jjabba
12th July 2006, 12:58
I have a anime-DVD with a very peculiar frame pattern.
Some frames are progressive and some are interlaced.
The source is MPEG2 is tagged interlaced and the actual framerate is 29.97 (NTSC).
But it is not a hardcoded 3:2 pulldown stored at 29.97.
The interlaced frames seems to be composed of fields from 2 progressive frames with a slight timedifference, almost the
same way the frames in a "real" interlaced cam-recording would look.
These interlaced frames appear every so often, perhaps every 5th or 6th frame. But I can't see any pattern.
My usual solution would, of couse, be to somehow recreate the progressive material from each field in the interlaced frames.
But each field in those interlaced frames is the only record of the assumed progressive content. There are no fields in
the pre- or post- positioned frames that even remotely reassemble the combined fields in the interlaced frames.
thus making the recreation of the progressive material impossible by any conventional field combining.
Now I want to re-encode this DVD at a lower bitrate so I'm thinking the best way to go would be to just leave things unchanged and let DVDRB do it's best.
But It might also be a good idea to de-interlace/decomb the interlaced frames to make CCE encoding more efficient.
Any suggestions?
I've included a two different sequences of 3 successive frames grabbed with DGindex using the "ignore flag"-mode. Both shows the issue at hand.
Some frames are progressive and some are interlaced.
The source is MPEG2 is tagged interlaced and the actual framerate is 29.97 (NTSC).
But it is not a hardcoded 3:2 pulldown stored at 29.97.
The interlaced frames seems to be composed of fields from 2 progressive frames with a slight timedifference, almost the
same way the frames in a "real" interlaced cam-recording would look.
These interlaced frames appear every so often, perhaps every 5th or 6th frame. But I can't see any pattern.
My usual solution would, of couse, be to somehow recreate the progressive material from each field in the interlaced frames.
But each field in those interlaced frames is the only record of the assumed progressive content. There are no fields in
the pre- or post- positioned frames that even remotely reassemble the combined fields in the interlaced frames.
thus making the recreation of the progressive material impossible by any conventional field combining.
Now I want to re-encode this DVD at a lower bitrate so I'm thinking the best way to go would be to just leave things unchanged and let DVDRB do it's best.
But It might also be a good idea to de-interlace/decomb the interlaced frames to make CCE encoding more efficient.
Any suggestions?
I've included a two different sequences of 3 successive frames grabbed with DGindex using the "ignore flag"-mode. Both shows the issue at hand.