need to compress it to less than 4.7 GB (for single-layer DVDs) or 8.5 GB (for dual-layer DVDs). Whichever application you use to compress the file(s), take care to retain the maximum resolution and quality possible while still being able to fit the MPEG-2 files on your recordable DVD.

 

More helpful comments to come!

Yes, tell me more about TBC processing!!

If you want to correspond with me about dubbing VHS tapes to DVD, Id be happy to oblige. Just send mail to me.

Return to the Unterzuber.com web page.