Samsung S3C6400X User Manual page 646

MULTI-FORMAT VIDEO CODEC
S3C6400X RISC MICROPROCESSOR
For encoding case, typically, 4 frame buffers are required. These buffers are used for storing incoming image from
camera or pre-processor, encoding, storing the currently reconstructed image, and previously reconstructed
frame.
Rotation/Mirroring
FIMV-MFC V1.0 supports rotation together with mirroring function for both incoming image for encoding and
output image of the decoder for display. The former is done by the PrP rotator module and the latter by the PP
rotator module.
PrP rotator module
FIMV-MFC V1.0 uses output from the rotator module as input to the encoder without additional bandwidth
consumption on the external SDRAM for rotation itself. The rotated image is sent to the local buffer of the motion
estimation module, and re-used in the intra-prediction module for intra-mode decision and residual computation
module. The PrP rotator module does not work in decoding process.
Figure 21.11. PrP rotator data flow
PP rotator module
The rotation/mirroring process in decoding process requires additional bandwidth because the video codec IP has
to re-use the un-rotated image for decoding the next image. Therefore, the rotated image is written to other
memory space. In this scheme, the display I/F has not to change memory space for displaying the decoded image
because subsequent rotated image is written to the same space. Of course, change of target frame buffer is
possible by setting registers that specify its base addresses. The PP rotator module does not work in encoding
process.
Preliminary product information describe products that are in development,
for which full characterization data and associated errata are not yet available.
21-20
Specifications and information herein are subject to change without notice.