This page describe the design principles and notes of new (as in 1.1.0 version) input rotation code. Input rotation intelligently handles directory containing almost-homogeneous content, and perform a smart contatenation of inputs (as opposed to "blind cat" previously performed on 1.0.x and former releases). Application scenario and intended use of new code is described first; this lead to explanation of design princples. Design notes and some implementation details are then provided.
| Intended application cases | describe the common use-cases scenario(s) for multiple inputs |
| Design principles | design principles of input rotation infrastructure |
| Design notes - probing | design notes about probing and input rotation |
| Design notes - importing | design notes on how transcode (main program) deals with input rotation |
New input rotation code introduced in 1.1.0 is a revision from the ground up of how multiple input should be handled in transcode. First and foremost, some basic usage cases are identified:
Jack wants to make a video clip from a bunch of JPEG images.
Tyler records TV using transcode itself, and he wants now to transcode a group of low-compressed AVIs into XviD.
Marla has some NUV files and she wants to join them and transcode it to DVD.
Some key aspects emerge from examination of intended use cases:
Each input file can contains one or more source frame(s).
Multiple input need to be intelligently merged before to import (think to AVI case).
New code should handle grafefully and efficiently a lot of input files.
body of paragraph
body of paragraph
body of paragraph