Curriculum Corporation
Video Research Project

Illustrating the Zoom

Clip 7a (280KB)

Observe the loss in image detail in this scene of trams as the operator zooms

Clip 7b (78KB)

Observe the loss image detail in this river scene during a zoom

Illustrating the Dissolve

Clip 8 (260KB)

Observe how the image breaks up when a dissolve transition is used during editing

Clip 9 (265KB)

Compare this to the shorter period of degradation when a clean cut is used

Highlighting the Effects of Noise

Clip 10 (412K)

Compressed from original {glossary:minidv}Mini-DV{~glossary} source this clip compresses well with good image detail on the work and faces in the background

Clip 11 (409K)

This 2nd generation {glossary:vhs}VHS{~glossary} dub does not compress as cleanly due to the amount of noise in the source. The level of detail is considerably reduced, particularly in faces.

Shooting and Editing Video for the Web

Useful background whitepapers

Key concepts

  • Applying practical video shooting and editing techniques that are derived from an appreciation of basic video Compression theory can visibly improve the quality of video shot for the web.

Introduction

When shooting video destined for the web or for a {glossary:cdrom}CD-ROM{~glossary} project, are there things to keep in mind to ensure the best possible compressed result? Absolutely!

Keeping the final output in mind during editing can also enhance your web video clips.

Guidelines to maximise "Compressability"

These guidelines for shooting video for the web derive from an understanding of how video {glossary:compression}Compression{~glossary} codecs squeeze uncompressed video into clips compact enough for delivery over the web or on CD as {glossary:quicktime}QuickTime{~glossary}, {glossary:windowsmedia}Windows Media{~glossary}, {glossary:realvideo}Real{~glossary} or {glossary:mpeg4}MPEG4{~glossary} formatted files (as discussed in the Distribution of Video whitepaper).

(Note that because the {glossary:dvd}DVD{~glossary} format provides such a high {glossary:datarate}Data Rate{~glossary}, it is not necessary to adopt special shooting and editing techniques.)

Video can be compressed by eliminating information that is repeated both within a single frame and from one frame to the next. (See the Image {glossary:compression}Compression{~glossary} and Video {glossary:compression}Compression{~glossary} whitepapers respectively.)

Therefore the more repetitive the video sequence is from one frame to another, the easier it will be to compress. Expressed another way, the more one frame has in common with its predecessor, the less new information needs to be sent down the line to describe it.

From this basic understanding of {glossary:compression}Compression{~glossary}, we can derive four guidelines for maximising the ‘compressibility’ of your footage:

  1. Place the video camera on a tripod. A large {glossary:camcorder}Camcorder{~glossary} that sits on the shoulder is quite stable but holding a small video camera steady by hand is difficult. The results look unprofessional and make for difficult-to-compress video because the entire video frame is constantly in motion.

    If objects in the scene such as buildings are stationary, they should also be stationary in your footage. A tripod stabilises the camera resulting in a more professional looking video and enables any intentional movement of the camera to be performed smoothly.

  2. Shoot your subject against a static background. Data must be used to describe the motion of objects in your video; therefore, the data should describe the movement of your subject, rather than superfluous background detail such as leaves blowing about or traffic passing by.

  3. Avoid pans and tilts. Panning is horizontal movement of the camera and is used to follow action or reveal more of a scene. In Figure 1 the camera is being panned left to follow the van as it negotiates the corner. Panning the camera creates motion within the video frame that can starve the crucial foreground subject of data.

    Figure 1: Panning left
    Figure 1 Panning left

    Ideally, you should plan your web video production so that the story can be told using locked-off shots with fixed framing. If panning is necessary in order to follow the action, it should be done with a high-quality {glossary:fluidhead}Fluid Head{~glossary} tripod to facilitate smooth and consistent panning.
    While panning involves horizontal movement of the camera, vertical movement is known as tilting and is often used to keep a seated subject in the video frame as they stand. Again, if vertical camera movement is necessary, use a {glossary:fluidhead}Fluid Head{~glossary} tripod.

  4. Avoid the use of the zoom lens. Excessive use of the zoom lens on a {glossary:camcorder}Camcorder{~glossary} looks unprofessional and results in video that is particularly difficult to compress. Panning, tilting and even the camera shake that is associated with handheld operation merely result in the movement of pixels horizontally and vertically from frame-to-frame. Most codecs can handle this reasonably well because the pixels themselves are unaltered as they move. However, zooming causes every {glossary:pixel}Pixel{~glossary} of the image to change as objects in the frame alter their size.

    In compressed clips that contain zooms, a noticeable pixelisation of the image can occur, as the {glossary:codec}Codec{~glossary} is unable to keep up with this rapid change. As soon as the zoom ends the image recovers. Instead of zooming, try shooting a wide {glossary:establishingshot}Establishing Shot{~glossary} followed by a {glossary:closeup}Close-Up{~glossary} of the subject in question, or cut the zoom out later during editing.

Editing suggestions

Ideally, your video material will have been shot according to the guidelines outlined above. However, most of the time you will be reusing material shot without the {glossary:bandwidth}Bandwidth{~glossary} constraints of the web in mind. Even here though, decisions made when editing can affect the final quality of the compressed output.

When editing two shots together, a hard cut is best in terms of {glossary:compression}Compression{~glossary}. Hard cutting joins the final frame of the first clip directly to the first frame of the next shot (see the top filmstrip in Figure 2) without any complex transitions.

The worst transition from a {glossary:compression}Compression{~glossary} standpoint is the {glossary:dissolve}Dissolve{~glossary}. This transition should be avoided. During a {glossary:dissolve}Dissolve{~glossary}, one shot is blended or gradually mixed into the next (see the bottom filmstrip in Figure 2). During the period of the {glossary:dissolve}Dissolve{~glossary} every {glossary:pixel}Pixel{~glossary} changes as the relative proportions of the two shots is progressively altered.

Figure 2 compares the {glossary:datarate}Data Rate{~glossary} impact of hard cuts and dissolves.

Figure 2: Data rate impact of hard cuts and dissolves
Figure 2 Data rate impact of hard cuts and dissolves

In the top graph, the orange bars indicate a low {glossary:datarate}Data Rate{~glossary} associated with a {glossary:lockedoff}Locked Off{~glossary} shot of a building with light traffic passing by in the foreground. Then an instantaneous spike of {glossary:datarate}Data Rate{~glossary} as the first frame of the new scene is introduced. In the bottom graph, instead of using a cut, a prolonged {glossary:dissolve}Dissolve{~glossary} has been employed resulting in one video sequence slowly blending into the other. It is obvious that during this period the {glossary:datarate}Data Rate{~glossary} must be increased because over the duration of the {glossary:dissolve}Dissolve{~glossary}, each frame is effectively a new frame, created from ever shifting proportions of pixels from the two separate scenes. Once the {glossary:dissolve}Dissolve{~glossary} is complete, the {glossary:datarate}Data Rate{~glossary} once again drops to the same low level associated with a static shot.

Aesthetically the impact of a {glossary:dissolve}Dissolve{~glossary}, which can be quite compelling on film or television, is often lost on the computer screen, deteriorating into a mess of chunky pixels. It is much better to make a clean break between scenes. In this way we take all the pain in one hit, rather than prolonging it over a second or more. It’s the video {glossary:compression}Compression{~glossary} equivalent of quickly ripping off a bandaid rather than peeling it off slowly.

If a transition other than a cut is required, then a ‘mechanical’ transition such as a {glossary:wipe}Wipe{~glossary} will be more readily compressed.

Key terms

{glossary:closeup}Close-Up{~glossary}
{glossary:dissolve}Dissolve{~glossary}
{glossary:establishingshot}Establishing_Shot{~glossary}
{glossary:fluidhead}Fluid Head{~glossary}
{glossary:lockedoff}Locked Off{~glossary}
{glossary:pan}Pan{~glossary}
Tilt
Transition
{glossary:wipe}Wipe{~glossary}