CAPTURE GUIDES -> AVI vs. MPEG
All too often, videographers and hobbyists make this appear to be a harder choice than it really is. Choosing AVI over MPEG, or MPEG over AVI, is a simple choice to make, and should take maybe 10 seconds to decide.
The quick question
Ask yourself this one question: Do you plan to edit the video?
- If yes, then use an uncompressed (YUY2) or low compression (HuffYUV or MJPEG) codec using the AVI file format. Use this AVI to edit in your favorite editing program (like Adobe Premiere) and then encode to MPEG-2 (for DVD) or other desired final format.
- If no, and you merely want to convert the video to VCD or DVD format, then encode directly to MPEG-1 or MPEG-2. You gain no benefits by capturing AVI then encoding to MPEG afterwards. In fact, all it does is take more time.
Editing note! Removing commercials and cutting away unwanted footage is NOT considered editing. That is basic cutting and splicing of footage. You can do this with either AVI or MPEG without reconverting the video or harming quality.
MPEG capture note! Capturing in MPEG format assumes you have a decent MPEG capture card and capture program, such as an ATI All In Wonder card using ATI MMC. Many cards and many software give poor MPEG capture results. Non-ATI users can try other programs, but the options are few and disappointing. This is the main reason I suggest ATI cards. Cards like Matrox, Canopus and Hauppauge can also give great results, using software like MainConcept and maybe PowerVCRII.
What is AVI and when it should be used ?
The history of AVI and technical specifications
AVI is one of the oldest video formats still in common use, having been created by Microsoft more than a decade ago. An AVI is an Audio-Video Interleave based on RIFF headers (resource interchange file format). For years, it was the only viable digital video format available.
An AVI is actually not a video format in and of itself. The "AVI" file is merely a wrapper, what I like to call a "container" file. It holds video and audio information that would typically be unable to exist as its own file format. This video and audio information is written to and read from the AVI by codecs (compressor-decompressor files).
Popular codecs include MPEG-4, DivX, Xvid, MJPEG, HuffYUV, and other FourCC variations. Even MPEG and Windows Media Video (WMV) can exist within an AVI file, as the AVI is just video data with a RIFF header surrounding it.
The AVI format has several distinct drawbacks, with the major one being huge file sizes when using low-compression codecs. (Older variations of AVI present in Windows 3.x and Windows 95/98/NT4 had limitations that prevented file size of larger than 2GB or 4GB, however that has since been removed with the newer operating systems and updates for 98/NT4.)
When AVI should be used and proper settings
An AVI is a temporary working format. An AVI serves no purpose outside of an editing environment on a computer using Windows. (Television formats like MPEG-1 and MPEG-2, and streaming formats like QuickTime, RealMedia and WMV, are all end-product formats.) Apple uses Sorenson-based QT files as their working format.
Again, as stated above, an AVI is a middleman format used to edit. It has little compression applied, and can easily be manipulated a few times before degradation occurs to the video information.
Since most of this site audience is concerned with VHS to DVD conversions, I suggest 352x480 or 720x480 AVI capturing. Both of them are fine. I suggest uncompressed AVI, but those with less hard drive space available may choose to use the HuffYUV or MJPEG codecs. Be sure the HuffYUV is interlaced or that the MJPEG quality is at least 18 or higher.
Myths about AVI quality
An AVI is not any better than an MPEG. It is just different. In fact, using a low bit-rate DivX codec AVI encode will always look worse than a high bit-rate MPEG. Quality depends on how you use the files. MPEG is an end product and is an official video format. An AVI is not even a video format, as it relies on the video data between the RIFF headers.
If editing is not needed, then capturing directly to MPEG is acceptable and quality will be just fine, assuming the capture card and capture program are decent. This is why most people assume MPEG is worse: their hardware and software is not great.
An AVI uses CODECS, which are compression-decompression algorithms.
Be aware that many of the so-called "lossless" codecs are still quite lossy (especially if it is not 4:4:4). When it comes to non-studio home solutions, the copy will never be as perfect as the original. It will be close, but not exact. At this time, in the year 2004, while digital imaging has proceed at a rate faster than anticipated five years ago, it is still unable to surpass the quality, clarity, depth and color palette of good film (although a few studio options do come quite close at achieving some of this).
And again, capturing AVI then converting to MPEG, due to some perceived "quality enhancement" is normally psychological perception, and not an actual quality difference (assuming the hardware and capture software are decent). Whether the video is compressed now or later makes no difference. This situation reminds me of the Bugs Bunny cartoon where Daffy asks Elmer, "Would you like to shoot me now, or wait until you get home?" Seriously, there is no difference. Daffy was somehow fooled into believing one was better than the other, yet the outcome was the same. He had his bill blown off.
What is MPEG and when it should be used ?
The history of MPEG and real-world usage
MPEG is a final product video format. After a video has been converted to MPEG, it should not be edited. Approximately a decade old, yet newer than AVI, the MPEG format was an attempt to provide quality video using smaller file sizes, having been invented by the Moving Pictures Expert Group.
JPEG, MJPEG and MPEG are similar, but please do not confuse them. A JPEG image takes like information, and squeezes it together in the file, compressing the file size. A Motion JPEG (known as MJPEG) does the same, at a frame-by-frame level. Each frame is analyzed for similarities, compressed, then the process is started over at the next frame. MPEG does this, as well as incorporates temporal compression using groups of pictures (the GOP). An MPEG is made up with I-frames, P-frames, and B-frames. The I-frame is the reference mark, and the P-frames and B-frames record only data that is difference from the I-frames. The intricacies of the actual encoding are far more complex than this, but this gives you a basic idea of how an MPEG works.
MPEG also has an option of VBR (Variable Bit-rate) or CBR (Constant Bit-rate). The VBR encode only takes the amount of data needed to make the picture look good, and has both maximum and minimum variables to work with. The CBR gives each frame the same amount of bit-rate and forces the picture to use it all, whether it was needed or not, resulting in larger file sizes.
This being said, the only MPEG file suitable for editing is an MPEG-2 with I-frames only from a CBR encode, as it is essentially an MJPEG since the compression was done within the frame and not temporal.
The MPEG format has seen several variations over the years, having been one of the first streaming formats using SGI WebForce MediaBase (which failed), becoming the MPEG-1 VCD specification, moving to the digital satellite DBS/DVB/DVB-II MPEG-2 formats, becoming the de facto DVD MPEG-2 format, using the popular MPEG-Layer III (MP3) audio format, and most recently as the MPEG-4 DivX video format for video swapping online.
Resolutions vs. D-values
Understand that various resolution are often referred to as D-values. Full D1 video is 720x480. Half D1 video is 352x480. Quarter D1 (actually SIF) video is 352x240. Two-Thirds D1 video is 480x480. DVB-II uses a mutant MPEG resolution, sometimes called Three-Fourths D1 at 544x480.
More is not always better, at least not when the bit-rate surpasses the resolution. Each resolution has an approximate bit-rate maximum, whereas sufficient bits (bandwidth) can be allocated to each pixel. While the MPEG will support bit-rates well into the double-digits, it will likely not use it, and the outcome is simply a bloated file.
352x240/288 = 2.0 MB/s or 2000k max
352x480/576 = 4.0 MB/s or 4000k max
720x480/576 = 8.0 MB/s or 8000k max (same for the 704x480/576 variation)
On a rare occasion an extra meg (or an extra 1000k) can help out. That is normally limited to situations where the video has extremely fast action or contains multitudes of minute details.
The following bell-curves illustrate the quality that can be found with various bit-rates.
How to analyze the graphs:
- Note the quick rise in quality with small jumps in bit-rate.
- At a certain point, you'll reach a plateau of quality.
- The suggested max bit-rates are the earliest point at which this plateau is met in most situations.
- The next number in the middle of the plateau represents a second safety marker, needed only when extreme aspects of the video require larger bit-rates. This can happen with high action or with high detail. WWE Wrestlemania is a perfect example of program that needs higher-than-normal bit-rates, due to fast action and very detailed crowds.
- The two numbers on the curve, found just prior to the suggested bit-rate (2000,4000,8000), are bit-rates that can often provide excellent quality, though some of that is dependent on the encoder software and the hardware being used.
- The number on the middle of the curve provide barely-watchable quality. Note that VCD falls in this range. These are often full of macroblocks.
- The number on the bottom of the bell curve is the last bit-rate that can be safely viewed (though still a miserable viewing experience). Numbers between it and zero will often be so full of artifacts that the image is lost.
- These bit-rates are not VBR/CBR specific. Using VBR may gain slightly better results. Using CBR may yield slightly lower quality results.
MPEG formats for VCD, SVCD, XVCD, CVD and DVD
- An MPEG-1 for VCD (Video CD) is defined as an MPEG-1 file at 352x240 resolution using a constant bit-rate (CBR) of 1150k and MPEG Layer-II (MP2) audio at 44hz. The PAL variation uses a resolution of 352x288 progressive. Please note that most DVD players will accept 1856k as a valid VCD bit-rate, and the higher the bit-rate, the better the quality, especially with MPEG-1 video.
- An MPEG-2 for SVCD (Super Video CD) is defined as an MPEG-2 file at 480x480 resolution using a combined audio-video variable bit-rate (VBR) of 2520k and MPEG Layer-II audio at 44hz. The PAL variation uses 480x576 resolution.
- An MPEG-2 for CVD (Chinese Video Disc) is defined as an MPEG-2 file at 352x480 resolution using a combined audio-video variable bit-rate (VBR) or 2520k and MPEG Layer-II audio at 44hz or 48hz. The PAL variation uses 352x576 resolution.
- An XVCD format can use either MPEG-1 or MPEG-2 and is really not a format at all. The "X" in XVCD does not stand for anything. This is a format used by video hobbyists using improper settings, altering VBR/CBR, bit-rates, audio formats/quality, and resolution. These can only be played by a few DVD players, especially when more than one setting is altered.
- An MPEG-1 for DVD (Digital Versatile Disc) is defined as an MPEG-1 files at 352x240 resolution using a constant bit-rate with a range of 1150k to 1856k and 48hz audio (as PCM or AC3). The PAL variation uses a resolution of 352x288 in either interlaced or progressive format, and may also use MPEG-Layer II audio at 48hz.
- An MPEG-2 for DVD is defined as an MPEG-2 file at a resolution of 352x240 progressive, 352x480 progressive or interlaced, 704x480 (or 720x480) progressive or interlaced, either CBR or VBR, with a bit-rate no greater than 9.8 MB/s (including audio), and using PCM or AC3 48hz audio. The PAL variation uses the 352x288, 352x576 or 720x576 resolutions, either progressive or interlaced, VBR or CBR, and may also use MPEG-Layer II audio at 48hz. DVD MPEG-2 at 720x480 can also use 16:9 aspect ratio for wide-screen video.
MPEG files names
M1V = MPEG1 video
M2V = MPEG2 video
MP2 = MPEG Layer II audio
MP3 = MPEG Layer III audio
MP4 = MPEG-4 video file (should be in an AVI), misnamed
MPA = MPEG Layer II audio
MPV = MPEG1 or MPEG2 video
MPG = .3 extension version of MPEG, from DOS/Win16 era, still used Win32/NT
MPEG = MPEG video, audio or muxed video-audio file
VOB = MPEG1/2 with DVD NAV info
DAT = MPEG1 with VCD data
MP2 is audio, if contains video, misnamed (ATI MMC did this in the old days)
When MPEG should be used
Again, MPEG is a final product video format. When conversion is the only goal, typically transferring VHS to DVD or VCD, then capturing to MPEG, with good equipment, is the best option. Especially since more encoding time is not needed.
PAGE LAST UPDATED | JUNE 4th 2004
Creating DV-AVI and MPEG files from a DV tape
If you wish to store video recorded on a DV tape on a computer hard disk, you can choose between two file formats: DV-AVI and MPEG (MPEG-1 or MPEG-2). DV-AVI files allow you to store video without loss of quality. This means that if you write a DV-AVI file out to a new tape, its quality will be the same as the original video. However, DV-AVI files need large space on disk (about 13 GB per hour of video). MPEG-1 and MPEG-2 allow compressing video; video stored as MPEG-2 provides better quality but requires larger disk space than MPEG-1 (about 4.5 GB vs. 700 MB per hour of video). MPEG-2 is the best format for archiving and distribution, while DV-AVI is very good for editing.
* DV-AVI files should not be confused with �classic� AVI files (the standard Video for Windows), which have the same extension *.avi but contain much less information. In order to play DV-AVI files on a computer, you need special software called codec. This software must be the same used by the DV device to create the video file (see below).
* You can play MPEG-1 files with Windows Media Player without specific codec installed. To play MPEG-2 files, you need either a software DVD player or Windows Media Player provided that a MPEG-2 codec has been installed.