TMPGEnc - Motion Search Precision

[theboywonder]

Motion Search Precision:

NORMAL --- HIGH --- HIGHEST?

What do you use?

Switching from HIGHEST to HIGH cuts my encode time by quite a lot.
(a 29min movie took 9.5hrs with highest, and only 5.5 hrs with high (svcd cbr + toolame for audio))

... however, i'm worried i'm sacrificing quality ...
i looked but couldnt really notice (the bitrate was set @ ~3500kbits)

... but for a 'normal' movie encode (45-60 mins per disc) using lower bitrates, would the difference be significant?

Thanx for your opinions
Robin
__________________________

550mhz athlon
256mb sdram (pc100)
16mb voodoo 3 3000
(just incase anyone wonders about the encode times)

[theboywonder]

162 views but no-one can be arsed to answer....

THANX guys / gals .....

[kwag]

Try "Motion estimation search (Fast)". It make a better job on reducing artifacts 'Mosquito effect" around edges.

-kwag

[theboywonder]

is that for svcd only, or vcd aswell?

i know that the default templates with tmpgenc specify 'motion estimation search' for svcd and 'normal' for vcd...

?

[adam]

Between highest and high you're not going to notice much difference. Between all the other modes you probably will. I would not recommend going lower than high or normal. I always use high for both svcd and vcd.

[markums2k]

162 views but no-one can be arsed to answer....

THANX guys / gals .....

Yowza... here's your definitive answer, Mr. Grumpy: use high.

Thank yew, thank yew.

[sterno]

For real stuff I use "high", like most other people. If I'm just encoding a short test clip I usually use "motion estimation search". The general consensus I've seen is that highest does give better quality than high, but it's not a huge improvement and it is a huge increase in encoding time.

[Deusxmachina]

I just did a test of these again today, actually. If you use a decent quality, such as an average bitrate of around 2000 or CQ of 85, etc, I could barely tell any difference between any of the settings. Reason being, there's more than enough bits to not macroblock during action anyway.

When I did the test at 70cq, the differences and variance in encoding time was more apparent. Especially the encoding time. The encoding time was almost the same up to normal, and then high was a bit longer, and then highest was slow as hell. It makes sense since with less bits to use, it has to take more time to make use of what bits it has.

[theboywonder]

thanx for the responses guys.... i'll stick with HIGH all round then!

(probably still use HIGHEST on my spare computer though - it doesnt get used except for video editing).

markums2k, i wasn't being grumpy, I just thought that no replies out of 162 views was rather disappointing! I always try and help out on forums

No harm anyways