CUBE's explanation of its inverse telecine technology in DVx.........
c-cube.com
Click on the link and scroll down to see a drawing that explains the process, or read the text below:
Inverse Telecine
Telecine, or 3:2 pull-down, is a process that converts film captured at 24 frames
per second into video running at 60 fields per second. Frames are scanned in an
interlaced fashion to create fields, but scanning each frame twice to create two
fields per frame would leave only 48 fields, so alternate frames are scanned
three times producing three fields instead of two. This means incoming frames
are scanned in a 3:2:3:2 cadence, so that 24 frames become 60 fields.
Because there are only two fields in a single frame, the first and third fields are
duplicated when a frame is scanned three times. However, information
indicating which fields are repeated is typically not provided to the encoder.
Therefore, the encoder must determine which fields are duplicates, so the
duplication can be eliminated before encoding takes place. The technique for
doing this is called inverse telecine.
Figure 5. Telecine and Inverse Telecine
Inverse telecine is a requirement for high-quality DVD authoring encoders
because a large amount of content that users encode onto DVD discs was
originally film material. As with most other aspects of MPEG-2 encoding, the
task is not as simple as it may appear. First, finding duplicate fields is difficult
because noise is introduced during the process (telecine is still primarily an
analog process), so pixel values from identical fields will usually not be
numerically identical.
Second, once a duplicate field is encountered, the encoder cannot assume a
3:2:3:2 cadence ensues from that point. Editing is frequently done on telecine
material, so the cadence can change at any time. This means that the encoder
must constantly search for repeated fields, whether they conform to a particular
cadence or not. This not only requires a great deal of processing power, it also
may lead to the encoder cutting out too many "repeated" fields in scenes with
minimal motion. That's because low-motion scenes can trick the encoder into
treating some fields as repeated fields when in fact they came from different
frames.
The codec architecture described in this paper provides hardware that performs
efficient computation and analysis of pixel data. This hardware combined with a
sophisticated software algorithm running on the CPU analyzes this data and
makes optimal field deletion decisions for inverse telecine. |
