FORMAT



Autor: Redaktion


Deinterlacing

Deinterlacing is the process of taking an interlaced video sequence and changing it into a progressive video sequence.

The process involves converting the different interlaced fields into frames. 
Deinterlacing is a very complex process and results can vary widely, with the simple techniques often resulting in image defects. Mixing progressive and interlaced materials is possible on a timeline, but the output will need to be converted to either a progressive or interlaced sequence when you output the final programme.

There are several methods for deinterlacing materials, and each have their own advantages and disadvantages. Some of these methods are available within popular non-linear editing tools or transcoding products.

Simple deinterlacing techniques include Weaving (joining together the fields, although this can prove problematic if an object has moved within the frame), Bobbing (simply taking the odd or even fields and doubling them to fill the empty fields), and Skip Field (simply ignoring the missing fields). The more common deinterlacing techniques are: linear, motion adaptive and object-based deinterlacing.

Linear Deinterlacing

For the linear technique, the system looks at adjacent lines of a field and then calculates an average between the two for the missing field. For example, if you wanted to fill the gap between even lines 2 and 4, you would generate a line (for the odd field, line 3) that matched the average between them.

Blend Deinterlacing

A similar technique to Linear deinterlace but this time it is done temporally.  The resulting image is calculated from preceding and following fields. This can be difficult to achieve for handheld shots or sequences with lots of movement in the frame. It works better for static fixed shots.

Motion Adaptive Deinterlacing

The motion adaptive technique attempts to detect changes in groups of pixels across different fields and frames. A computer algorithm is used to change between linear, bob, weaving and blending etc. to obtain the best fit of errors to fill in the gaps between lines and fields. This is an advanced process that is dependent on the deinterlacing system understanding the movement of individual pixels across a small number of fields and frames. This is very processor intensive to do, but can create excellent conversions if it gets the motion detection right. 

Object-based Deinterlacing

Object-based technology combines the approach of motion adaptive deinterlacing with the intelligence of trying to understand and track objects within a scene. The deinterlacer tracks different points in the image across multiple fields and frames to calculate the change and then create a progressive sequence that is object and frame motion generated. Again this is very processor intensive to do, but can create excellent conversions if it gets the motion detection right. 

Of course, the best way to avoid having to deinterlace in the first place is to plan thoroughly BEFORE the production to ensure that a continuous approach (progressive or interlaced) is chosen for the entire production.  Programme contribution guides and delivery specifications are available from the broadcasters and should be checked prior to starting a production.

Related Articles: