BluDChyLD
7th September 2003, 13:54
Following the info in some threads I made I decided to make some more tests using fast PV. I encoded "irreversible" with no PV and again with fast PV.
I've studied sections of each encode and I can't see any differance - macroblocking and "mosquito" noise appear in the same areas on both encodes and freeze frames look the same during fast motion scenes... Fast PV adds an extra 1hr 15mins onto my encoding time, and (in this case) for very little benefit!
I then ran DRFAnalyzer on both avis, here are the results -
No PV used -
DRF=1&2: 16918 12.1%
DRF=3: 68335 48.9%
DRF=4: 52945 37.9%
DRF=5: 0 0.0%
DRF=6: 1497 1.1%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.54%
KeyDRF<4: 748
KeyDRF=4: 1
KeyDRF>4: 1
AverageKeyDRF: 2.33
MAXDRF: 6
AverageDRF: 3.30
Deviation: 0.72
Fast PV used -
DRF=1&2: 8530 6.1%
DRF=3: 67476 48.3%
DRF=4: 60926 43.6%
DRF=5: 0 0.0%
DRF=6: 2723 1.9%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.57%
KeyDRF<4: 787
KeyDRF=4: 3
KeyDRF>4: 0
AverageKeyDRF: 2.46
MAXDRF: 6
AverageDRF: 3.45
Deviation: 0.70
Although it isn't noticeable during playback, the encode with Fast PV actually uses higher quantizer values which (in theory) results in lower quality. Im curious - has anyone noticed a quality increase using PV in the new version?
I've studied sections of each encode and I can't see any differance - macroblocking and "mosquito" noise appear in the same areas on both encodes and freeze frames look the same during fast motion scenes... Fast PV adds an extra 1hr 15mins onto my encoding time, and (in this case) for very little benefit!
I then ran DRFAnalyzer on both avis, here are the results -
No PV used -
DRF=1&2: 16918 12.1%
DRF=3: 68335 48.9%
DRF=4: 52945 37.9%
DRF=5: 0 0.0%
DRF=6: 1497 1.1%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.54%
KeyDRF<4: 748
KeyDRF=4: 1
KeyDRF>4: 1
AverageKeyDRF: 2.33
MAXDRF: 6
AverageDRF: 3.30
Deviation: 0.72
Fast PV used -
DRF=1&2: 8530 6.1%
DRF=3: 67476 48.3%
DRF=4: 60926 43.6%
DRF=5: 0 0.0%
DRF=6: 2723 1.9%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.57%
KeyDRF<4: 787
KeyDRF=4: 3
KeyDRF>4: 0
AverageKeyDRF: 2.46
MAXDRF: 6
AverageDRF: 3.45
Deviation: 0.70
Although it isn't noticeable during playback, the encode with Fast PV actually uses higher quantizer values which (in theory) results in lower quality. Im curious - has anyone noticed a quality increase using PV in the new version?