One aspect of video editing that takes some time for some people to understand is Fields. This includes getting it wrong in the first video, resulting in a few jerky sequences and transitions. Although most know what fields are, many don't realized all their implications when it comes to editing.

For the benefit of anyone else who has trouble or is confused with this, here is an explanation of fields as they affect video editing, as well as some tips on how to get best results.

Each TV frame is produced by scanning the screen twice, arranged so that the lines of the second scan fill in the gaps left by the first. Each of these scans is a field. So a 25 frame/sec TV picture is actually 50 fields/sec (30 & 60 respectively for NTSC - I'm English so I think PAL).

So to get one frame with maximum detail, you need to combine the information in both fields. Easy so far, but what happens when motion is introduced? Because the two fields are scanned sequentially in the camera, anything in the image that is moving is in a different place in the second field than it is in the first. This helps make for smooth motion in a TV picture, but is also the reason why fields can cause trouble when it comes to editing.

When a computer plays video on its monitor it only displays a sequence of complete frames, it doesn't use the TV trick of interlacing fields. So video formats designed for computer displays don't use fields and nor does MPEG-1. Some video editing software intended only for use with such formats is unable to handle fields properly, which makes it unsuitable for getting good results with video targeted for TV use. It's important to use an editor that can process fields if the target medium is intended for TV replay.

Capture card drivers often offer the option of capturing one field per frame (sometimes you can choose the first or second), or capturing both fields. In addition, when capturing both fields there is a question of what order the fields are packed into the frames of the file on disk. Most drivers pack them in the order they are scanned, field-A first, but some pack them field-B first. Your capture card's documentation should tell you which field order it uses, but if you can't find it try order A and do a field-dependent trial edit. If the result plays badly, try order B. DV Type-1 is always order A.

If you capture one field then the resulting video file is frame-based. In other words, it has one image per frame. If you capture both fields the file is field-based - each frame consists of two distinct field images, separated by 1/50 (1/60) second.

For best quality you will want to capture both fields, so knowing about field mode is important.

Most of an AVI file compiled by a video editor program consists of frames that are simply copied from an input file to the output file (assuming you're not changing the compression scheme). For this material it doesn't matter whether the files are frame or field based, the frame data is just copied verbatim - the output is the same as the input. But when you start adding effects, transitions, speed changes, etc. the difference becomes very important.

If the editor doesn't know that the input frames contain pairs of fields, it may do completely wrong things with the images. In addition, by creating the output in field mode, synthesised motion produced by transitions, titles, etc. is considerably smoother.

AVI files contain a lot of information identifying the format of their contents, but the ability to mark a file as containing dual-field frames was not put into the AVI specification. As a result you have to know what field layout your files contain, and set the editor options correctly yourself. Exactly how you do this depends on the video editor you are using: check your documentation.

As a general rule, if you capture your clips using both fields, set the appropriate field order as the default for imported clips, and use the same setting for creating the edited result. If you import clips in field mode but create video in frame mode, then video creation will be extremely slow as every frame will be re-rendered. If, on the other hand, you have a particular reason to convert a field-mode video to frame mode, then the following is the way to do it.

This is where it is very important that field mode files are correctly identified. If you reduce the speed of a clip, the effect is achieved by repeating frames. If the editor treats the clip as frame mode, then it repeats a whole frame. If this frame consists of two fields A & B, the resulting field sequence is A-B-A-B. This will jitter visibly when replayed. If it treats the clip as field mode, it creates the sequence A-A-B-B. In other words, each field becomes one frame. This can actually be quite useful for other purposes, as we can see later.

A similar effect occurs if you reverse a clip. If the clip is treated as frame mode, the frames are played in reverse, but each pair of fields within the frame are still played forwards. The result looks horrible! Identifying the clip as field mode ensures that the editor reverses the order of the fields within the frames as well.

If you want to capture a still image of a single frame from a video file, then the multiple fields can be a serious problem. If you use a video capture program to grab a single frame, the resulting image will probably contain both fields. This gives you the best detail if the scene is static, but if there is any significant motion present the difference between the fields shows up as a blurring. Let's assume you don't want this!

One solution is to re-capture the sequence using single-field mode in the capture driver. However, this may lower the resolution too far, or you just may not be able to re-run the original. Much simpler is usually to use your video editor to create the image, as it will look after the field problem at the same time.

Editors will normally let you select a single frame and save it to disk as an image file (MediaStudio lets you save a block of frames as an Image Sequence, for a single frame you just create a sequence of one). By setting the clip's field properties correctly, the editor should use just one field when creating the image, thus avoiding the field-blur problem. In MediaStudio you simply have to mark the clip as field-mode. In other editors you may need to specify de-interlace.

Another way to break a clip down into individual frames in an editor is to define the speed of the clip as 50% (half-speed). As described above, this has the effect of turning each input field into one output frame. If you expand the editor's timeline so you can see each frame, each frame of the expanded clip is actually one field of the original. You can now save whichever frame you want as a clean image.

If you capture clips using both fields, then set the default import mode in your editor to the correct field-order for your capture hardware. Also set the same field-order for creating result-files and previews. You shouldn't get any peculiar effects caused by the existence of two fields per frame.

If you're doing anything more subtle, especially working with still images, then understanding what fields are all about will help you get the best results