Merging VCube, Руководство пользователя

Страница: 249 / 291

Appendices : MPEG Settings (MainConcept Encoder) Page 249
So you get the sequence of fields: TB TBT BT BTB or grouped as frames: TB TB TB TB TB. The above would be consid-
ered 2:3 pulldown as it is 2 fields, 3 fields, 2 fields etc.
3:2 is the reverse:
frame 1: tff = 1, rff = 1 fields displayed: TBT frame 2: tff = 0, rff = 0 fields displayed: BT frame 3: tff = 0, rff = 1 fields
displayed: BTB frame 4: tff = 1, rff = 0 fields displayed: TB.
In this case you get the sequence of fields: TBT BT BTB TB or grouped as frames: TB TB TB TB TB.
In most cases the MPEG Encoder adjusts the necessary settings automatically, so that the Pulldown option
remains disabled.
GOP structure (interval between frame types)
I frames: These frames are also called Key Frames. All GOPs start with an I frame. I frames contain information for a
complete picture, and can be decoded independent of any other frame. I frames are the largest (and least com-
pressed) frames.
P frames: P frames are encoded using information from the previous I or P frame, and can only be decoded cor-
rectly if the previous I / P frame is available. P frames are smaller than I frames.
B frames : B frames are usually encoded using information from the previous I or P frame and the next I or P frame.
In this case, B frames can only be decoded correctly if the previous and the next I / P frames are available. B frames
are smaller than P frames. In addition, B frames can be encoded using only information from the next I / P frame
but then they are larger than if they were encoded using both the previous and next frame information.
As a general rule for practical settings: The GOP size (in frames) is specified with the I frame setting and it must be
a multiple of the P frame setting. When I frame is set to 1, all frames in the video will be I frames. When I frame is
larger than 1, it specifies the size of the GOP, and the P frame setting specifies how often P frames occur in the GOP.
If P frame is set to 1, the video will consist of only I and P frames. If P frame is larger than 1, B frames are placed
between the P frames and the video will consist of I, P and B frames. Larger GOPs will yield greater compression
but will possibly cause a loss of quality. We recommend using the default settings.
Auto GOP: This function always starts a new GOP when there is a scene change, I.e. the encoder sets an I frame. If
you choose None from the drop-down menu, there will be no scene detection. The Fast option is a quick method
of scene detection where no VCSD happens. During the motion search the application checks if a scene change
occurs, and - if yes - the P frame is encoded as an I frame. Then the encoder starts a new GOP. VCSD is the abbrevi-
ation for Visual Content
Scene Detection, which is a better way of doing scene detection. At first, the VCSD is carried out, I.e. the analysis of
the frames, and then the GOP planning. It will yield a slightly slower encoding.
Closed GOP interval : This value specifies how often the GOPs should be closed and is only of importance if there
are B frames present in the GOPs. A value of 0 means do not close any of the GOPs, a value of 1 means close every
GOP and a value of 2 means close every other GOP etc. If a GOP is closed, it can be decoded by itself. If a GOP is not
closed, the first few B frames of the GOP will be dependent on the last P frame of the previous GOP and cannot be
decoded correctly without decoding the previous GOP first. When a GOP is closed, the first few frames of a GOP
are encoded so they only depend on the I frame in the GOP (the previous GOP is not required). This can be useful
for setting “chapter“points so a player can jump to these GOPs and can start decoding immediately without hav-
ing to read the previous GOP (or discarding the first few B frames).
Bitrate
Bitrate type: Shows current Bitrate mode Constant, Variable or Constant Quantization
Constant bitrate (CBR): Fixed bitrate (the relevant input prompt will be enabled if selected)
Variable bitrate (VBR): The minimum and maximum values define the bitrate range the encoder should stay within
while encoding. The average value is the desired average bitrate of the video stream. The relevant input prompts
will be enabled if selected.
Constant quantization affects the macroblock quantization value, approximately the “compression“of the macrob-
locks. Lower numbers yield better quality and larger files (larger bitrate results in less compression). The range is 1
... 31; 1 is probably excessive in that the quality does not improve much but the file size increases quite a bit. A
range is probably 3 ... 15 for constant quantization operation. In normal VBR/CBR modes, the encoder changes the
macroblock quantization value to adjust the bitrate; in constant quality mode it does not.