Report on field synchronisation problems that can cause interlacing artefacts when encoding telecine material from video tape.
Continued from Part 1.
How does this affect telecine transfers?
So far we have only looked at the issues that relate to transferring video camera originated video to the digital domain. Because each field in a video frame is both spatially and temporally different from its neighbour, interlacing artefacts will always be present unless the image is processed in some way to reduce or remove the artefacts.
When film is transferred to video, adjacent fields may be duplicated (in the case of NTSC transfers), and pairs of fields may be generated from the same film frame which will make them temporally identical.
Which two fields do I choose to make a frame?
Definition of F1 and F2
Interlaced video signals have a natural two-field periodicity and F1 and F2 are the names we give to alternate fields in the sequence. If you look at the waveform of a video field on an oscilloscope, you can tell whether it is an F1 field or an F2 field by the shape of its sync pulses.
Sometimes the fields will be called ‘top’ or ‘bottom’ where ‘top’ refers to the field containing the first scan line and ‘bottom’ refers to the field containing the last scan line. Fields are also referred to as ‘odd’ and ‘even’ – this means nothing unless you know whether to start counting the first scan line as 0 or 1.
ITU-R BT.470-6 (formerly known as CCIR Report 624-1) defines “first field” (F1) and “second field” (F2) for 625-line PAL. We shall continue to use F1 and F2 terminology only.
The way in which two fields should be interlaced together to produce a picture depends on:
- which field is an F1 field and which field is an F2 field;
- whether the fields are from a 525- or 625-line signal.
It does NOT depend on:
- what “order” the F1 and F2 fields are in;
- anything relating to field dominance.
According to ITU-R BT.470-6 a 625 line video picture should be produced like this:
There are 288 “active” lines in each frame which make 576 in total. The remaining lines in a 625 line video picture are not visible. In fact, only the second half of the first line and the first half of the last line contain any picture data. When you look at a PAL video signal displayed on a good quality over-scanned monitor it is easy to see these half lines.
Video acquires field dominance as soon as a frame must be defined. A frame is either an F1 field followed by an F2 field or it is an F2 field followed by an F1 field. The field dominance is determined by the field which comes first in a frame.
An F1 dominant video frame is an F1 field followed by an F2 field. This is the protocol recommended by ITU- R BT.470-6 for PAL video.
Where does field dominance come from?
When video is created it has no field dominance. The output from a camera is a stream of F1 and F2 fields and there is no need for any concept involving frames:
At some point you will want to edit this material onto a tape containing different material. Assume that the existing material has no field dominance – it’s all black and burst for example. As most VTRs work in frames you need to make a decision: do you cut the new material into the old at an F1 field or at an F2 field? The inserted video will always be a whole number of frames (or an even number of fields) long.
Once you have made that first choice, all of the subsequent edits must be begin with the same field type. To see why this is so, assume we have edited together material A, B, and C following F1 dominance:
Now say we try and edit in material D to replace material B following F2 dominance:
We have created an unpleasant edit where one field of material B still pops up at the edit point. Field dominance is the protocol which video engineers invented to prevent this problem.
Most VTRs can be set to insert new edits starting with an F1 field, or an F2 field. This does NOT mean that a VTR can record an F2 field in place of an F1 field or vice versa. A VTR will ALWAYS replace a F1 field with an F1 field and an F2 field with an F2 field. Remember that the PAL and NTSC standards require the video signal to contain alternate F1 and F2 video fields.
By maintaining the same field dominance throughout the editing process, video engineers have made sure that no orphaned fields will be left at edit points. In fact, early editors will have been more concerned to make edits on 8 field boundaries in order to avoid problems with the PAL colour signals that have an 8-field periodicity.
Fields, dominance, and timecode signals
Timecodes on video tape only count in whole frames, so how are timecodes affected by field dominance?
Longitudinal time code (LTC) is recorded as a separate track next to the video signal on the tape. Sometimes one of the audio tracks is sacrificed for LTC timecode. The LTC time code is encoded into a signal that is exactly one video frame time long. As specified in SMPTE 12M 1999, the start of a LTC time code signal always coincides with a certain line near the beginning of an F1 field, regardless of the field dominance of the video material.
Vertical interval time code (VITC) is encoded into a non-visible line in each field. The same line in every field is used and that line can be any line from line 6 to line 22 for PAL. The first visible line in a PAL signal is line 23, so VITC does not interfere with the picture. The VITC code has a “Field Mark” bit which specifies if the field is F1 or F2. A VITC hh:mm:ss:ff value will always increment at a transition from F2 to F1, regardless of the field dominance of the video material.
So if your material is F1 dominant, then LTC and VITC behave as you would expect:
But if your material is F2 dominant, then LTC and VITC straddle your edit points like this:
This means that one frame of F2 dominant material consists of two fields with different timecodes!
MPEG-2 encoding of video
The MPEG-2 format is a frame-based format. In order for the compression algorithms to operate efficiently, two video fields are combined into a single frame. The resulting image is then compressed. All MPEG-2 encoders follow the PAL and NTSC standards in order to determine how to combine the F1 and F2 fields in order to make a frame. For PAL video, the MPEG-2 encoder will combine an F1 field with the next F2 field.
Telecine transfer to PAL video
The film we are interested in is shot at 24 frames per second, which is fairly close to the 25fps required for PAL video. The easiest way to transfer this film material to video is to speed it up to 25fps. This has two side effects. First, the film length is reduced by 4%. This is not usually a problem. Second, the pitch of the soundtrack will increase. If this is a problem the pitch can be reduced using a phase vocoder – this allows the audio to be speeded up by 4% without a change in pitch.
When the PAL video is created two video fields are captured from each film frame. These two fields will be an F1 and an F2 field, but for now, the order is unimportant. As each film frame is a still, there will be no movement of the image between the time that the F1 and the F2 fields are captured. This means that when the two fields are combined to display a still image in the digital environment there should be no comb artefacts or tearing at all.
The three images on the following pages are frames captured from an MPEG-2 encoding of a British Gaumont newsreel. There is no apparent combing or tearing of the image as you would expect for material telecine’d to PAL video. Furthermore, dust artefacts only appear on one frame. For example, there is a white patch in the top right quadrant of frame 15 which does not appear on frame 14 or frame 16.
The following images show the F1 and F2 field masked out from frame 15. The particular deck used is set to superimpose the VITC time code on the image and it displays a * after the timecode for the F2 field. The field images are created by superimposing a mask for F1 where all the F2 lines are black and the F1 lines are transparent. Vice versa for F2.
There are no problems with the preceding images. They illustrate that 24fps film telecine’d to PAL video should have no artefacts associated with interlacing at all.
However, we have found that some material does have artefacts. The image of the lorry at the beginning of this document illustrates that clearly. So what is going on? The next three images are still frames from a video tape where the telecine’d material does show interlacing artefacts. The images show what happens at a point in the film where there is a scene change.
The first and third frame both have interlacing artefacts that we have seen earlier, but most importantly, you can see that the middle frame comprises a field from the first scene, interlaced with a field from the second scene.