ryeguy
5th February 2007, 00:34
I am making a movie for a game. A movie exists that has great quality with xvid that I am trying to reproduce. The author of the movie doesn't want to give out his settings, so I'm trying to dig deep. The game is a first person shooter, so it's high motion and movement. Both of our movies are at 1280x720 @ 25fps. His quant distribution, using DRF analyzer, is pretty steady, yet no matter what I do I can't get the distribution he has with the quality but with the small filesize. I'm aiming for a filesize under 45mb/min, 50mb/min at most. One thing i DO know he did was use 2pass 5000 bitrate. My xvid settings are:
http://img249.imageshack.us/img249/5927/quantsjj7.jpg
Analysis of HIS movie:
Frame Type Statistics :
I Frames: 4.86%
P Frames: 67.84%
B Frames: 27.30%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 123 5.4%
DRF=3: 1020 44.9%
DRF=4: 501 22.1%
DRF=5: 298 13.1%
DRF=6: 207 9.1%
DRF=7: 108 4.8%
DRF=8: 10 0.4%
DRF=9: 5 0.2%
DRF>9: 0 0.0%
KeyF/DeltaF: 5.11%
KeyDRF<4: 108
KeyDRF=4: 7
KeyDRF>4: 1
AverageKeyDRF: 2.67
MAXDRF: 9
AverageDRF: 3.87
Deviation: 1.31
MY movie:
Frame Type Statistics :
I Frames: 4.71%
P Frames: 63.73%
B Frames: 31.56%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 1 0.1%
DRF=3: 559 60.1%
DRF=4: 62 6.7%
DRF=5: 0 0.0%
DRF=6: 92 9.9%
DRF=7: 212 22.8%
DRF=8: 4 0.4%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 4.95%
KeyDRF<4: 45
KeyDRF=4: 1
KeyDRF>4: 0
AverageKeyDRF: 3.00
MAXDRF: 8
AverageDRF: 4.23
Deviation: 1.72
I just don't get it, his is almost like a bell curve and mine is just random spikes at all quants. I've also noticed something: whatever I set the max quant to, the max is always used. Look at the percentages; the DRF3 is high because the Pframe quant is 2-3 (I know this is really low range, but I've tried larger ranges like 2-5 and they just group at 5 for example), and the 6 and 7's are high because of the Bframe quant ratio. How can I smooth out my distribution more to like his? The only factor anyone here might suggest is the source footage: it's not exactly the same, but both movies are "clips" of footage, so the compressibility of each second of the clip can be much different.
http://img249.imageshack.us/img249/5927/quantsjj7.jpg
Analysis of HIS movie:
Frame Type Statistics :
I Frames: 4.86%
P Frames: 67.84%
B Frames: 27.30%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 123 5.4%
DRF=3: 1020 44.9%
DRF=4: 501 22.1%
DRF=5: 298 13.1%
DRF=6: 207 9.1%
DRF=7: 108 4.8%
DRF=8: 10 0.4%
DRF=9: 5 0.2%
DRF>9: 0 0.0%
KeyF/DeltaF: 5.11%
KeyDRF<4: 108
KeyDRF=4: 7
KeyDRF>4: 1
AverageKeyDRF: 2.67
MAXDRF: 9
AverageDRF: 3.87
Deviation: 1.31
MY movie:
Frame Type Statistics :
I Frames: 4.71%
P Frames: 63.73%
B Frames: 31.56%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 1 0.1%
DRF=3: 559 60.1%
DRF=4: 62 6.7%
DRF=5: 0 0.0%
DRF=6: 92 9.9%
DRF=7: 212 22.8%
DRF=8: 4 0.4%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 4.95%
KeyDRF<4: 45
KeyDRF=4: 1
KeyDRF>4: 0
AverageKeyDRF: 3.00
MAXDRF: 8
AverageDRF: 4.23
Deviation: 1.72
I just don't get it, his is almost like a bell curve and mine is just random spikes at all quants. I've also noticed something: whatever I set the max quant to, the max is always used. Look at the percentages; the DRF3 is high because the Pframe quant is 2-3 (I know this is really low range, but I've tried larger ranges like 2-5 and they just group at 5 for example), and the 6 and 7's are high because of the Bframe quant ratio. How can I smooth out my distribution more to like his? The only factor anyone here might suggest is the source footage: it's not exactly the same, but both movies are "clips" of footage, so the compressibility of each second of the clip can be much different.