octapus
23rd February 2002, 22:21
I made two tests concerning differences between internal and external 2pass in XVID.
I used Koepi's 17 February version.
Target: Rush Hour 2
Duration: 1:26 hours
Resolution: 640x272
Neutral Bicubic
Target Video Size: 659 MB or 1067 kbits/sec ( 1kbit=1000 bits)
1st Pass as made using H.263 and motion search precision 6 and of cource no luminance masking
1) 2nd pass using stats file edited by guardianknot. the last 110 seconds of credits were given a 0% weight
2) 2nd pass using internal method and again setting the last 110 seconds in credits section at 5%
Both second passes had modulated quant type, luminance off, smooth quantizer flunctuation, 2-31 min-max quantizer, 2-31 IFRAME min-max quantizer,motion search precision 6
i encoded two parts from rush hour
a) the very first 1001 frames at the beginning
internal method gave the following results:
MPEG frames=385 - H.263 frames=617
MPEG bytes=3.116.056 - H.263 bytes=2.631.219
Total size(contains avi overhead)=5.779.456
Average quantizer=3.833333
external method gave the folowing results:
MPEG frames=571 - H.263 frames=431
MPEG bytes=2.679.381 - H.263 bytes=3.004.852
Total size(contains avi overhead)=5.715.968
Average quantizer=3.41018
b) 3003 frames chosen randomly from file (using selection start-selection end in virtualdub)
This happened to have some still scenes. And it was at about 1/3 of movie.
internal method gave the folowing results:
MPEG frames=2304 - H.263 frames=700
MPEG bytes=17.904.580 - H.263 bytes=2.117.572
Total size(contains avi overhead)=20.103.168 bytes
Average quantizer=3.248668
external method gave the following results:
MPEG frames=2269 - H.263 frames=735
MPEG bytes=17.652.217 - H.263 bytes=2.514.890
Total size(contains avi overhead)=20.248.576 bytes
Average quantizer=3.04494
The quality of the four clips was quite good but i like most the gurainknot-edited two of them
I used Koepi's 17 February version.
Target: Rush Hour 2
Duration: 1:26 hours
Resolution: 640x272
Neutral Bicubic
Target Video Size: 659 MB or 1067 kbits/sec ( 1kbit=1000 bits)
1st Pass as made using H.263 and motion search precision 6 and of cource no luminance masking
1) 2nd pass using stats file edited by guardianknot. the last 110 seconds of credits were given a 0% weight
2) 2nd pass using internal method and again setting the last 110 seconds in credits section at 5%
Both second passes had modulated quant type, luminance off, smooth quantizer flunctuation, 2-31 min-max quantizer, 2-31 IFRAME min-max quantizer,motion search precision 6
i encoded two parts from rush hour
a) the very first 1001 frames at the beginning
internal method gave the following results:
MPEG frames=385 - H.263 frames=617
MPEG bytes=3.116.056 - H.263 bytes=2.631.219
Total size(contains avi overhead)=5.779.456
Average quantizer=3.833333
external method gave the folowing results:
MPEG frames=571 - H.263 frames=431
MPEG bytes=2.679.381 - H.263 bytes=3.004.852
Total size(contains avi overhead)=5.715.968
Average quantizer=3.41018
b) 3003 frames chosen randomly from file (using selection start-selection end in virtualdub)
This happened to have some still scenes. And it was at about 1/3 of movie.
internal method gave the folowing results:
MPEG frames=2304 - H.263 frames=700
MPEG bytes=17.904.580 - H.263 bytes=2.117.572
Total size(contains avi overhead)=20.103.168 bytes
Average quantizer=3.248668
external method gave the following results:
MPEG frames=2269 - H.263 frames=735
MPEG bytes=17.652.217 - H.263 bytes=2.514.890
Total size(contains avi overhead)=20.248.576 bytes
Average quantizer=3.04494
The quality of the four clips was quite good but i like most the gurainknot-edited two of them