Plugin Solutions For Final Cut Pro, Color, & Motion

Film Effects - Technical Information

How is film transferred to video?

This depends on whether the video is NTSC or PAL, or more specifically whether the video is 30 or 25 frames per second. Actually, each frame of video is made up of two fields. This dates back to the earliest days of television when the electronics of the time couldn't cope with displaying all 625 (PAL) or 525 (NTSC) lines each 1/25th or 1/30th of a second. To solve this problem, they changed the system so that it displayed half of the lines at double the frame rate. So now we have systems which work at 50 fields per second (PAL) or 60 fields per second (NTSC). Each field has half the vertical resolution, and, more importantly, is separated in time by 1/50th or 1/60th of a second. If an object in your video is moving fast enough then your camera will record it being in a different position for each of the two fields. If you view a still frame on the computer monitor, you will be able to see the two interlaced fields with a tearing effect at the edges of moving objects. This is perfectly normal, and will not be visible on a TV monitor when playing the video at full speed - because TV monitors display interlaced video, and computer monitors do not.

The traditional frame rate of movie film is 24fps (frames per second), and for transfer to video, these 24fps must be evenly distributed over the 25 or 30 fps. If the distribution is not even, the transfer will look bad. Because of the two different frame rates used in video (PAL and NTSC), there are two methods used to convert the 24fps.

Because of interlace, PAL and NTSC video are often referred to as 60i and 50i in terms of their frame rate, the "i" indicating interlaced. High Definition video shot at 24 frames per second doesn't have interlaced lines - it's called progressive scan, hence the frame rate is referred to as 24p. Standard definition video can also be recorded progressive, and it's referred to as 25p and 30p.

PAL (25fps)

The divisors of 24 are 1, 2, 3, 4, 6 & 12. The divisors of 25 are 1 & 5. They don't have any common divisors - the numbers 24 and 25 are said to be mutually prime. Fortunately, 24 and 25 are very close to each other, so to convert film's 24fps to PAL's 25fps, each frame of film is mapped to each frame of video, speeding up the film slightly (about 4%). Because film is not interlaced, the resulting video has no temporal difference between fields - they both represent the same instant in time.


NTSC (30fps, 60 fields per second)

A different method was invented to map film frames over to NTSC video frames. The divisors of 24 are 1, 2, 3, 4, 6 & 12. The divisors of 60 are 1, 2, 3, 4, 5, 6, 10, 12, 15 & 30. There are some common divisors of 24 and 60 so this means that, for instance, 6 divides into 24 and 6 divides into 60. 24/6 = 4, and 60/6 = 10, so each 4 frames of film must map to 10 fields of video, or 1 frame of film maps to 2.5 fields of video. Because we can't have half fields of video, we alternate the mapping of each frame of film to 3 then 2 fields of video which averages out to 2.5 This is called 3:2 pulldown.

Ok - enough mathematics - what does this mean for me?

It means that making PAL video look like film is a lot easier than making NTSC video look like film! The look of film on PAL is that of (in motion characteristics at least) de-interlaced video. De-interlacing is a term for a number of different techniques that remove the interlaced look from video, and hence allow the viewer of the video to only see 25 or 30 discreet image time instances each second, rather than the usual 50 or 60.
The most basic method is just to remove one of the fields, and fill in the gaps with the other one. This method works, but you will loose 50% of your vertical resolution, and this looks bad!
The second method is to blend between the two fields. This can add a slight amount of motion blur, but should keep a good proportion of the vertical resolution you started with. This method looks good.
The third method is "smart" or "adaptive" de-interlacing. There are many distinct ways to do this, but all rely on analyzing which areas of the video image are moving (and hence will exhibit interlace tearing) and which are not. Only those areas which are moving need de-interlacing, and hence the overall vertical resolution will approach a theoretical maximum, depending on the video being analyzed and the actual details of the method employed.
G Film Effects allows you to use field blending or smart de-interlacing for both PAL and NTSC. The field blending algorithm is, however slightly faster than smart de-interlacing. Also, the different look that each provides can be chosen for artistic reasons.

To make NTSC video look like film is more tricky. First we must select the which of the 24 frames we have extracted from the 30 we started with we want to add 3:2 pulldown to. There is not a right answer to which 24 we pick (because we can't have fractional fields), which is why G Film Effects has 5 different methods for you to choose from.

Once the 24 frames are chosen, they are spread back out in the 3:2 pattern to fit back into 60 fields per second. The 3:2 pulldown pattern causes a specific characteristic - that of Whole and Split Frames. Whole frames are caused when a frame of film lands in the 30fps video so that it occupies at least one whole frame of video - no interlace tearing will be visible. Split frames occur when it lands so that it takes up only one field of a frame, the other field being used by the next frame of film. There will be visible interlace tearing on the split frame, and this is perfectly correct and normal - it will look fine when playing on a TV monitor.

In G Film Effects you can choose the pulldown order by selecting from WWSSW, WWWSS, SWWWS, SSWWW, WSSWW, which are the only five patterns possible. This will not really change the visual effect of the filter, but may be useful for matching clips so that they all flow in a sequence (constant cadence).

G Film Effects also allows for other frame rates - 12, 15, and 30 (based upon using NTSC source video). They use a combination of the above methods, depending on the number chosen.

If you're using PAL, you will get 25fps, 20fps, 12.5fps and 10fps.



The Look of Film

The look of film is not just made up of the temporal effects discussed above. Film has a wider dynamic range than video, and does not record all brightness levels linearly. If you were to measure how different brightness levels are recorded on film, you'd get a graph that looks a little like a letter "S". G Film Effects has special controls to allow you to create and adjust the S shaped curve to your liking, and to adjust it for Red, Green and Blue separately. This allows you to change the colour balance of your video image for great effect.



The Film Effect

By combining temporal processing and image manipulation with correctly shot video, you have the basic film effect. Everything else in the G Film Effects package is designed to allow you to tweak the look of your film effect for artistic results.


©2007 Nattress Productions Inc.