[mythtv-commits] Ticket #1558: Ignore timecode when decoding IVTV, EIA-608 Closed Captions

[MythTV]

#1558: Ignore timecode when decoding IVTV, EIA-608 Closed Captions
------------------------------+---------------------------------------------
 Reporter:  gtgj at pacbell.net  |       Owner:  ijr
     Type:  patch             |      Status:  new
 Priority:  minor             |   Milestone:     
Component:  mythtv            |     Version:     
 Severity:  medium            |  
------------------------------+---------------------------------------------
 This patch works around a couple of issues:

 - Wrong timestamps used for IVTV CC.  AVF "lastccptsu" is taken from video
 packets, which is appropriate for ATSC CC but not for IVTV VBI packets.
 This issue results in the problem reported by Robert Johnston here:
 http://www.gossamer-threads.com/lists/mythtv/dev/184534#184534
 (It is possible to fix this by using the pkt->pts from the VBI packets but
 that timestamp does not seem to be normalized.)

 - ATSC CC timestamps not always in sequential order.  Although AVF
 DecodeDTVCC now receives data in the correct order, the corresponding
 "lastccptsu" is not always incrementing.  (There is probably a better way
 of calculating "lastccptsu".)

 -----
 This patch allows the CC decoder to ignore the timecode.   The main use of
 the timecode in the original (backend) V4L CC decoding was to detect false
 duplicate codes because VBI reads were not reliably synchronized to the
 frame rate.  When decoding in the frontend, embedded VBI codes should be
 reliable.  (The incorrect timecodes was defeating this duplicate
 detection, resulting in extra "erase" codes.)

 Ignoring the timecode also disables logic which postpones "scroll-up"
 caption updates by accumulating characters in the same caption line.  This
 was originally used to reduce the number of CC records when recording to
 NUV but this is not really an issue for AVF decoding.  Also, disabling
 this logic for AVF allows "scroll-up" captions to be displayed more
 accurately (less "bursty").  However, the depth of NVP txtbuffers needed
 to be increased to handle the larger number of updates.

 Some VERBOSE debugging has been added in CCDecoder.

 NOTE:  I do not have an IVTV card to test these changes but I have tested
 it on a sample IVTV stream provided by Robert.

-- 
Ticket URL: <http://cvs.mythtv.org/trac/ticket/1558>
MythTV <http://www.mythtv.org/>
MythTV