7.10. Making a high quality MPEG-4 ("DivX") rip of a DVD movie

One frequently asked question is "How do I make the highest quality rip for a given size?". Another question is "How do I make the highest quality DVD rip possible? I do not care about file size, I just want the best quality."

The latter question is perhaps at least somewhat wrongly posed. After all, if you do not care about file size, why not simply copy the entire MPEG-2 video stream from the the DVD? Sure, your AVI will end up being 5GB, give or take, but if you want the best quality and do not care about size, this is certainly your best option.

In fact, the reason you want to transcode a DVD into MPEG-4 is specifically because you do care about file size.

It is difficult to offer a cookbook recipe on how to create a very high quality DVD rip. There are several factors to consider, and you should understand these details or else you are likely to end up disappointed with your results. Below we will investigate some of these issues, and then have a look at an example. We assume you are using libavcodec to encode the video, although the theory applies to other codecs as well.

If this seems to be too much for you, you should probably use one of the many fine frontends that are listed in the MEncoder section of our related projects page. That way, you should be able to achieve high quality rips without too much thinking, because most of those tools are designed to take clever decisions for you.

7.10.1. Preparing to encode: Identifying source material and framerate

Before you even think about encoding a movie, you need to take several preliminary steps.

The first and most important step before you encode should be determining what type of content you are dealing with. If your source material comes from DVD or broadcast/cable/satellite TV, it will be stored in one of two formats: NTSC for North America and Japan, PAL for Europe, etc. It is important to realize, however, that this is just the formatting for presentation on a television, and often does not correspond to the original format of the movie. In order to produce a suitable encode, you need to know the original format. Failure to take this into account will result in ugly combing (interlacing) artifacts in your encode. Besides being ugly, the artifacts also harm coding efficiency: You will get worse quality per bitrate.

7.10.1.1. Identifying source framerate

Here is a list of common types of source material, where you are likely to find them, and their properties:

  • Standard Film: Produced for theatrical display at 24fps.

  • PAL video: Recorded with a PAL video camera at 50 fields per second. A field consists of just the odd- or even-numbered lines of a frame. Television was designed to refresh these in alternation as a cheap form of analog compression. The human eye supposedly compensates for this, but once you understand interlacing you will learn to see it on TV too and never enjoy TV again. Two fields do not make a complete frame, because they are captured 1/50 of a second apart in time, and thus they do not line up unless there is no motion.

  • NTSC Video: Recorded with an NTSC video camera at 60000/1001 fields per second, or 60 fields per second in the pre-color era. Otherwise similar to PAL.

  • Animation: Usually drawn at 24fps, but also comes in mixed-framerate varieties.

  • Computer Graphics (CG): Can be any framerate, but some are more common than others; 24 and 30 frames per second are typical for NTSC, and 25fps is typical for PAL.

  • Old Film: Various lower framerates.

7.10.1.2. Identifying source material

Movies consisting of frames are referred to as progressive, while those consisting of independent fields are called either interlaced or video - though this latter term is ambiguous.

To further complicate matters, some movies will be a mix of several of the above.

The most important distinction to make between all of these formats is that some are frame-based, while others are field-based. Whenever a movie is prepared for display on television (including DVD), it is converted to a field-based format. The various methods by which this can be done are collectively referred to as "pulldown", of which the infamous NTSC "3:2 telecine" is one variety. Unless the original material was also field-based (and the same fieldrate), you are getting the movie in a format other than the original.

There are several common types of pulldown:

  • PAL 2:2 pulldown: The nicest of them all. Each frame is shown for the duration of two fields, by extracting the even and odd lines and showing them in alternation. If the original material is 24fps, this process speeds up the movie by 4%.

  • PAL 2:2:2:2:2:2:2:2:2:2:2:3 pulldown: Every 12th frame is shown for the duration of three fields, instead of just two. This avoids the 4% speedup issue, but makes the process much more difficult to reverse. It is usually seen in musical productions where adjusting the speed by 4% would seriously damage the musical score.

  • NTSC 3:2 telecine: Frames are shown alternately for the duration of 3 fields or 2 fields. This gives a fieldrate 2.5 times the original framerate. The result is also slowed down very slightly from 60 fields per second to 60000/1001 fields per second to maintain NTSC fieldrate.

  • NTSC 2:2 pulldown: Used for showing 30fps material on NTSC. Nice, just like 2:2 PAL pulldown.

There are also methods for converting between NTSC and PAL video, but such topics are beyond the scope of this guide. If you encounter such a movie and want to encode it, your best bet is to find a copy in the original format. Conversion between these two formats is highly destructive and cannot be reversed cleanly, so your encode will greatly suffer if it is made from a converted source.

When video is stored on DVD, consecutive pairs of fields are grouped as a frame, even though they are not intended to be shown at the same moment in time. The MPEG-2 standard used on DVD and digital TV provides a way both to encode the original progressive frames and to store the number of fields for which a frame should be shown in the header of that frame. If this method has been used, the movie will often be described as "soft-telecined", since the process only directs the DVD player to apply pulldown to the movie rather than altering the movie itself. This case is highly preferable since it can easily be reversed (actually ignored) by the encoder, and since it preserves maximal quality. However, many DVD and broadcast production studios do not use proper encoding techniques but instead produce movies with "hard telecine", where fields are actually duplicated in the encoded MPEG-2.

The procedures for dealing with these cases will be covered later in this guide. For now, we leave you with some guides to identifying which type of material you are dealing with:

NTSC regions:

  • If MPlayer prints that the framerate has changed to 24000/1001 when watching your movie, and never changes back, it is almost certainly progressive content that has been "soft telecined".

  • If MPlayer shows the framerate switching back and forth between 24000/1001 and 30000/1001, and you see "combing" at times, then there are several possibilities. The 24000/1001 fps segments are almost certainly progressive content, "soft telecined", but the 30000/1001 fps parts could be either hard-telecined 24000/1001 fps content or 60000/1001 fields per second NTSC video. Use the same guidelines as the following two cases to determine which.

  • If MPlayer never shows the framerate changing, and every single frame with motion appears combed, your movie is NTSC video at 60000/1001 fields per second.

  • If MPlayer never shows the framerate changing, and two frames out of every five appear combed, your movie is "hard telecined" 24000/1001fps content.

PAL regions:

  • If you never see any combing, your movie is 2:2 pulldown.

  • If you see combing alternating in and out every half second, then your movie is 2:2:2:2:2:2:2:2:2:2:2:3 pulldown.

  • If you always see combing during motion, then your movie is PAL video at 50 fields per second.

Hint:

MPlayer can slow down movie playback with the -speed option. Try using -speed 0.2 to watch the movie very slowly and identify the pattern, if you cannot see it at full speed.

7.10.2. Constant quantizer vs. multipass

It is possible to encode your movie at a wide range of qualities. With modern video encoders and a bit of pre-codec compression (downscaling and denoising), it is possible to achieve very good quality at 700 MB, for a 90-110 minute widescreen movie. Furthermore, all but the longest movies can be encoded with near-perfect quality at 1400 MB.

There are three approaches to encoding the video: constant bitrate (CBR), constant quantizer, and multipass (ABR, or average bitrate).

Note:

Most codecs which support ABR encode only support two pass encode while some others such as x264 and libavcodec support multipass, which slightly improves quality at each pass, yet this improvement is no longer measurable nor noticeable after the 4th or so pass. Therefore, in this section, two pass and multipass will be used interchangeably.

In each of these modes, libavcodec breaks the video frame into 16x16 pixel macroblocks and then applies a quantizer to each macroblock. The lower the quantizer, the better the quality and higher the bitrate. The method libavcodec uses to determine which quantizer to use for a given macroblock varies and is highly tunable. (This is an extreme over-simplification of the actual process, but the basic concept is useful to understand.)

When you specify a constant bitrate, libavcodec will encode the video, discarding detail as much as necessary and as little as possible in order to remain lower than the given bitrate. If you truly do not care about file size, you could as well use CBR and specify a bitrate of infinity. (In practice, this means a value high enough so that it poses no limit, like 10000Kbit.) With no real restriction on bitrate, the result is that libavcodec will use the lowest possible quantizer for each macroblock (as specified by vqmin, which is 2 by default). As soon as you specify a low enough bitrate that libavcodec is forced to use a higher quantizer, then you are almost certainly ruining the quality of your video. In order to avoid that, you should probably downscale your video, according to the method described later on in this guide. In general, you should avoid CBR altogether if you care about quality.

With constant quantizer, libavcodec uses the same quantizer, as specified by the vqscale option, on every macroblock. If you want the highest quality rip possible, again ignoring bitrate, you can use vqscale=2. This will yield the same bitrate and PSNR (peak signal-to-noise ratio) as CBR with vbitrate=infinity and the default vqmin of 2.

The problem with constant quantizing is that it uses the given quantizer whether the macroblock needs it or not. That is, it might be possible to use a higher quantizer on a macroblock without sacrificing visual quality. Why waste the bits on an unnecessarily low quantizer? Your CPU has as many cycles as there is time, but there is only so many bits on your hard disk.

With a two pass encode, the first pass will rip the movie as though it were CBR, but it will keep a log of properties for each frame. This data is then used during the second pass in order to make intelligent decisions about which quantizers to use. During fast action or low detail scenes, higher quantizers will likely be used, and during slow moving or high detail scenes, lower quantizers will be used.

If you use vqscale=2, then you are wasting bits. If you use vqscale=3, then you are not getting the highest quality rip. Suppose you rip a DVD at vqscale=3, and the result is 1800Kbit. If you do a two pass encode with vbitrate=1800, the resulting video will have higher quality for the same bitrate.

Since you are now convinced that two pass is the way to go, the real question now is what bitrate to use? The answer is that there is no single answer. Ideally you want to choose a bitrate that yields the best balance between quality and file size. This is going to vary depending on the source video.

If size does not matter, a good starting point for a very high quality rip is about 2000Kbit plus or minus 200Kbit. For fast action or high detail source video, or if you just have a very critical eye, you might decide on 2400 or 2600. For some DVDs, you might not notice a difference at 1400Kbit. It is a good idea to experiment with scenes at different bitrates to get a feel.

If you aim at a certain size, you will have to somehow calculate the bitrate. But before that, you need to know how much space you should reserve for the audio track(s), so you should rip those first. You can compute the bitrate with the following equation: bitrate = (target_size_in_Mbytes - sound_size_in_Mbytes) * 1024 * 1024 / length_in_secs * 8 / 1000 For instance, to squeeze a two-hour movie onto a 702MB CD, with 60MB of audio track, the video bitrate will have to be: (702 - 60) * 1024 * 1024 / (120*60) * 8 / 1000 = 740kbps

7.10.3. Constraints for efficient encoding

Due to the nature of MPEG-type compression, there are various constraints you should follow for maximal quality. MPEG splits the video up into 16x16 squares called macroblocks, each composed of 4 8x8 blocks of luma (intensity) information and two half-resolution 8x8 chroma (color) blocks (one for red-cyan axis and the other for the blue-yellow axis). Even if your movie width and height are not multiples of 16, the encoder will use enough 16x16 macroblocks to cover the whole picture area, and the extra space will go to waste. So in the interests of maximizing quality at a fixed filesize, it is a bad idea to use dimensions that are not multiples of 16.

Most DVDs also have some degree of black borders at the edges. Leaving these in place can hurt quality in several ways.

  1. MPEG-type compression is also highly dependent on frequency domain transformations, in particular the Discrete Cosine Transform (DCT), which is similar to the Fourier transform. This sort of encoding is efficient for representing patterns and smooth transitions, but it has a hard time with sharp edges. In order to encode them it must use many more bits, or else an artifact known as ringing will appear.

    The frequency transform (DCT) takes place separately on each macroblock (actually each block), so this problem only applies when the sharp edge is inside a block. If your black borders begin exactly at multiple-of-16 pixel boundaries, this is not a problem. However, the black borders on DVDs rarely come nicely aligned, so in practice you will always need to crop to avoid this penalty.

In addition to frequency domain transforms, MPEG-type compression uses motion vectors to represent the change from one frame to the next. Motion vectors naturally work much less efficiently for new content coming in from the edges of the picture, because it is not present in the previous frame. As long as the picture extends all the way to the edge of the encoded region, motion vectors have no problem with content moving out the edges of the picture. However, in the presence of black borders, there can be trouble:

  1. For each macroblock, MPEG-type compression stores a vector identifying which part of the previous frame should be copied into this macroblock as a base for predicting the next frame. Only the remaining differences need to be encoded. If a macroblock spans the edge of the picture and contains part of the black border, then motion vectors from other parts of the picture will overwrite the black border. This means that lots of bits must be spent either re-blackening the border that was overwritten, or (more likely) a motion vector will not be used at all and all the changes in this macroblock will have to be coded explicitly. Either way, encoding efficiency is greatly reduced.

    Again, this problem only applies if black borders do not line up on multiple-of-16 boundaries.

  2. Finally, suppose we have a macroblock in the interior of the picture, and an object is moving into this block from near the edge of the image. MPEG-type coding cannot say "copy the part that is inside the picture but not the black border." So the black border will get copied inside too, and lots of bits will have to be spent encoding the part of the picture that is supposed to be there.

    If the picture runs all the way to the edge of the encoded area, MPEG has special optimizations to repeatedly copy the pixels at the edge of the picture when a motion vector comes from outside the encoded area. This feature becomes useless when the movie has black borders. Unlike problems 1 and 2, aligning the borders at multiples of 16 does not help here.

  3. Despite the borders being entirely black and never changing, there is at least a minimal amount of overhead involved in having more macroblocks.

For all of these reasons, it is recommended to fully crop black borders. Further, if there is an area of noise/distortion at the edge of the picture, cropping this will improve encoding efficiency as well. Videophile purists who want to preserve the original as close as possible may object to this cropping, but unless you plan to encode at constant quantizer, the quality you gain from cropping will considerably exceed the amount of information lost at the edges.

7.10.4. Cropping and Scaling

Recall from the previous section that the final picture size you encode should be a multiple of 16 (in both width and height). This can be achieved by cropping, scaling, or a combination of both.

When cropping, there are a few guidelines that must be followed to avoid damaging your movie. The normal YUV format, 4:2:0, stores chroma (color) information subsampled, i.e. chroma is only sampled half as often in each direction as luma (intensity) information. Observe this diagram, where L indicates luma sampling points and C chroma.

LLLLLLLL
CCCC
LLLLLLLL
LLLLLLLL
CCCC
LLLLLLLL

As you can see, rows and columns of the image naturally come in pairs. Thus your crop offsets and dimensions must be even numbers. If they are not, the chroma will no longer line up correctly with the luma. In theory, it is possible to crop with odd offsets, but it requires resampling the chroma which is potentially a lossy operation and not supported by the crop filter.

Further, interlaced video is sampled as follows:

Top fieldBottom field
LLLLLLLLšššššššš
CCCCšššššššš
ššššššššLLLLLLLL
LLLLLLLLšššššššš
ššššššššCCCC
ššššššššLLLLLLLL
LLLLLLLLšššššššš
CCCCšššššššš
ššššššššLLLLLLLL
LLLLLLLLšššššššš
ššššššššCCCC
ššššššššLLLLLLLL

As you can see, the pattern does not repeat until after 4 lines. So for interlaced video, your y-offset and height for cropping must be multiples of 4.

Native DVD resolution is 720x480 for NTSC, and 720x576 for PAL, but there is an aspect flag that specifies whether it is full-screen (4:3) or wide-screen (16:9). Many (if not most) widescreen DVDs are not strictly 16:9, and will be either 1.85:1 or 2.35:1 (cinescope). This means that there will be black bands in the video that will need to be cropped out.

MPlayer provides a crop detection filter that will determine the crop rectangle (-vf cropdetect). Run MPlayer with -vf cropdetect and it will print out the crop settings to remove the borders. You should let the movie run long enough that the whole picture area is used, in order to get accurate crop values.

Then, test the values you get with MPlayer, using the command line which was printed by cropdetect, and adjust the rectangle as needed. The rectangle filter can help by allowing you to interactively position the crop rectangle over your movie. Remember to follow the above divisibility guidelines so that you do not misalign the chroma planes.

In certain cases, scaling may be undesirable. Scaling in the vertical direction is difficult with interlaced video, and if you wish to preserve the interlacing, you should usually refrain from scaling. If you will not be scaling but you still want to use multiple-of-16 dimensions, you will have to overcrop. Do not undercrop, since black borders are very bad for encoding!

Because MPEG-4 uses 16x16 macroblocks, you will want to make sure that each dimension of the video you are encoding is a multiple of 16 or else you will be degrading quality, especially at lower bitrates. You can do this by rounding the width and height of the crop rectangle down to the nearest multiple of 16. As stated earlier, when cropping, you will want to increase the Y offset by half the difference of the old and the new height so that the resulting video is taken from the center of the frame. And because of the way DVD video is sampled, make sure the offset is an even number. (In fact, as a rule, never use odd values for any parameter when you are cropping and scaling video.) If you are not comfortable throwing a few extra pixels away, you might prefer instead to scale the video instead. We will look at this in our example below. You can actually let the cropdetect filter do all of the above for you, as it has an optional round parameter that is equal to 16 by default.

Also, be careful about "half black" pixels at the edges. Make sure you crop these out too, or else you will be wasting bits there that are better spent elsewhere.

After all is said and done, you will probably end up with video whose pixels are not quite 1.85:1 or 2.35:1, but rather something close to that. You could calculate the new aspect ratio manually, but MEncoder offers an option for libavcodec called autoaspect that will do this for you. Absolutely do not scale this video up in order to square the pixels unless you like to waste your hard disk space. Scaling should be done on playback, and the player will use the aspect stored in the AVI to determine the correct resolution. Unfortunately, not all players enforce this auto-scaling information, therefore you may still want to rescale.

First, you should compute the encoded aspect ratio: ARc = (Wc x (ARa / PRdvd )) / Hc

where:

  • Wc and Hc are the width and height of the cropped video,

  • ARa is the displayed aspect ratio, which usually is 4/3 or 16/9,

  • PRdvd is the pixel ratio of the DVD which is equal to 1.25=(720/576) for PAL DVDs and 1.5=(720/480) for NTSC DVDs,

Then, you can compute the X and Y resolution, according to a certain Compression Quality (CQ) factor: ResY = INT(SQRT( 1000*Bitrate/25/ARc/CQ )/16) * 16 and ResX = INT( ResY * ARc / 16) * 16

Okay, but what is the CQ? The CQ represents the number of bits per pixel and per frame of the encode. Roughly speaking, the greater the CQ, the less the likelihood to see encoding artifacts. However, if you have a target size for your movie (1 or 2 CDs for instance), there is a limited total number of bits that you can spend; therefore it is necessary to find a good tradeoff between compressibility and quality.

The CQ depends both on the bitrate and the movie resolution. In order to raise the CQ, typically you would downscale the movie given that the bitrate is computed in function of the target size and the length of the movie, which are constant. A CQ below 0.18 usually ends up in a very blocky picture, because there are not enough bits to code the information of each macroblock (MPEG4, like many other codecs, groups pixels by blocks of several pixels to compress the image; if there are not enough bits, the edges of those blocks are visible). It is therefore wise to take a CQ ranging from 0.20 to 0.22 for a 1 CD rip, and 0.26-0.28 for 2 CDs.

Please take note that the CQ is just an indicative figure, as depending on the encoded content, a CQ of 0.18 may look just fine for a Bergman, contrary to a movie such as The Matrix, which contains many high-motion scenes. On the other hand, it is worthless to raise CQ higher than 0.30 as you would be wasting bits without any noticeable quality gain.

7.10.5. Audio

Audio is a much simpler problem to solve: if you care about quality, just leave it as is. Even AC3 5.1 streams are at most 448Kbit/s, and they are worth every bit. You might be tempted to transcode the audio to high quality Vorbis, but just because you do not have an A/V receiver for AC3 pass-through today does not mean you will not have one tomorrow. Future-proof your DVD rips by preserving the AC3 stream. You can keep the AC3 stream either by copying it directly into the video stream during the encoding. You can also extract the AC3 stream in order to mux it into containers such as NUT or Matroska.

mplayer source_file.vob -aid 129 -dumpaudio -dumpfile sound.ac3

will dump into the file sound.ac3 the audio track number 129 from the file source_file.vob (NB: DVD VOB files usually use a different audio numbering, which means that the VOB audio track 129 is the 2nd audio track of the file).

But sometimes you truly have no choice but to further compress the sound so that more bits can be spent on the video. Most people choose to compress audio with either MP3 or Vorbis audio codecs. While the latter is a very space-efficient codec, MP3 is better supported by hardware players, although this trend is changing.

First of all, you will have to convert the DVD sound into a WAV file that the audio codec can use as input. For example:

mplayer source_file.vob -ao pcm:file=destination_sound.wav -vc dummy -aid 1 -vo null

will dump the second audio track from the file source_file.vob into the file destination_sound.wav. You may want to normalize the sound before encoding, as DVD audio tracks are commonly recorded at low volumes. You can use the tool normalize for instance, which is available in most distributions. If you are using Windows, a tool such as BeSweet can do the same job. You will compress in either Vorbis or MP3. For example:

oggenc -q1 destination_sound.wav

will encode destination_sound.wav with the encoding quality 1, which is roughly equivalent to 80Kb/s, and is the minimum quality at which you should encode if you care about quality. Please note that MEncoder currently cannot mux Vorbis audio tracks into the output file because it only supports AVI and MPEG containers as an output, each of which may lead to audio/video playback synchronization problems with some players when the AVI file contain VBR audio streams such as Vorbis. Do not worry, this document will show you how you can do that with third party programs.

7.10.6. Interlacing and Telecine

Almost all movies are shot at 24 fps. Because NTSC is 30000/1001 fps, some processing must be done to this 24 fps video to make it run at the correct NTSC framerate. The process is called 3:2 pulldown, commonly referred to as telecine (because pulldown is often applied during the telecine process), and, naively described, it works by slowing the film down to 24000/1001 fps, and repeating every fourth frame.

No special processing, however, is done to the video for PAL DVDs, which run at 25 fps. (Technically, PAL can be telecined, called 2:2 pulldown, but this does not become an issue in practice.) The 24 fps film is simply played back at 25 fps. The result is that the movie runs slightly faster, but unless you are an alien, you probably will not notice the difference. Most PAL DVDs have pitch-corrected audio, so when they are played back at 25 fps things will sound right, even though the audio track (and hence the whole movie) has a running time that is 4% less than NTSC DVDs.

Because the video in a PAL DVD has not been altered, you needn't worry much about frame rate. The source is 25 fps, and your rip will be 25 fps. However, if you are ripping an NTSC DVD movie, you may need to apply inverse telecine.

For movies shot at 24 fps, the video on the NTSC DVD is either telecined 30000/1001, or else it is progressive 24000/1001 fps and intended to be telecined on-the-fly by a DVD player. On the other hand, TV series are usually only interlaced, not telecined. This is not a hard rule: some TV series are interlaced (such as Buffy the Vampire Slayer) whereas some are a mixture of progressive and interlaced (such as Angel, or 24).

It is highly recommended that you read the section on How to deal with telecine and interlacing in NTSC DVDs to learn how to handle the different possibilities.

However, if you are mostly just ripping movies, likely you are either dealing with 24 fps progressive or telecined video, in which case you can use the pullup filter -vf pullup,softskip.

7.10.7. Encoding interlaced video

If the movie you want to encode is interlaced (NTSC video or PAL video), you will need to choose whether you want to deinterlace or not. While deinterlacing will make your movie usable on progressive scan displays such a computer monitors and projectors, it comes at a cost: The fieldrate of 50 or 60000/1001 fields per second is halved to 25 or 30000/1001 frames per second, and roughly half of the information in your movie will be lost during scenes with significant motion.

Therefore, if you are encoding for high quality archival purposes, it is recommended not to deinterlace. You can always deinterlace the movie at playback time when displaying it on progressive scan devices, and future players will be able to deinterlace to full fieldrate, interpolating 50 or 60000/1001 entire frames per second from the interlaced video.

Special care must be taken when working with interlaced video:

  1. Crop height and y-offset must be multiples of 4.

  2. Any vertical scaling must be performed in interlaced mode.

  3. Postprocessing and denoising filters may not work as expected unless you take special care to operate them a field at a time, and they may damage the video if used incorrectly.

With these things in mind, here is our first example:

  mencoder capture.avi -mc 0 -oac lavc -ovc lavc -lavcopts \
  vcodec=mpeg2video:vbitrate=6000:ilmv:ildct:acodec=mp2:abitrate=224

Note the ilmv and ildct options.

7.10.8. Filtering

In general, you want to do as little filtering as possible to the movie in order to remain close to the original DVD source. Cropping is often necessary (as described above), but do not scale the video. Although scaling down is sometimes preferred to using higher quantizers, we want to avoid both these things: remember that we decided from the start to trade bits for quality.

Also, do not adjust gamma, contrast, brightness, etc. What looks good on your display may not look good on others. These adjustments should be done on playback only.

One thing you might want to do, however, is pass the video through a very light denoise filter, such as -vf hqdn3d=2:1:2. Again, it is a matter of putting those bits to better use: why waste them encoding noise when you can just add that noise back in during playback? Increasing the parameters for hqdn3d will further improve compressibility, but if you increase the values too much, you risk degrading the image visibily. The suggested values above (2:1:2) are quite conservative; you should feel free to experiment with higher values and observe the results for yourself.

7.10.9. Encoding options of libavcodec

Ideally, you would probably want to be able to just tell the encoder to switch into "high quality" mode and move on. That would probably be nice, but unfortunately hard to implement as different encoding options yield different quality results depending on the source material. That is because compression depends on the visual properties of the video in question. For example, anime and live action have very different properties and thus require different options to obtain optimum encoding. The good news is that some options should never be left out, like mbd=2, trell, and v4mv. See below for a detailed description of common encoding options.

Options to adjust:

  • vmax_b_frames: 1 or 2 is good, depending on the movie. Note that libavcodec does not yet support closed GOP (the option cgop does not currently work), so DivX5 will not be able to decode anything encoded with B-frames.

  • vb_strategy=1: helps in high-motion scenes. Requires vmax_b_frames >= 2. On some videos, vmax_b_frames may hurt quality, but vmax_b_frames=2 along with vb_strategy=1 helps.

  • dia: motion search range. Bigger is better and slower. Negative values are a completely different scale. Good values are -1 for a fast encode, or 2-4 for slower.

  • predia: motion search pre-pass. Not as important as dia. Good values are 1 (default) to 4. Requires preme=2 to really be useful.

  • cmp, subcmp, precmp: Comparison function for motion estimation. Experiment with values of 0 (default), 2 (hadamard), 3 (dct), and 6 (rate distortion). 0 is fastest, and sufficient for precmp. For cmp and subcmp, 2 is good for anime, and 3 is good for live action. 6 may or may not be slightly better, but is slow.

  • last_pred: Number of motion predictors to take from the previous frame. 1-3 or so help at little speed cost. Higher values are slow for no extra gain.

  • cbp, mv0: Controls the selection of macroblocks. Small speed cost for small quality gain.

  • qprd: adaptive quantization based on the macroblock's complexity. May help or hurt depending on the video and other options. This can cause artifacts unless you set vqmax to some reasonably small value (6 is good, maybe as low as 4); vqmin=1 should also help.

  • qns: very slow, especially when combined with qprd. This option will make the encoder minimize noise due to compression artifacts instead of making the encoded video strictly match the source. Do not use this unless you have already tweaked everything else as far as it will go and the results still are not good enough.

  • vqcomp: Tweak ratecontrol. What values are good depends on the movie. You can safely leave this alone if you want. Reducing vqcomp puts more bits on low-complexity scenes, increasing it puts them on high-complexity scenes (default: 0.5, range: 0-1. recommended range: 0.5-0.7).

  • vlelim, vcelim: Sets the single coefficient elimination threshold for luminance and chroma planes. These are encoded separately in all MPEG-like algorithms. The idea behind these options is to use some good heuristics to determine when the change in a block is less than the threshold you specify, and in such a case, to just encode the block as "no change". This saves bits and perhaps speeds up encoding. vlelim=-4 and vcelim=9 seem to be good for live movies, but seem not to help with anime; when encoding animation, you should probably leave them unchanged.

  • qpel: Quarter pixel motion estimation. MPEG-4 uses half pixel precision for its motion search by default, therefore this option comes with an overhead as more information will be stored in the encoded file. The compression gain/loss depends on the movie, but it is usually not very effective on anime. qpel always incurs a significant cost in CPU decode time (+20% in practice).

  • psnr: does not affect the actual encoding, but writes a log file giving the type/size/quality of each frame, and prints a summary of PSNR (Peak Signal to Noise Ratio) at the end.

Options not recommended to play with:

  • vme: The default is best.

  • lumi_mask, dark_mask: Psychovisual adaptive quantization. You do not want to play with those options if you care about quality. Reasonable values may be effective in your case, but be warned this is very subjective.

  • scplx_mask: Tries to prevent blocky artifacts, but postprocessing is better.

7.10.10. Example

So, you have just bought your shiny new copy of Harry Potter and the Chamber of Secrets (widescreen edition, of course), and you want to rip this DVD so that you can add it to your Home Theatre PC. This is a region 1 DVD, so it is NTSC. The example below will still apply to PAL, except you will omit -ofps 24000/1001 (because the output framerate is the same as the input framerate), and of course the crop dimensions will be different.

After running mplayer dvd://1, we follow the process detailed in the section How to deal with telecine and interlacing in NTSC DVDs and discover that it is 24000/1001 fps progressive video, which means that we needn't use an inverse telecine filter, such as pullup or filmdint.

Next, we want to determine the appropriate crop rectangle, so we use the cropdetect filter:

mplayer dvd://1 -vf cropdetect

Make sure you seek to a fully filled frame (such as a bright scene), and you will see in MPlayer's console output:

crop area: X: 0..719  Y: 57..419  (-vf crop=720:362:0:58)

We then play the movie back with this filter to test its correctness:

mplayer dvd://1 -vf crop=720:362:0:58

And we see that it looks perfectly fine. Next, we ensure the width and height are a multiple of 16. The width is fine, however the height is not. Since we did not fail 7th grade math, we know that the nearest multiple of 16 lower than 362 is 352.

We could just use crop=720:352:0:58, but it would be nice to take a little off the top and a little off the bottom so that we retain the center. We have shrunk the height by 10 pixels, but we do not want to increase the y-offset by 5-pixels since that is an odd number and will adversely affect quality. Instead, we will increase the y-offset by 4 pixels:

mplayer dvd://1 -vf crop=720:352:0:62

Another reason to shave pixels from both the top and the bottom is that we ensure we have eliminated any half-black pixels if they exist. Note that if your video is telecined, make sure the pullup filter (or whichever inverse telecine filter you decide to use) appears in the filter chain before you crop. If it is interlaced, deinterlace before cropping. (If you choose to preserve the interlaced video, then make sure your vertical crop offset is a multiple of 4.)

If you are really concerned about losing those 10 pixels, you might prefer instead to scale the dimensions down to the nearest multiple of 16. The filter chain would look like:

-vf crop=720:362:0:58,scale=720:352

Scaling the video down like this will mean that some small amount of detail is lost, though it probably will not be perceptible. Scaling up will result in lower quality (unless you increase the bitrate). Cropping discards those pixels altogether. It is a tradeoff that you will want to consider for each circumstance. For example, if the DVD video was made for television, you might want to avoid vertical scaling, since the line sampling corresponds to the way the content was originally recorded.

On inspection, we see that our movie has a fair bit of action and high amounts of detail, so we pick 2400Kbit for our bitrate.

We are now ready to do the two pass encode. Pass one:

mencoder dvd://1 -ofps 24000/1001 -oac copy -vf crop=720:352:0:62,hqdn3d=2:1:2 -ovc lavc \
-lavcopts vcodec=mpeg4:vbitrate=2400:v4mv:mbd=2:trell:cmp=3:subcmp=3:mbcmp=3:autoaspect:vpass=1 \
-o Harry_Potter_2.avi

And pass two is the same, except that we specify vpass=2:

mencoder dvd://1 -ofps 24000/1001 -oac copy -vf crop=720:352:0:62,hqdn3d=2:1:2 -ovc lavc \
-lavcopts vcodec=mpeg4:vbitrate=2400:v4mv:mbd=2:trell:cmp=3:subcmp=3:mbcmp=3:autoaspect:vpass=2 \
-o Harry_Potter_2.avi

The options v4mv:mbd=2:trell will greatly increase the quality at the expense of encoding time. There is little reason to leave these options out when the primary goal is quality. The options cmp=3:subcmp=3:mbcmp=3 select a comparison function that yields higher quality than the defaults. You might try experimenting with this parameter (refer to the man page for the possible values) as different functions can have a large impact on quality depending on the source material. For example, if you find libavcodec produces too much blocky artifacting, you could try selecting the experimental NSSE as comparison function via *cmp=10.

For this movie, the resulting AVI will be 138 minutes long and nearly 3GB. And because you said that file size does not matter, this is a perfectly acceptable size. However, if you had wanted it smaller, you could try a lower bitrate. Increasing bitrates have diminishing returns, so while we might clearly see an improvement from 1800Kbit to 2000Kbit, it might not be so noticeable above 2000Kbit. Feel free to experiment until you are happy.

Because we passed the source video through a denoise filter, you may want to add some of it back during playback. This, along with the spp post-processing filter, drastically improves the perception of quality and helps eliminate blocky artifacts in the video. With MPlayer's autoq option, you can vary the amount of post-processing done by the spp filter depending on available CPU. Also, at this point, you may want to apply gamma and/or color correction to best suit your display. For example:

mplayer Harry_Potter_2.avi -vf spp,noise=9ah:5ah,eq2=1.2 -autoq 3

7.10.11. Muxing

Now that you have encoded your video, you will most likely want to mux it with one or more audio tracks into a movie container, such as AVI, MPEG, Matroska or NUT. MEncoder is currently only able to output audio and video into MPEG and AVI container formats. for example:

mencoder -oac copy -ovc copy  -o output_movie.avi -audiofile input_audio.mp2 input_video.avi

This would merge the video file input_video.avi and the audio file input_audio.mp2 into the AVI file output_movie.avi. This command works with MPEG-1 layer I, II and III (more commonly known as MP3) audio, WAV and a few other audio formats too.

MEncoder features experimental support for libavformat, which is a library from the FFmpeg project that supports muxing and demuxing a variety of containers. For example:

mencoder -oac copy -ovc copy  -o output_movie.asf -audiofile input_audio.mp2 input_video.avi -of lavf -lavfopts format=asf

This will do the same thing as the previous example, except that the output container will be ASF. Please note that this support is highly experimental (but getting better every day), and will only work if you compiled MPlayer with the support for libavformat enabled (which means that a pre-packaged binary version will not work in most cases).

7.10.11.1. Limitations of the AVI container

Although it is the most widely-supported container format after MPEG-1, AVI also has some major drawbacks. Perhaps the most obvious is the overhead. For each chunk of the AVI file, 24 bytes are wasted on headers and index. This translates into a little over 5 MB per hour, or 1-2.5% overhead for a 700 MB movie. This may not seem like much, but it could mean the difference between being able to use 700 kbit/sec video or 714 kbit/sec, and every bit of quality counts.

In addition this gross inefficiency, AVI also has the following major limitations:

  1. Only fixed-fps content can be stored. This is particularly limiting if the original material you want to encode is mixed content, for example a mix of NTSC video and film material. Actually there are hacks that can be used to store mixed-framerate content in AVI, but they increase the (already huge) overhead fivefold or more and so are not practical.

  2. Audio in AVI files must be either constant-bitrate (CBR) or constant-framesize (i.e. all frames decode to the same number of samples). Unfortunately, the most efficient codec, Vorbis, does not meet either of these requirements. Therefore, if you plan to store your movie in AVI, you will have to use a less efficient codec such as MP3 or AC3.

Having said all that, MEncoder does not currently support variable-fps output or Vorbis encoding. Therefore, you may not see these as limitations if MEncoder is the only tool you will be using to produce your encodes. However, it is possible to use MEncoder only for video encoding, and then use external tools to encode audio and mux it into another container format.

7.10.11.2. Muxing into the Matroska container

Matroska is a free, open standard container format, aiming to offer a lot of advanced features, which older containers like AVI cannot handle. For example, Matroska supports variable bitrate audio content (VBR), variable framerates (VFR), chapters, file attachments, error detection code (EDC) and modern A/V Codecs like "Advanced Audio Coding" (AAC), "Vorbis" or "MPEG-4 AVC" (H.264), next to nothing handled by AVI.

The tools required to create Matroska files are collectively called mkvtoolnix, and are available for most Unix platforms as well as Windows. Because Matroska is an open standard you may find other tools that suit you better, but since mkvtoolnix is the most common, and is supported by the Matroska team itself, we will only cover its usage.

Probably the easiest way to get started with Matroska is to use MMG, the graphical frontend shipped with mkvtoolnix, and follow the guide to mkvmerge GUI (mmg)

You may also mux audio and video files using the command line:

mkvmerge -o output.mkv input_video.avi input_audio1.mp3 input_audio2.ac3

This would merge the video file input_video.avi and the two audio files input_audio1.mp3 and input_audio2.ac3 into the Matroska file output.mkv. Matroska, as mentioned earlier, is able to do much more than that, like multiple audio tracks (including fine-tuning of audio/video synchronization), chapters, subtitles, splitting, etc... Please refer to the documentation of those applications for more details.