A Different Drummer – Understanding Frame Rates and Timecode

Frame Rates

 

All video, just like film, is simply a series of still images recorded and played sequentially to give the impression of continuous motion. Each of these still images is a frame, with definite size as well as temporal boundaries. The more frequently these “stills” are updated, the more fluid – and “real” – the motion appears to be. The rate at which images are recorded and played back is known as the frame rate and is measured in frames per second (FPS). Like most things in the evolution of film and video, frame rates have changed over the years.

 

The generally accepted frame rate for filming motion pictures, intended for theatrical projection, is a true 24fps. This was the speed settled on for sync sound when the “talkies” came into being, but there is nothing magical about that number. It could have been something else, but for better or worse, that is the rate we associate with film. As black-and-white television developed, the frequency at which images were updated was locked to the rate of electrical line current (AC), which is 60Hz in the US. In order to conserve bandwidth, television developers adopted interlaced scanning techniques (two fields for each frame), so the signal was divided into 60 fields (30fps), corresponding to 60Hz.

 

So far so good, until color television came along. For a variety of reasons having to do with concerns about interference, it was decided that the NTSC color TV signal could not have exactly the same rate as the electrical current. To facilitate this change, the TV signal was slightly “slowed down”, running at 59.94Hz (a rounded value), rather than a true 60Hz. Early black-and-white TV sets were able to stay locked to this speed change along with the new color sets. As a result, B&W sets weren’t instantly made obsolete by this technology change. Under this new NTSC standard, the video frame rate was also “slower”, running at 29.97fps (rounded) compared with the previous B&W rate of a true 30fps.

 

In Europe, the PAL color TV standard developed on a later time frame, permitting those engineers to take a somewhat different approach than their US counterparts. The European electrical rate is 50Hz, so TV engineers referenced their interlaced signals to that, resulting in 50 fields or 25fps. The decision was made not to grandfather the black-and-white technology, so their frame rate was never changed with the introduction of PAL color TV. PAL today uses a true frame rate of 25fps.

 

There is no problem as long as you are working in any one of these standards. Only when you convert from one to the other do you have to compensate in various ways. For example, films shot at 24fps are transferred to NTSC and PAL video formats. In the NTSC world, these films actually run through the telecine transport at the equivalent rate of 23.976fps (also expressed as 23.98fps). In addition, film frames are repeated to “create” additional video fields in a cadence referred to as “pulldown”.

 

Pulldown is a scheme to add six more frames to make up the difference between 24fps (23.976) and 30fps (29.97); however, it is important to note that time doesn’t change – one second of film duration is still one second of video duration. Film images are simply buffered and repeated to add more video images into this string of sequential images that we view as motion. There is a slight speed difference since the film is slowed from a true 24fps to the video rate of 23.98fps, but that is generally not an issue to the ear of the viewer.

 

A different approach is used in PAL countries. 24fps films are sped up as they run through the telecine to match the PAL video rate of 25fps. There is no pulldown, but there is a noticeable 4% speed increase, resulting in higher-pitched soundtracks. Although European theatrical films are generally also filmed at 24fps, many film-originated television productions are shot at 25fps, so that a one-to-one relationship with video and sound is maintained. These various frame rates are perceived differently in other countries. American (and other NTSC) viewers are used to 30fps (29.97) interlaced video, so when they first see PAL images, the TV sets seem to have a more pronounced strobe. This quickly wears off as our eyes (and brains) adjust. Pulldown and the lower image resolution of NTSC tends to look odd to PAL viewers when they first see NTSC images.

 

The choice of frame rates has been exacerbated with the increased use of high-definition video. As HD evolved, it become evident that video equipment manufacturers were no longer totally locked into the frame rates dictated by the legacies of NTSC and PAL. The adoption of additional frame rate choices has filtered down into standard-definition video gear. Current equipment offers these choices: 1) progressive scanning at 23.98fps, 24fps, 25fps, 29.97fps, 30fps, 50fps, 59.94fps and 60fps; 2) interlaced scanning at 25fps (50 fields) and 29.97fps (60 fields). Remember that these are all different ways to “slice” the same interval of time. Note, though, that some of these rates are “video-friendly” and some are not.

 

For instance, you can shoot HD video at a true 24fps or a “video-friendly” 23.98fps. The 24fps video can only be edited in HD and becomes difficult to downconvert and deal with in the standard-definition video world. It was intended for going out to film. On the other hand, if you shoot at 23.98fps, downconversions to SD come out correct and it is still easy to get back to film if you want to. In another example, European producers may opt to shoot HD at 25fps while US producers might shoot the same subject at 23.98fps or 29.97fps. Neither is directly transferable to the other video standard (PAL-NTSC or NTSC-PAL). Standards conversion equipment is needed as well as downconversion from HD to SD. Knowing the required end result will dictate the rate at which you shoot things in the beginning.

 

Timecode

 

The way editors and editing systems keep track of all of this is timecode. Timecode is an eight-digit clock signal which tracks hours, minutes, seconds and frames. The numbers are consecutive and ascending, starting at 00:00:00:00 and ending at 23:59:59:29. Each and every frame has a unique numerical identifier, which makes timecode ideal for editing. Add a four-digit videotape reel number along with the timecode, and it would be possible to find ANY frame in 10,000 hours of tape – if one-hour reels were used – and even more for other tape lengths.

 

The timecode count is locked to the frame rate of the video signal, so the frame indicator will go from :00 to :29 in 30fps timecode, :00 to :23 in 24fps timecode, :00 to :24 in 25fps timecode and so on. Even though the video is actually running at 23.98fps or 29.97fps when compared to a very accurate clock, the count is still based on even frames and even numbers. There is no .03 frame!

 

Timecode started out this way, but immediately ran into some issues. Because of the cumulative timing errors caused by the 29.97 versus 30 issue, a duration of one hour, as indicated by standard timecode, was in fact 108 frames too long. This error of more than three seconds is unacceptable if you are timing a network show that has to be edited to an exact duration. Drop Frame timecode was introduced to correct this. People frequently refer to 29.97fps timecode as “drop frame”, but this is totally incorrect. All NTSC timecode is referenced to a rate of 29.97fps. “Standard” timecode (called Non Drop Frame) and Drop Frame both run at the same rate and both number each and every frame. Drop Frame uses a scheme to skip certain numbers from the counting sequence to equal 108 frames in an hour. No frames are actually dropped – only numbers in the sequence. One hour of duration – as indicated by the drop frame timecode number – is exactly one hour.

 

Non drop frame timecode values are expressed with colons (00:00:00:00) and drop frame values use semicolons (00;00;00;00). Some software applications, like After Effects, also express timecode values with semicolons. They aren’t actually converting anything to drop frame values, but rather apply a drop frame counting scheme in order to express an exact duration. Drop frame timecode is generally used for broadcast television, while non drop frame timecode is used for nearly everything else. Modern editing system can easily mix and match and convert between both formats, so shooting field tapes in one code-base and editing in another is no problem.

 

Editing systems and editing software algorithms evolved around interlaced video. Consequently, editing HD formats with frame rates higher than 30fps poses certain challenges, because these systems aren’t set up to edit at the higher intervals. Video editing has always been frame and not field-based. As a result, producers who choose to work in the 720P format (1280×720 at 60fps progressive) have had some difficulties in finding systems which could be used for HD editing. Most of the NLEs don’t permit editing on a 60-frame timeline or controlling a VTR for edit-to-tape functions with 60fps timecode. One workaround that many manufacturers are investigating is the field marker. Most NLEs have tracked fields (1/60th of a second), even though they’ve only been able to edit on frames (1/30th of a second). This method might be used to develop 60fps editable timelines in upcoming software versions.

 

© 2003 Oliver Peters