I need a "blind" Singlepass duplicate remover >> Food and Drink >> Our Forum
Our Forum

Our Forum



SearchSearch   Users   Registration   Entrance
Today: 23.07.2025 - 16:30:00
Pages:  1  

I need a "blind" Singlepass duplicate remover

Advertising


MessageAuthor

I have a tv cap of telecined progressive pal (I think) footage and the duplicates vary from every 5th frame to every 7th frame with odd duplicates apearing every 2nd 3rd or 4th frame from time to time , so Decimate doesn't work on it. What I'm looking for is a filter that will blindly remove any duplicates it finds, that only uses one pass and won't mistake still scenes as dupes.

---------------------

320Power

users




Statistics:
Messages: 398
Registration: 12.07.2003
29.11.20 - 16:56:33
Message # 1
RE: I need a "blind" Singlepass duplicate remover

What's the frame rate of your TV cap ? What frame rate are you aiming for ? Are you planning to encode it using variable frame rate -VFR ? What you describe is commonly known as "hybrid" material. Take a look at the apppropiate modes for processing this kind of material with Decimate. Also, take a look at the very flexible TDecimate alternative, included in the TIVTC package from Tritical.

---------------------
Both sold. Now: Black 94 E34 530i-5sp

Carzzi

users




Statistics:
Messages: 745
Registration: 04.18.2002
29.11.20 - 17:01:00
Message # 2
RE: I need a "blind" Singlepass duplicate remover

Is there any greater pattern? If you can find one, you can use tdecimate with cycle&cycler, eg, cycle=41, cycleR=7 would drop 7 of every 41. If not, or if the patterns are fixed for long stretches, it's probably hybrid, but it's a sort of hybrid no tools currently cleanly deal with (tdecimate's hybrid modes only work on cycle=5).

---------------------

95M3032679

users




Statistics:
Messages: 44
Registration: 11.29.2003
29.11.20 - 17:04:51
Message # 3
RE: I need a "blind" Singlepass duplicate remover
FTP losing extensions : Previous topic
Pages:  1  

The administrator has prohibited guests from replying to messages! To register, follow the link: register


Participants