ViewCast 2120 Camcorder User Manual


 
Appendix: Advanced Settings for H.264
84 ViewCast
Setting
Explanation
Because teas (10) is not achievable in real time encoding there
is no reason to use it.
dia (diamond) is the simplest search. It starts at the best
predictor, checks the motion vectors at one pixel upwards,
left, down, and to the right, chooses the best, and repeats the
process until it no longer finds any better motion vectors. This
setting is the fastest.
hex (hexagon) is the default setting. It uses a similar strategy
to dia, except it uses a range-2 search of 6 surrounding points.
It is considerably more efficient than dia and hardly any
slower. This setting is a good choice for general use.
umh (uneven multi-hex) is considerably slower than hex. It
searches a complex multi-hexagon pattern to avoid missing
harder-to-find motion vectors. umh is about 40 % slower than
hex.
esa (exhaustive) is a highly optimized intelligent search of
the entire motion search space within range of the best
predictor. It is the mathematically equivalent to the brute
force method of searching every single motion vector in that
area, though faster. It is still considerably slower than umh.
teas (transformed exhaustive) attempts to approximate the
effect of running a Hadamard transform comparison at each
motion vector. It’s a little bit better than esa but a little bit
slower, too slow for practical use.
MotionEstimationRange
This value controls the maximum range of the motion search.
For hex and dia, the encoder only uses between 4 and 16. umh
and esa allow you to increase past 16 to allow a wider range of
motion search, which is useful for high-definition footage and
for high-motion footage. Increasing MotionEstimationRange
significantly slows down encoding.
MotionEstimationSubPelQuality
This value controls the subpixel motion estimation quality.
Higher numbers are better. Levels 1 through 5 simply control
the subpixel refinement strength. Level 6 enables RDO for
mode decision, and level 8 enables RDO for motion vectors
and intra prediction modes. RDO levels are significantly slower
than the previous levels.
0 fullpel only (not recommended)
1 QPel SAD 1 iteration
2 QPel SATD 2 iterations
3 HPel on MB then Qpel
4 Always Qpel
5 Multi QPel + bime