|A CreativeCOW Adobe After Effects Tutorial
Barend Onneweer describes the how, what and why of several methods of de-interlacing video-footage using Adobe After Effects.
All PAL, NTSC or SECAM video signals throughout the world are field-based, instead of frame-based. This means that each video frame is composed of two interleaved fields. One field consists of the even-numbered horizontal lines, the other field is made up of the odd-numbered lines. The two fields of each frame are captured 1/50 (PAL and SECAM) or 1/60 (NTSC) second apart, and a TV or video-monitor displays the alternating fields in the same way. The field that contains the first scan line from the top is called the upper field, and the other field is called (you guessed it...) lower field. On playback or export of interlaced video it's crucial to get the field order right (either upper field first, or lower field first) or you'll get your field order reversed, resulting in stuttering motion, and comb-like artefacts around the edge of moving elements.
DV footage, PAL or NTSC is always lower field first.
The reason for all this apparently lies in the dark ages of television, when the phosphors of television screens weren't up to the job of keeping illumination up for an entire 1/60th of a second. To reduce flickering of the image, the frame was divided in 2 fields, maintaining the continuity of the image.
UPPER FIELD ONLY (300%)
COMBINED FIELDS (300%)
The above is an example of a scene that could have been shot on video, while the cow would be moving. The (enlarged) middle image shows only the uneven imagelines, and the image on the right shows that the two fields that are stored in a single frame are slightly offset in the area's where there's motion in the image, while there's no comb-effect at all in the background which is not moving (in case of a steady shot with no camera movement).
In modern times all the problems with fading phosphors have been solved. PAL and NTSC devices still output video as an interlaced signal, though. Although a couple of consumer and prosumer DV cameras like the Canon XL-1 claim to feature progressive scan of some sort, the PAL or NTSC output specifications still demand an interlaced signal from them. The difference here lies in that both upper and lower field were scanned at the same moment (often at the cost of resolution). Especially in PAL countries this is often used to get a 25-frames-per-second 'feel', approaching the 24 fps film look. True broadcast progressive scan though, is currently limited to High Definition Television. The progressive scan offered by DV camera's results in both fields being recorded at the same moment, but on playback on a TV the fields will still be displayed alternating.
So from here you basically have two choices: either you output to interlaced video, which means everything you do should be interpreted and rendered in fields, including titles, graphics and animations, or you decide to go for full frames and de-interlace the video footage and output 'no fields'.
There are many reasons to go with either approach, but basically the choice depends on the final medium on which the material will be viewed.
- For TV broadcast or distribution, you would normally go for field-interlaced video.
- For viewing on video-beamers you could probably go either way, but I prefer to work full frame, no fields (although most vido-beamers incorporate hardware-based de-interlacing and line-doubling circuits).
- For web-streaming or distribution on a CD-rom (basically anything that's going to be viewed on a computer monitor) you'll definitely want to work full-frame, no fields.
- For transfer to film, there are several concerns. If you're going to do a lot of compositing and animation or titling, you should work full frame, 24 or 25fps if possible. But if you're just going to transfer video footage to film, it's best to consult with the transfer facility, since they often have their own de-interlacing and pull-down processes.
Outputting interlaced video
When you're working with interlaced footage, or are designing motion graphics for display on PAL or NTSC monitors, you might want to make sure you have all your settings right. It's especially important get the footage interpretation right, that the software knows whether to put the upper field first, or the lower field first. As mentioned before, DV footage is always lower field dominant, but for other video sources it's important to verify that you have the right interpretation selected. Especially if you're going to do scaling, rotating, or moving the video around, separating fields and using the right field order is essential. Also if you are going to peed up or slow down your footage, you definitely need to separate the fields in the right order.
Animations that are done within software such as After Effects will automatically be rendered in fields if the field-render option is activated in the output options. If you're implementing material from other sources like 3D Animation software, it's best to render out your animation in the 3D application at 50 (or 60 if you're in NTSC) fps.
Upon importing the material in After Effects, set the frame rate to 50fps, drag into your 25fps comp and on output After Effects will create a field from every frame (if field-rendering is activated in the output module), thus maintaining the smooth motion inherent in field-based video.
De-interlacing video footage
There are a couple of reasons why you might not want to work with fields. I mentioned above that for projection or viewing on a computer monitor you'll want full-frame no-fields video. Adding to this, if you're going to take a still from video-footage to freeze-frame for an amount of time, you also need to de-interlace, even if you're outputting to interlaced video.
For animations and motion graphics this is easy. Just render out without field rendering enabled and you'll be fine.
For video, there are several methods of creating full frames from an interlaced video signal. The following is a description of the different procedures and how to apply them in Adobe After Effects, but similar operations should be possible in your own compositing application.
METHOD 1: PS-Deinterlace
METHOD 2: Blending the fields
METHOD 3: Nudge the footage
METHOD 4: The After Effects way
METHOD 5: ReelSmart FieldsKit
As shown in the procedures above, quite satisfactory results can be achieved using After Effects built-in resources. The ReelSmart FieldsKit is a safe buy though, and at the price I wouldn't look much further. The de-interlacer is very good, and the highly adjustable motionmask function works very well for most types of footage, preserving image quality and resolution where it can.
--Barend Onneweer is a leader in the Adobe After Effects Creative COW . Drop by and discuss this or other effects. Like to see who Barend is? Click here.