|
While trying to accomplish the same thing (de-interlacing) there are practical differences in de-interlacing methods depending on whether the target frames are for display or for capture. While the Bob approach is good for display purposes it is not one that in practice can be used for capture and further processing. Primarily this is true because of the doubled frame rate. The expense of 60 frames/sec (either in disk i/o, bits used for further compression, etc) far surpasses the advantages of bob. Typically for capture/compression purposes one needs to stay in a 30frame/sec mode and can't afford to go to 60 frames/second. Thus any de-interlacing algorithm that includes temporal information from two fields in the de-interlaced frame result will have some motion blur / ghosting. Note that this is typically only an issue if the playback device or encoder can not maintain the 30Hz operation. Sure a single frame snapshot of a de-interlaced frame will have a ghost effect but, if played back at normal frame rates, our brains will tend to interpret this blur as natural motion. Wave your hand in front of your face real fast. You'll see blur motion there as well. But, as stated, if the playback is not a full rates or if you're compressing the captured video and the compression can not keep up at full rates then the blur becomes a distraction.
Note that these comments are applicable specifically to interlaced video. For film / telecined material you should do an inverse-telecine operation and get 24fps and not bother with de-interlacing.
As for the o500, the hw de-interlacer will tend to be better overall since it happens on the source video before any further video processing is done (scaling, cropping, color format conversion, etc). It also has some noise reduction elements thrown in it as well that the sw de-interlacer does not.
|
