Texas Instruments DM365, User Manual

Page: 125 / 131

E-1
Appendix E
Simple Two-pass Encoding Sample
Usage
Multi-pass encoding can be used to improve the quality of the H264 encoded
video. This version of H264 encoder on DM365/DM368 supports simple two
pass (STP) encoder. In STP encoder, two encoders run sequentially for every
frame captured, first the low-resolution encoder and then the high-resolution
encoder.
The low-resolution encoder accumulates the frame specific information
(metadata) in the structure described in the Section 4.4. After completion of
low pass encoding, metadata is passed to the high-resolution encoder. The
high-resolution encoder uses metadata appropriately to improve the quality of
the encoded video
Various example cases of simple two encoding are given below:
1) Case-1: If both encoder runs at same frame rate and no frame skip
occurs; generation and consumption of the frame level information
(metadata) happens for each frame.
2) Case-2: If there is a frame skip at low-resolution encoder; no
information is passed to high-resolution encoder for the corresponding
frame.
3) Case-3: If there is a frame skip at high-resolution encoder, the frame
level information is retained and is consumed at next frame. (For
example, if
SceneChange
occurs at Nth frame and it is frame skip, then
the IDR frame is inserted at high-resolution at (N+1)th frame.) This way
information is preserved and utilized at high-resolution encoder.
4) Case-4: If the low-resolution encoder is running at low frame rate and
high resolution encoder is running at high frame rate, the metadata is
consumed by the corresponding frame at high-resolution encoder. If
the corresponding frame is frame skip then metadata is used by
subsequent frame. This process continues if the frame skip persist and
the metadata is pass to next frame until the new metadata is received
from low-resolution encoder.
5) Case-5: If the low-resolution encoder is running at high frame rate and
high-resolution encoder is running at low frame rate, the metadata
information is retained inside the low-resolution codec until encoding at
high reolution encoder starts and used accordingly at the high-
resolution encoder.
.