RealVideo provides the highest quality compressed video available, whether for streaming or downloading. This chapter explains how RealProducer encodes a source video. It covers video production techniques, providing tips for capturing high-quality input and working with digitized video source files. This information will help you to produce high-quality clips.
A video consists of two parts: the visual track and the soundtrack. In a
RealVideo clip, the soundtrack is encoded with RealAudio codecs, and the
visual track is encoded with a RealVideo codec. RealProducer packages both
tracks in a RealVideo clip that, like a RealAudio clip, uses the file extension
.rm, .rmvb (variable bit-rate clip), or .rv (video file created by RealPlayer). The
following sections explain the basic aspects of creating a RealVideo clip.
The most important factor for creating a good streaming video is to start with good source material. The RealVideo encoding process does not improve the quality of the input, although filters can increase the visual contrast or remove certain imperfections. The more you compress video, the more its quality tends to degrade. So it's important to understand how the encoding choices you make affect the clip's final quality, and to follow good practices that keep the quality as high as possible throughout the video production cycle.
As RealProducer converts a source video into a RealVideo clip, it compresses the video information so that the clip can stream at a certain bandwidth. Think of targeting a certain streaming audience as having a fixed data budget to spend. For low-bandwidth audiences such as users on 56 Kbps modems, you have a smaller data budget (about 34 Kilobits a second) than when you target broadband audiences such as DSL and cable modem users (225 Kilobits or more per second). The following table lists some common target audiences, along with the maximum, recommended streaming speed (that is, your total data budget) for each.
For a given data budget, you have to spend data on the following things, which are in your control:
CBR clips, which are described in "Constant Bit Rate Video", are the safer choice for streaming video to any bandwidth. Explained in "Variable Bit Rate Video", VBR clips produce higher-quality results, but are suited only for video downloads and certain high-bandwidth streaming environments.
Most videos include a soundtrack that requires a fixed amount of bandwidth. If you give too much data to the soundtrack, the video's visual quality suffers. Chapter 4 explains the RealAudio codecs used for RealVideo soundtracks. The section "Soundtrack Bandwidth" provides details about the bandwidth division between a video's soundtrack and visual track.
You want to use the largest width and height for the video that you can. However, choosing too large of a video size takes too much data away from the other quality factors. The section "Video Capture" provides recommended dimensions.
The RealVideo codec choice, and its encoding complexity level, also affect the clip quality. By default, RealProducer produces the best results by using the highest-quality codec set to the highest complexity level. However, you may want to choose different settings to speed encoding or reach older versions of RealPlayer. See "RealVideo Codecs".
You do not set the following quality factors directly. Instead, other choices you make affect their outcomes in the encoded clip. In general, following good production practices and making smart choices with the factors that you can control result in good quality in these areas:
The frame rate affects how smoothly the motion flows in the video. If the frame rate encodes too low, the video will look jerky. For more on this, see "Encoded Frame Rates".
If the visual quality comes out too low, the video will look fuzzy. For more information, see "Visual Clarity".
Because RealVideo uses RealAudio to encode a video's soundtrack, a chunk of the clip's bandwidth first goes toward the audio. The visual track is then squeezed into the bandwidth that's left. For 56 Kbps modems, for example, RealVideo clips stream at 34 Kbps, leaving 22 Kbps of modem bandwidth for overhead. How much bandwidth the visual track gets depends on how the audio is encoded. With an 8 Kbps RealAudio voice codec for the soundtrack, the visual track gets 26 Kbps. With a 16 Kbps music codec, though, the visual track gets just 18 Kbps.
At low bandwidths, how you encode the soundtrack greatly affects how the visual track looks. Music codecs typically consume more bandwidth than do voice codecs. Music's greater frequency range requires more data than does speech, so a music soundtrack consumes more bandwidth than a spoken one. A video with an audio narration might therefore look better than one accompanied by music, as there would be more bandwidth available for the visual track.
At higher bandwidths, the soundtrack consumes proportionally less of the available clip data, so differences in soundtrack encoding affect visual quality less. At speeds above 100 Kbps, you get high-quality sound that uses no more than a quarter of the clip's bandwidth. The following table lists the standard target audiences for constant bit rate RealVideo streams, giving the clip streaming speeds and the RealAudio codecs used for the soundtracks, broken out by audio type.
| Target Audience | Clip Speed | Voice Codec | Music Codec |
|---|---|---|---|
| 28.8 Kbps modem | 20 Kbps | 6.5 Kbps voice | 8 Kbps Music - RealAudio |
| 56 Kbps modem | 34 Kbps | ||
| 64 Kbps single ISDN | 50 Kbps | 8.5 Kbps voice | 11 Kbps Music - RealAudio |
| 128 Kbps dual ISDN | 100 Kbps | 16 Kbps voice | 20 Kbps Music - RealAudio |
| Corporate LAN | 150 Kbps | 32 Kbps voice | 32 Kbps Stereo Music High Response - RealAudio |
| 256 Kbps DSL/cable | 225 Kbps | 44 Kbps Stereo Music High Response - RealAudio | |
| 384 Kbps DSL/cable | 350 Kbps | 64 Kbps voice | 64 Kbps Stereo Music - RealAudio |
| 512 Kbps DSL/cable | 450 Kbps | 96 Kbps Stereo Music - RealAudio | |
| 768 Kbps DSL/cable | 700 Kbps |
One way that RealProducer compresses clips is by reducing the input video's frame rate. Each RealProducer audience has a frame rate target, typically 15 or 30 frames per second (fps). Clips encoded at broadband rates usually meet their frame rate targets. At slower streaming speeds, RealProducer attempts to encode the target rate, but will scale the rate down as necessary depending on other factors such as video dimensions and audience bandwidth. So although you cannot control the frame rate precisely in these instances, using good production practices results in higher frame rates.
The higher the frame rate, the smoother the motion will appear in the video:
Most source videos start out at 15 to 30 fps. During encoding, RealVideo adjusts this frame rate downward as necessary. Thus, your encoded clip will not have just one frame rate, but a mix of frame rates that varies with its content. If you follow good production practices, your clips will typically stream over slow- to medium-speed connections at 7 to 15 fps. At higher speeds, you'll get the 15 to 30 fps maximum set in the audience. Other factors besides bandwidth, though, affect a RealVideo clip's frame rate:
RealVideo codecs include Scalable Video Technology (SVT), which affects playback, but has no effect on encoding. SVT scales down frame rates when clips play on slower computers. High frame rates take a lot of processing power to decompress. Although fast PCs handle high frame rates well, slower PCs may have trouble. With SVT, RealPlayer can lower the frame rate "on the fly" to keep a PC's CPU from sputtering. So although a given scene is encoded at 15 fps, it may play on some RealPlayers at 8 fps if those computers lack the power to decompress 15 frames per second.
In addition to lowering the frame rate, RealVideo compresses clips by throwing out pixel data. A video stores information about each pixel in the frame. RealVideo, on the other hand, stores data for pixel groups. When bandwidth is tight, RealVideo shoehorns pixels with slightly different RGB values into the same group. These pixels then look identical rather than nearly identical. This may result in a loss of detail if compression is too high. The following illustration compares a smooth video with one that has lost detail through too much compression.
Bandwidth is the primary factor that affects a clip's visual clarity. By using good production practices as described in this chapter, you can help keep the video's clarity intact during encoding. Also note the following points:
 |
Tip: If you plan to launch a video in double- or full-screen mode, boost video clarity as much as possible during production and encoding. RealPlayer enlarges the clip by duplicating its pixels, which magnifies any defects. |
RealVideo 10 is the default RealVideo codec used with RealProducer 10, but you can also encode with older RealVideo codecs. RealNetworks recommends using RealVideo 10 unless you need faster encoding performance during broadcasts, or you need to stream video to RealPlayer 8.
 |
For More Information: See also "Encoding Complexity Modes" for information about how a video codec's complexity mode affects video quality. |
The RealVideo 10 codec creates the highest quality compressed video possible. It offers improved visual quality over RealVideo 9 and RealVideo 8, especially with fast-action scenes and on-screen text. Because RealVideo 10 performs more complex analysis of video data than earlier codecs, encoding may take more than twice the time required with RealVideo 9. To help shorten the encoding time and improve the quality of live broadcasts, RealProducer provides the following features:
RealVideo 10 is compatible with RealOne Player and later. Users of older RealPlayers are prompted to update to RealPlayer 10 when they attempt to play RealVideo 10 content. Playback of RealVideo 10 content consumes the same amount of system resources on the viewer's computer as playback of RealVideo 9. Viewers, therefore, will not notice any performance slowdown when playing a RealVideo 10 clip compared to a RealVideo 9 clip.
RealVideo 9 improves on RealVideo 8 with higher compression and improved visual quality. RealOne Player and later can play RealVideo 9 clips. Users who have older versions of RealPlayer are prompted to autoupdate to RealPlayer 10 when the viewer attempts to play a RealVideo 9 clip.
The RealVideo 8 codec is backwards-compatible to RealPlayer 8. The video quality is not as high as with RealVideo 9 and 10, but encoding is faster. Additionally, RealVideo 8 requires fewer resources on the RealPlayer machine to decompress. This makes it suitable for the slower processors of mobile, handheld devices.
When you create a RealVideo clip, you can choose constant bit rate (CBR) or variable bit rate (VBR) encoding. In RealProducer, each audience template is either CBR or VBR. CBR encoding is the more traditional method of encoding streaming video. It maintains a consistent bit rate for the stream, such as a constant 34 Kbps when streaming to 56 Kbps modems. You should generally use CBR video when streaming at bandwidths below 350 Kbps, and anytime you want to use SureStream to encode multiple bandwidths into the same clip or broadcast. At high bandwidths, however, you can use CBR or VBR video.
Using SureStream technology, you can encode a constant bit rate RealAudio or RealVideo clip for multiple bandwidths. For example, you can encode a single RealVideo clip for 56 Kbps modems, 112 Kbps dual ISDN, 256 Kbps DSL, and so on. The clip's playback quality improves with each faster speed. When a viewer clicks a link to a SureStream clip, RealPlayer and Helix Server determine which stream to use based on the available bandwidth, as shown in the following illustration.
Each stream that you add to SureStream clip increases the clip's file size, as well as the encoding time. Streams for higher bandwidths increase the file size more than streams for lower bandwidth because high-bandwidth encodings include more data. As explained in "Broadcast Load Management", the encoding time can become a critical issue during live broadcasts, so you need to choose the audiences you use carefully.
 |
Note: The RealVideo codec you choose encodes all of a clip's SureStream streams. You cannot encode half the streams with the RealVideo 10 codec, for example, and the other half with the RealVideo 8 codec. |
Helix Server and RealPlayer can switch streams to compensate for network conditions. If a fast connection becomes bogged down because of high network traffic, Helix Server switches to a lower-bandwidth stream to prevent the presentation from stalling. When the congestion clears, Helix Server switches back to the higher-bandwidth stream. RealPlayer doesn't need to rebuffer data during this shifting.
|
Note: With the exception of the RealAudio lossless codec, all RealAudio codecs use constant bit rate encoding. Any audio stream other than a lossless stream can therefore be included in a SureStream clip, whether as a soundtrack to a video, or as audio-only voice or music. |
Because SureStream can downshift during network congestion, it's a good idea always to include one or more substreams in a CBR clip. A substream is simply a stream encoded at a lower bandwidth than your target audience speed. RealProducer predefines three substream audiences, two for 28 Kbps modems and one for 56 Kbps modems:
Substreams are particularly important for modem users because modem bandwidth can fluctuate widely. For a 56 Kbps modem audience, the 56 Kbps modem template encodes video at 34 Kbps. If the SureStream clip includes the 26 Kbps substream, the video downshifts to that stream if the user's available bandwidth falls below 34 Kbps. The clip can continue to downshift to slower substreams if needed. This ensures that the presentation can still stream during network congestion.
RealProducer labels certain audience templates as "substreams" because they are not intended to be the clip's primary stream. Any CBR audience can act as a substream for a higher-bandwidth audience, however. You can include the 64k Single ISDN audience as a substream for your 150k LAN users, for example, even if you do not intend to stream the clip over ISDN. To be effective, however, a substream should be no more than 100 Kbps slower than the primary audience.
Variable bit rate (VBR) encoding generally provides superior video quality to
constant bit rate (CBR) encoding. It gives more bandwidth to scenes that are
hard to compress, making the most visible difference in videos that have fast-
moving, high-action scenes. Clips encoded with VBR use the file extension
.rmvb. VBR is not compatible with SureStream technology, however, so you can
encode a VBR clip only for a single bandwidth. VBR encoding is suited for
bandwidths of 350 Kbps or higher.
Unlike a CBR clip, a VBR clip does not maintain a constant streaming rate. Instead, it has a target average bit rate (or quality) and a maximum bit rate. For example, RealProducer's 450 Kbps VBR download audience averages 450 Kilobits of data per second. However, the audience has a maximum of 900 Kbps, meaning that the stream can, if necessary, consume up to 900 Kilobits of data a second. These data spikes typically occur during high-action sequences, resulting in greater video clarity and higher frame rates than CBR encoding.
|
Tip: In RealProducer, audience templates that use VBR encoding include "VBR" in the audience name. If the template does not say "VBR," it is a CBR template. |
|
For More Information: Two-pass encoding is an important component for creating a high-quality VBR file. For more information, refer to "Two-Pass Encoding". |
VBR clips are better suited for downloading than are CBR clips. Because a viewer downloads the entire clip before playing it, the bandwidth spikes inherent in a VBR clip will not cause playback problems. As you encode clips for download, you can consider the VBR encoding speeds as a guide to quality and file size. For example, a 450 Kbps VBR download will generally have lower quality and a smaller file size than a 750 Kbps VBR download.
|
Tip: Quality depends on the source content. For a video that has relatively little action and small dimensions, for example, 750 Kbps VBR encoding may offer little improvement over 450 Kbps VBR encoding. |
Although VBR audience templates are geared for video downloads, you can also use VBR encoding for streaming clips and broadcasting live events. The primary reason to do this is that VBR offers better quality than CBR. For example, depending on the content, a 350 Kbps VBR stream may have roughly the same visual quality as a CBR stream encoded at 450 Kbps.
To benefit from VBR, the streaming network must be able to accommodate bandwidth spikes. Local area networks (LANs) and cable modem users are good candidates for VBR streaming. Because all viewers in these networks share a large pool of bandwidth, a VBR clip's intermittent bandwidth spikes tend not to overload a single viewer's connection bandwidth. Connections in which bandwidth is not shared, such as DSL, can prove problematic. A 450 Kbps VBR clip has a maximum bandwidth of 900 Kbps, for instance. If a DSL connection has a maximum throughput of 500 Kbps, the VBR clip may stall.
When you stream a prerecorded VBR clip on demand, each viewer generally starts playback at a different time. This means that the bandwidth spikes inherent in the clip are spread out over time for all viewers. When you broadcast a VBR stream, however, bandwidth spikes occur for each viewer at the same time. When broadcasting, therefore, your network needs to handle higher cumulative spikes than when you stream prerecorded clips.
|
Note: Streaming a VBR clip requires that you use Helix Server version 9 or higher. |
|
Tip: Multicasting, which is available on some intranets, helps to overcome the bandwidth spikes of standard unicasting by delivering one stream to all viewers, rather than a separate stream to each viewer. For more information, refer to the multicasting chapter of Helix Server Administration Guide. |
For each VBR audience template, three settings affect how the clip or broadcast is encoded: maximum bit rate, average bit rate, and quality. Each VBR clip uses two of these three settings to determine how it encodes its data. To create a VBR clip appropriate for your needs, it's important to understand how these settings interact.
|
For More Information: The section "Creating and Editing Audiences" explains how to change the VBR encoding settings through the graphical application. |
|
Tip:
In an audience file, you can define two additional
variations of VBR encoding, vbrUnconstrainedQuality and
vbrUnconstrainedBitrate. For information about setting VBR
properties through an audience file, refer to "Video Stream
Properties".
|
A VBR clip's average bit rate value reflects the average bandwidth of the clip measured as the total number of Kilobits consumed divided by the timeline in seconds. The actual bandwidth at any point during clip playback may be lower or higher than this value. Through most of the video, however, the second-by- second bandwidth use will be close to this value.
A VBR clip's maximum bit rate caps the bandwidth that the clip can consume. It is typically set by default to twice the average bit rate. A 450 Kbps VBR clip has a maximum bandwidth of 900 Kbps, for example. So during a high-action scene, for example, a video encoded with this audience may have a bandwidth spike of up to 900 Kbps. These spikes are only occasional, however, and the overall clip playback rate stays close to the average bit rate.
Although you can set the maximum bit rate value higher or lower, a higher setting is unlikely to result in significantly greater quality. You can lower the setting if you want to reduce the bandwidth spikes on your network. Keep in mind, however, that lowering this maximum decreases the benefit of VBR encoding. The closer the maximum bit rate value approaches the average bit rate value, the more the clip behaves like a CBR clip.
|
Tip: Keep the maximum bit rate value between 50 percent to 100 percent greater than the average bit rate setting. For a 450 Kbps template, for example, the maximum rate should be 675 to 900 Kbps. |
The quality setting gears the encoding process to achieve a certain level of visual quality. With a quality of 100 percent, for example, RealProducer attempts to reproduce the visual quality of the input nearly perfectly. A lower quality, such as 90 or 80 percent, allows for more visual imperfections, but results in a smaller clip that uses less bandwidth. Keep in mind, however, that the quality setting does not guarantee a faithful reproduction of content. It only attempts to achieve the quality level within the input constraints, such as the video dimensions, target frame rate, content type, and maximum bit rate.
Because encoding for quality can vary the average bit rate greatly, the average bit rate setting is ignored. (That is, the average bit rate and quality settings are mutually exclusive, and you can define only one of the values.) If you set a maximum bit rate of 900 Kbps and a quality of 100 percent, for example, the average bit rate may turn out to be close to 900 Kbps. At a lower quality target, such as 80 percent, the average bit rate will probably be lower. The actual, average bit rate depends greatly on content, though, and will vary for each clip.
Because the average bit rate can vary greatly, quality-based encoding is better suited for downloaded clips. However, you can stream or broadcast a quality- encoded stream as long as your network has the capacity to provide to each user the bandwidth indicated by the maximum bandwidth target. Typically, each stream will use less bandwidth than this, and you can compute the average bandwidth of a clip by dividing the file size in Kilobits by the number of seconds in the timeline. For a live broadcast, though, it is impossible to know the average bandwidth of a quality-encoded stream in advance.
|
Tip: By editing an VBR audience file, you can vary the relationships between bit rate and quality. See "Stream Encoding Types". |
If you intend to shoot a new video rather than use existing video content, this section provides tips for capturing high-quality input. Because video loses image quality if it's highly compressed, always start with the best video source available.
|
For More Information: For pointers on recording audio, see "Audio Capture". |
Consider the video's final frame size before you shoot the first frame. Streaming over 56 Kbps modems requires a small video window, so you need to frame important visual elements well. For recommended clip dimensions, see "Video Encoding Dimensions".
The fewer things that change from frame to frame, the sharper the image will appear in a low-bandwidth video. You can do the following to cut down on unnecessary movement:
Of course, you don't want to eliminate all dynamic elements. When you do include rapid movement, allow enough time for objects to resolve. Because of low frame rates and high compression, objects coming to rest may appear blurry at first. If you have a dialog box popping up on a computer screen, for example, have the box remain stationary for a few seconds so that the image resolves.
Bright lighting at a constant exposure keeps the foreground detail crisp. Use uniformly dark colors for backgrounds, and uniformly light colors (but not whites) for clothing. Complex textures such as paisley and stripes degrade the final image quality with unwanted visual effects.
Video playback devices commonly have at least two common output types— S-video and composite. Use S-video, as it produces better results. Professional-grade devices typically have other, high-quality output modes that can connect to a video capture card.
Always use 24-bit or 32-bit color. Lower color resolution results in poor clips.
Whether you shoot a video yourself or digitize existing material, start with high-quality video media. The following are common video formats, listed in order of descending quality:
The following sections provide recommendations on frame rates and video dimensions when capturing video input into a digitized file, and encoding the video into a streaming or downloadable clip. When you encode directly from a capture source, you do not create an input file first. However, it is still important to choose your encoded output dimensions correctly to produce a high-quality clip or broadcast.
|
For More Information: See "Factors for Creating a Good Streaming Video" for background on the relationship between dimensions, bandwidth, frame rate, and visual clarity. |
If you capture video to a digitized file format, such as AVI or MPEG, you can edit the video using video editing software before encoding it with RealProducer. In this case, digitize the video at 320 pixels wide by 240 pixels high unless you are short on disk space or your video capture card recommends different dimensions.
You may want to capture full-motion video at the full-screen size of 640 by 480 pixels if all of the following are true:
When you capture content to a source file first, digitize the video at 15 frames per second (fps) if you plan to stream the clip at less than 150 Kbps. For these low speeds, 15 fps is the maximum rate that the standard RealProducer audiences encode. Above speeds of 150 Kbps, RealProducer can encode up to 30 fps, so it is better to capture the source input at 30 fps.
|
For More Information: For more information about the frame rate for encoded clips, see "Encoded Frame Rates". |
Because video capture places a large burden on a computer's CPU and hard drive, use the fastest computer you have available. On Windows computers, you can use any video capture card that supports Video for Windows or DirectShow.
Use the following formula to calculate the approximate size in megabytes of a digitized video file:
(pixel width) x (pixel height) x (color bit depth) x (fps) x (duration in seconds) |
Suppose you want to capture a three-minute video at 15 frames per second, with 24-bit color, in a window that is 320 by 240 pixels. As you can see from the following equation, your digitized source file would be approximately 622 MB:
(320) x (240) x (24) x (15) x (180) / 8,000,000 = 622 Megabytes |
If necessary, you can conserve disk space by decreasing the clip dimensions or lowering the frame rate, or both.
Some computer file systems limit a single file to 2 GB (2048 MB) in size. At a 320-by-240 size and 15 fps, this translates to about 9.5 minutes of video. Certain video production programs support the OpenDML (AVI 2.0) standard, which allows the creation of files larger than 2 GB. RealProducer may be able to accept a video source file larger than 2 GB as input, depending on the production software used to create the file.
If you plan to produce long videos or videos with large dimensions, check whether or not your video production software is limited to a 2 GB output file size. If it is not limited to 2 GB, create a video file larger than 2 GB and test to determine if RealProducer can accept the file as input.
|
Tip: If you are limited to 2 GB for the video source file and you need to produce a larger video, you can create separate video source files (each 2 GB or smaller) and encode them as separate RealVideo clips. Then, merge the clips using the RealMedia editor described in Chapter 12. |
When you capture video to a digitized input clip, you want to capture the largest size possible to preserve as much quality as you can. When you encode the file as RealVideo, however, you may need to reduce the video dimensions. Choosing dimensions too large for a given target bandwidth can result in a low frame rate or a large number of visual artifacts, rendering the video jerky or fuzzy.
There are no specific rules for which video dimensions to use, other than to maintain the aspect ratio of the digitized source. The primary consideration for selecting encoding dimensions is bandwidth, though other factors can affect the quality. For example, to keep its frame rate higher, a fast-action clip may require smaller dimensions than a low-action clip.
|
For More Information: To resize a video, you can scale the source file with your video editing software. Or, you can crop or resize the RealVideo clip as you encode it. For more on resizing, see "Resize Filter". |
Most videos encoded for streaming to a desktop media player use a 4:3 aspect ratio to fit the dimensions of standard computer monitors. The following are general recommendations for encoded video dimensions based on your target audience's bandwidth:
|
Tip: RealVideo 10 provides higher quality at high compression rates than older RealVideo codecs. When developing video for low-bandwidth audiences, using RealVideo 10 provides higher quality at larger dimensions. |
Mobile devices such as personal digital assistants and smartphones may have different screen sizes, so it's useful to know the specifications for the devices you are targeting. A common screen resolution of most smartphones is 176 pixels by 144 pixels. This size does not have the 4:3 aspect ratio common to television and desktop video. If you are starting with a larger, 4:3 source such as 320 by 240, you can do two things:
|
For More Information: For more about RealPlayer for mobile devices, visit http://www.realnetworks.com/industries/mobile/index.ht ml. |
If you want to encode a video clip or broadcast for both low-bandwidth and high-bandwidth audiences, you can adopt two different strategies:
Using SureStream technology, you can create a single RealVideo clip that streams at many bandwidths. However, if you create the video at a large size such as 320-by-240, the clip will not stream well to slow connections. Using a smaller size benefits modem users, but does not take full advantage of the greater bandwidth of faster connections.
Creating separate clips allows you to encode a larger clip for high-bandwidth audiences, and a smaller clip for low-bandwidth audiences. You can make the clips available through separate links, or use a SMIL <switch> tag to let RealPlayer choose which version to play.
|
Note: Using a job file, you can encode two separate clips at once, sizing each clip separately. This allows you to create a larger clip for high-bandwidth audiences, and a smaller clip for low-bandwidth audiences in a single encoding pass. For more information, see "Media Profile". |
|
For More Information: For information about using SMIL to select clips, refer to the switching chapter of RealNetworks Production Guide. |
As noted in "Video Source File Size Limit", file size limits imposed by an operating system's file system are generally more of a problem when capturing video than when encoding it. A compressed RealVideo clip is often smaller than the input clip. However, there are some cases in which a a RealVideo clip may exceed the operating system's maximum:
When a clip reaches the operating system's file size limit, RealProducer
automatically creates, or rolls, a new clip. The new clip will have the same name
as the original clip, but with a number added to the file name. For example,
movie.rm rolls to movie1.rm when movie.rm reaches the operating system limit.
The clip movie1.rm rolls to movie2.rm if it grows too large, and so on.
|
For More Information: By using the command-line application or editing a job file, you can set lower limits on file rolling, and roll files by time, such as every 15 minutes. For job file information, refer to "File Destinations". The command-line options are explained in "Output and Destination Options". |
|
Tip: You can combine rolled files into a single sequence using a Ram file or a SMIL file. Transitions may not be seamless, and may contain audio and video gaps, however. For information about creating clip sequences, refer to Introduction to Streaming Media or RealNetworks Production Guide. |
RealProducer includes video filters that can improve the appearance of the encoded clip. However, you should understand when to use each type of filter. Some filters can degrade the video quality when used incorrectly. As well, the use of some filters can significantly increase the encoding time.
|
For More Information: For information about using filters with the command-line application, refer to "Prefilter Options". See "Prefilters" for information about defining filters in the job file. |
A by-product of poor quality in one or more links in the video production chain, video noise (which has nothing to do with the audio quality) can distort the encoded clip. These distortions are similar to the "snow" that often shows up in TV signals received over an antenna. The source of the noise is typically hardware, such as the video tape, capture card, or camera. Using professional- quality equipment and media helps eliminate video noise at the source. If your source video is high quality to start with, you won't need the noise filters.
|
For More Information: The section "Video Noise" explains how to apply this filter using the graphical application. |
If your video input has a small amount of noise, turn on the low-noise filter. Because it has a small impact on processing power and won't degrade a video's appearance, the low noise filter is safe to leave on at all times. It's better practice, though, to use it only when necessary.
If noise greatly distorts the source video, try the high noise filter. Use it only if necessary, though, because it can add 30% or more to the encoding time. The high noise filter can also remove slight details, making highly textured surfaces look more smooth, which may not be desirable in all cases.
RealProducer's resizing filter allows you to crop or resize the video as you encode it. You can select whether to do this as a fast resize or a high-quality resize. These resize options affect the video only when you make it smaller. The minimum size for a resized or cropped video is 32-by-32 pixels. The width and height of a resized or cropped video must be a multiple of 4, such as 160 pixels, 240 pixels, and so on.
RealProducer can resize content using a quick method (fast resize), or through a complex analysis (high-quality resize). A fast resize has a small impact on encoding time, but the resulting image may have some distortion. A high- quality resize results in a superior image, but it may double or triple the encoding time because it carefully analyzes the video source before resizing. Because of its impact on speed, the high-quality resize filter is not recommended for broadcasts.
|
Tip: If you are encoding a letterbox clip for display on computer screens that use a traditional 4:3 aspect ratio, crop out the black bars at the top and bottom of the video image. This helps to improve the video quality. |
|
For More Information: The sections "Cropping" and "Resizing the Video" explains how to crop and resize a video using the graphical application. |
The Inverse-Telecine filter is for cinematic film that was transferred to NTSC video, whether the VHS or BETA version. Film is usually photographed at 24 frames per second (fps), whereas the NTSC standard is 30 fps. The film-to- video conversion (called "telecine") duplicates some frames to bring the film input up to the NTSC frame rate. American theatre-release films transferred to video, for example, undergo the telecine process.
Use the inverse-telecine filter when encoding NTSC video that was transferred from film and has a frame rate of 25 to 30 fps. (The filter is not necessary if the video frame rate is below 25 fps.) The filter strips out redundant frames, letting RealProducer focus on image quality. This improves the clip's overall look. Although the inverse-telecine filter is safe to use on all input, it slows performance marginally and should be used only when the source is NTSC video that originated from film.
|
Note: PAL video, which is widely used in Europe, does not require the inverse-telecine filter because the conversion from 24 fps film to 25 fps PAL does not use the telecine process. |
|
For More Information: The section "De-Interlace and Inverse- Telecine" explains how to apply this filter using the graphical application. |
The de-interlace filter removes jaggedness in interlaced NTSC or PAL video. A video camera running at 30 frames per second captures the odd-numbered lines of a frame in 1/60th of a second, and the even-numbered lines in the next 1/60th of a second. It then interlaces the two to create the frame. Because half the frame's lines are captured a fraction of a second later than the other half, fast-moving objects may appear jagged, the result of the object advancing slightly within 1/60th of a second. The following figure illustrates this jaggedness in a detail of an interlaced video.
The next figure shows the jaggedness removed with the de-interlace filter.
The de-interlace filter has a modest impact on encoding speed, but is useful only for interlaced source video that is 240 lines or higher. Typical source video used for television is 480 lines high. If you digitize the video with a video capture card that captures at 240 lines high or less, the card throws out either the odd or the even lines, de-interlacing the video itself. The de-interlace filter is safe to leave on, though, because RealProducer never applies it to a video less than 240 lines high.
|
For More Information: The section "De-Interlace and Inverse- Telecine" explains how to apply this filter using the graphical application. |
The black-level correction filter improves a video's contrast by making near- black pixels pure black and near-white pixels pure white. This improves the video's appearance if it looks "washed out" because of a lack of contrast. Using this filter improves the RealVideo codec efficiency by increasing the number of pixels that have the same color value. Applying this filter has only a modest impact on encoding speed.
|
For More Information: The section "Black-Level Correction" explains how to apply this filter using the graphical application. |
The RealVideo options allow you to modify how RealProducer encodes RealVideo clips. You can generally change these options separate for each audience template. As with the RealVideo filters, you should understand how each options works before changing the default values. Setting an option incorrectly can degrade the video quality.
With two-pass encoding, which is used only when encoding from a digitized source file, RealProducer runs through the entire source video once to gather information about how best to encode the streaming clip. It then makes a second pass to encode the streams. Two-pass encoding can substantially increase clip quality, but it requires more encoding time. The first pass takes about as long as it would to encode the source file for one target audience.
Although two-pass encoding helps when you use constant bit rate encoding, it provides greater benefit for variable bit rate (VBR) encoding, described in "Variable Bit Rate Video". With two pass encoding, RealProducer can analyze the entire video file to determine how best to vary the playback bit rate through the length of the clip. Without two-pass encoding, RealProducer sequentially analyzes small sections of the source file during encoding, creating a string of VBR sections within the clip.
|
Tip: Use two-pass encoding whenever you encode from a digitized file. Turn it off only if you must decrease the encoding time. When you use live input, RealProducer deactivates two-pass encoding automatically. |
|
For More Information: The section "Using Two-Pass Encoding" explains how to disable two-pass encoding through the graphical application. See "Job Properties" for information about setting this feature directly in the job file. The section "Disable Two-Pass Encoding (-dt)" explains how to override this feature using the command-line application. |
RealProducer uses encoding complexity modes of low, medium, and high that
affect the RealVideo 9 and RealVideo 10 codecs, as well as the RealAudio
lossless codec. The default value of high produces the best possible results, but
also requires the most processing, resulting in the longest encoding times.
Lowering the complexity level to medium or low results in faster encoding
times, but reduced visual quality (for video) or a larger file size (for lossless
audio).
With RealProducer Plus, the complexity mode for video encoding is user- definable for each audience template through the template's advanced video options. You can also set the complexity for video or lossless audio through the command line, and capture the complexity setting to a job file. Note the following about the encoding complexity settings:
low encoding complexity is generally equivalent in quality and encoding time to RealVideo 9 set to the high encoding complexity.high. See "Broadcast Load Management".-eco option that allows you to override the video or lossless audio encoding complexity mode selected in the audience file. See "Encoding Complexity Override (-eco)". |
For More Information: The section "Creating and Editing Audiences" explains how to edit audience templates to change the RealVideo complexity mode. See "Video Stream Properties" for information about setting this feature directly in the audience file. |
Each RealVideo clip has a maximum startup latency that determines how long video requires to display after RealPlayer begins to receive the stream. The default value of 4 ensures that the video requires no more than four seconds of buffering once the stream begins. This four second latency does not include the time it takes to launch RealPlayer, find the host Helix Server, send the request, and receive the server's response, however.
If necessary, you can increase the startup latency in whole values up to 60 seconds. This may be particularly useful for videos that stream at low bit rates and start out with high action sequences. The longer latency creates a larger data buffer for the starting sequence, and generally improves the video's appearance. Bear in mind, however, that a long latency time may cause restless viewers to stop the presentation before it begins playback.
|
Note: The startup latency does not affect how quickly a downloaded clip begins to play. Increasing the latency value, however, can improve the visual quality in downloaded clips that begin with fast-action sequences. |
|
For More Information: The section "Creating and Editing Audiences" explains how to edit audience templates to change this RealVideo option. See "Video Stream Properties" for information about setting this feature directly in the audience file. |
An uncompressed video records all data for each frame of the video. At 15 to 30 frames per second, the amount of data quickly escalates to a very large file size. When creating a RealVideo clip, RealProducer encodes the full frame data for only certain frames, called keyframes. The frames that follow a keyframe encode just the data that describes how that frame varies from the preceding frame. How often keyframes occur depend on the video contents. The first frame of a new scene typically requires a new keyframe. A fast-action video typically requires more keyframes than a slow-moving video.
|
For More Information: The section "Creating and Editing Audiences" explains how to edit audience templates to change this RealVideo option. See "Video Stream Properties" for information about setting this feature directly in the audience file. |
As an option for each audience template, you can set the maximum time between keyframes, which is 10 seconds by default for every audience template. This means that RealProducer adds a keyframe at least every 10 seconds. Most videos will have more frequent keyframes than this anyway, depending on the video contents. Under most circumstances, you should not change this default value. You may want to lower it, though, to provide certain benefits. Adding more frequent keyframes does the following:
Because keyframes encode much more data than other frames, lowering the maximum time between keyframes can either increase the clip file size, or lower the clip's image quality. In short, lowering the maximum keyframe rates provides more resilience against data loss, but degrades the overall compression efficiency. You should therefore change the keyframe rate only after careful consideration and testing to determine if the change produces the desired results.
RealProducer's loss-protection feature adds error-correction data to RealVideo packets, helping them maintain quality when they are streamed in lossy environments. You'll get more benefit from loss protection when streaming across the Internet than over an intranet. The feature is turned off by default, but is safe to turn on for all encoded content because RealProducer adds only as much error-correction data as it can without lessening the video quality.
|
Note: For most types of live broadcasts, RealProducer can also generate error correction packets to protect the stream as it is delivered to Helix Server. For more information, see "Forward Error Correction". |
|
For More Information: The section "Creating and Editing Audiences" explains how to edit audience templates to change this RealVideo option. See "Video Stream Properties" for information about setting this feature directly in the audience file. |
|
©2004 RealNetworks, Inc.
For more information, visit RealNetworks Click here if the Table of Contents frame is not visible at the left side of your screen. |