I got this through a PM and couldn't figure out a good solution, so looking for bright ideas:
This is progressive content, encoded interlaced, NTSC=>PAL DVD conversion with inserted fields - but not in a fixed pattern that I can see. The inserted fields lead to field shifting/phase shifting and "combing", that might be treated with adaptive field matching like TFM. But since it has both native animation sequences at 12fps and 24fps , there are inserted duplicates (and triplicates in the 12fps sequences) when doing basic field matching from the field inserts
1) The OP wants to edit manually in premiere but wants help detecting and identifying the inserted duplicates so he can remove them; then he wants to manually insert duplicates in sections of his choosing . He doesn't want any errors in decimating the wrong frames or inserting a wrong frame, so any proposed automated solution has to be perfect
I thought of maybe using ConditionalFilter() , YDifferenceFromPrevious - with duplicate detection to flag frames or print out txt file, or replace with red frames so you can see them on the premiere timeline easily - but there is no way to distinguish between "wanted duplicates" (12fps sequences) vs. "unwanted duplicates" (in the 24fps sequences) - all duplicates will be detected
I've tried several combinations of TFM/TDecimate (various modes, 2passes, mp2a=false with a large cycle range), but there is always a mistake somewhere. Even tried bob deinterlacing first instead of field matching before farther manipulations/decimation operations . I tried trimming the first few frames off (to start with a fresh scene in case there were decoder/GOP issues from the cut), but there is always a small issue somewhere.
If there was some set pattern, some combination of selectevery might work, but I cannot see a repeating pattern
Or is there another completely different approach to correctly identify the wanted vs. unwanted duplicates , and/or get this back to 24p properly?
Or perhaps somehow identifying the inserted fields before field matching ?
+ Reply to Thread
Results 1 to 10 of 10
Last edited by poisondeathray; 3rd Feb 2014 at 20:07.
Without a fixed duplicate pattern I don't see any way of differentiating wanted from unwanted duplicates. I don't know if it will help but you can use TFM() to get a smart bob:
Interleave(TFM(field=1), TFM(field=0)) # for TFF, swap 1 and 0 for BFF
As long as there are no orphaned fields or field blending you always get full film frames.
There are orphaned fields, single field inserts. It makes it difficult to do automatically (detect or fix)
At least there are no field blends (there is a transition that blends into the next scene, but that is intended)
If you ignore the 1st scene, the 1st orphan occurs right at the beginning of the 2nd scene . It's a single field without it's partner . This is a unique field. (in contrast to inserted field duplicates)
Sorry that's bad description. It occurs right at the beginning of the 2nd scene ~00:00:00.440 . The 1st "even" field (since we should probably ignore the 1st scene become issue might occur depending on the way it was cut and what decoder is used)
I can't believe he's been manually doing this (and wants to continue manually doing it) for the entire set
I see it now.
So your guy wants to keep the orphaned field but throw away a duplicate elsewhere? Presumably where a field or frame was duplicated to make up for the difference between 23.976 fps and 25 fps.
All he really wants is a way to detect/show the inserted fields (or frames) easily in premiere, so he can manually find them more easily and delete them and introduce new duplicates of his choosing elsewhere to preserve sync. He especially doesn't like the 24p sequences with inserted duplicates (e.g. ~30 seconds the boat moving out)
It's me that is more interested in an automated solution or approach
That' s not really what I would call a "pivotal" field; the information really isn't that much different than the next. If it were up to me I wouldn't care about it .
DupMC creates a log file which might be helpful to detect duplicates.