Video engineering & media pipelines

Interlaced or Progressive: Fields, Frames, Combing and Deinterlacing

Progressive video stores each frame as a complete picture captured at one moment. Interlaced video stores each frame as two fields, one holding the odd lines and one the even lines, captured a fraction of a second apart. Interlacing gave television smooth motion within limited bandwidth, but modern screens are progressive, so interlaced footage has to be deinterlaced for display; done badly, moving edges show comb-like stripes. Neither scan type is better in every case: progressive suits modern delivery, while interlaced originals are often best kept as they are in an archive.

7 min read · Updated

Fields versus frames

A progressive frame is what most people imagine video to be: every line of the picture captured at the same instant and shown together. Labels such as 1080p and 720p mean progressive.

An interlaced frame is two half-pictures woven together. The first field contains lines 1, 3, 5 and so on; the second contains lines 2, 4, 6. Crucially, the camera captured them at different moments, one field interval apart. So 1080i at 25 frames per second is really 50 separate moments per second, each with half the vertical detail. Labels such as 1080i, 576i and 480i mean interlaced, and the rate is sometimes written as fields per second, as in 1080i50.

Which field comes first is called the field order: top field first or bottom field first. HD formats generally use top field first, while standard-definition DV is commonly bottom field first. Getting the order wrong makes motion jerk backwards and forwards.

Why interlacing existed

Early television had to deliver smooth motion through limited broadcast bandwidth to cathode-ray tubes that would flicker visibly if refreshed too slowly. Sending half the lines in each pass doubled the rate at which the screen was refreshed without doubling the signal. On a CRT, each field was drawn in its own moment and the eye blended them, so interlacing looked natural. The link between field rates and mains frequency, and the 25 and 29.97 families that came from it, is explained in video frame rates.

Combing and other interlacing artifacts

Flat-panel displays, computers and phones show whole frames. When two fields captured at different moments are simply displayed together, anything that moved between them appears twice, offset line by line. The result is combing: fine horizontal teeth along the edges of moving objects, most visible on fast motion, pans and scrolling text.

Other artifacts include interline twitter, a shimmering on fine horizontal detail such as striped shirts or text, and field-order errors, where motion stutters back and forth. Not all footage with an interlaced flag actually shows these: film transferred to video and progressive material carried in an interlaced signal, sometimes called progressive segmented frames, can have both fields from the same moment, so they recombine perfectly.

Deinterlacing methods compared

Deinterlacing converts fields back into progressive frames. Every method trades sharpness, smoothness, artifacts and processing time.

Weave
Combines the two fields into one frame. Perfect for static scenes and for film or progressive content in an interlaced wrapper; combs on any motion
Bob
Expands each field into a full frame by interpolating missing lines, doubling the frame rate. Smooth motion, softer vertical detail, slight flicker on fine lines
Blend
Averages the two fields. Removes combing but leaves ghosted double images on motion
Discard
Keeps one field and drops the other. Simple and artifact-free, but loses half the vertical resolution and half the motion
Motion-adaptive
Weaves static areas and interpolates moving areas. A good general compromise and the basis of common tools such as yadif and bwdif in ffmpeg
Motion-compensated
Tracks motion across several fields to rebuild detail. High quality, much slower; QTGMC, used with AviSynth or VapourSynth, is a well-known example
Inverse telecine
For film transferred with pulldown, removes repeated fields to recover the original film frames rather than inventing any

A second choice is output rate. Single-rate deinterlacing produces one frame per original frame, for example 25p from 50i, and loses half the motion moments. Double-rate deinterlacing produces one frame per field, 50p from 50i, which keeps the fluid look of live television and sport. For a talking head, single rate is often fine; for movement, double rate looks closer to the original broadcast.

Deinterlacing needs re-encoding, with the usual cost in time and quality. If you are converting formats anyway, our guide to converting video formats covers the encoding side.

Where interlaced footage still comes from

Interlacing is far from gone. You meet it in:

  • Broadcast. HD television in many regions still uses 1080i, so off-air recordings and broadcast masters are often interlaced.
  • Older camcorders. DV, HDV and many AVCHD cameras recorded interlaced video, and some offered progressive only as an option.
  • Digitized tape. VHS, Hi8, Video8, Betacam and other analogue formats are inherently interlaced, so every digitized home video and archive tape is too.
  • DVDs. Standard-definition discs commonly carry 480i or 576i video, including film content with pulldown.
  • Some surveillance and industrial systems, which kept interlaced formats for compatibility.

The other direction exists as well: phones, webcams, screen recorders and almost all modern cameras record progressive.

How to tell whether a file is interlaced

  1. Check the metadata. MediaInfo reports scan type as interlaced, progressive or MBAFF, along with the scan order.
  2. Do not trust metadata alone. Files are sometimes flagged progressive when the picture is interlaced, or the reverse, especially after earlier conversions.
  3. Look. Step frame by frame through a section with horizontal motion on a computer screen. Combing on moving edges means interlaced fields from different moments.
  4. Measure. ffmpeg's idet filter analyzes frames and reports how many look interlaced, progressive or undetermined, which helps on long files.
Example: a digitized family tape

Hypothetically, a family digitizes a 1990s VHS tape of a birthday into a 576i file. Played on a laptop, a child running across the room shows heavy combing. They keep the digitized file untouched as the master and make a double-rate deinterlaced 50p access copy with bwdif for sharing. The running child looks smooth in the copy, and the soundtrack of the speeches is unaffected because only the picture was processed.

Trade-offs by situation

There is no universal winner; the right choice depends on what the file is for.

Archive or preservation master
Many archives keep the original interlaced scan and field order untouched, so better deinterlacers in future can be applied to an unaltered source
Web, phone and social delivery
Deinterlace, because these destinations display progressive frames and some platforms deinterlace poorly or not at all
Broadcast delivery
Follow the broadcaster's technical specification, which may require interlaced output
Editing mixed footage
Set the project's scan type deliberately and deinterlace the minority footage once, with a good method
Scaling or cropping
Deinterlace first; resizing interlaced video directly smears the two fields together
Speed of processing
Motion-adaptive methods are fast enough for most jobs; motion-compensated methods are worth the time for important archive access copies

Common mistakes follow from the same table: deinterlacing the only copy of an archive tape, scaling interlaced video before deinterlacing, using the wrong field order, and applying a deinterlacer to content that is really progressive, which softens it for nothing.

What interlacing means for audio and mydubly

Interlacing is purely a property of the picture. The audio stream is separate and is not changed by deinterlacing, so speech, music and timing are unaffected as long as the conversion keeps the original duration.

That is also why scan type does not matter to mydubly for transcription and translation. Only the soundtrack is used: the default audio track is decoded in your browser, downmixed to 16 kHz mono and sent in compressed chunks over HTTPS, while the picture stays on your device. For a dubbed video, the original picture is copied packet by packet without re-encoding, so interlaced footage stays interlaced and mydubly does not deinterlace it. If the video codec cannot go in MP4, the output is MKV. If you want a progressive version, deinterlace in an editor or with ffmpeg before or after. Digitized tapes bring their own audio challenges, covered in transcribing old tapes, and each file can be up to 2 hours long.

Next step: check your oldest footage

Open a broadcast recording, camcorder clip or digitized tape in a media inspector, note its scan type and field order, and step through a moment of motion to confirm. Keep any interlaced original as your master and make a deinterlaced copy for sharing. If you need its speech transcribed or translated, how to transcribe a video and the mydubly video translator work from the file as it is.

Frequently asked questions

Is 1080i worse than 1080p?

Not in every respect. 1080i captures twice as many moments per second as 1080p at the same frame rate, but each moment has half the vertical detail, and it must be deinterlaced for modern screens. For delivery to computers and phones, 1080p is simpler; for broadcast workflows, 1080i may be what the specification requires.

Why does my old video have lines on moving objects?

Those comb-like lines appear when interlaced fields captured at different moments are shown together on a progressive screen. Deinterlacing with a motion-adaptive or motion-compensated method removes them. Make the deinterlaced version a copy and keep the original file.

Should I deinterlace digitized VHS tapes?

For sharing and watching on modern devices, yes, using a good deinterlacer and preferably double rate to keep motion smooth. For the preservation master, many archivists keep the original interlaced capture unaltered. That way a better method can be applied later.

Which deinterlacing method should I use?

It depends on the footage and the time available. Motion-adaptive filters such as yadif and bwdif are fast and good for most material, motion-compensated methods give higher quality at much greater processing cost, and film content with pulldown should use inverse telecine instead. Static or progressive-source material may need no more than weaving the fields.

Does deinterlacing affect the audio?

No. Deinterlacing processes only the picture, so the audio stream is untouched. As long as the conversion keeps the original duration, sync stays as it was.

How do I find the field order of a video?

Media inspectors such as MediaInfo usually report it as scan order. If the metadata seems wrong, deinterlace a short section of motion with each field order and look: the wrong order produces jerky back-and-forth movement. HD footage is usually top field first and standard-definition DV commonly bottom field first.