[FFmpeg-devel] FFv1.3: Final testing - Results

[Paul B Mahol]

On 11/2/12, Peter B. <pb at das-werkstatt.com> wrote:
> On 11/01/2012 11:39 PM, Michael Niedermayer wrote:
>> take 10 files
>> take 10 ways to encode them
>> build a table of 100 cells each contains the percentage of how much
>> bigger the file gets with GOP 1 instead of GOP 100.
>> add 3 columns that contain the rowwise max, min and average
>> add 3 rows that contain the columnwise max, min and average
>>
>> please also make sure the result is correctly formatet and is
>> vertically aligned
> Here's what I got now, comparing GOP 1 to GOP 300 with different
> encoding parameters:
> (Properly aligned with a fixed-width font, but sorry it's wider than 80
> characters)
>
> //----------------------------------
>                                                    FFv1 (version
> 1)                                  FFv1 (version
> 3)
>
>
> Name                   PIX_FMT  Duration_TS raw       coder=0
> coder=1    context=0    context=1     coder=0     coder=1
> context=0    context=1    slices=4    slices=30    crc=0    crc=1
> min    avg    max
> akiyo_qcif             yuv420p          300 11140        0.96
> 0.9         0.96         0.87        0.91        0.84
> 0.91         0.79        0.91         0.81     0.91     0.91   0.79
> 0.89   0.96
> bowing_qcif            yuv420p          300 11140        0.97
> 0.92         0.97          0.9        0.94        0.88
> 0.94         0.85        0.94         0.86     0.94     0.94   0.85
> 0.92   0.97
> bridge_close_qcif      yuv420p         2001 74300        0.97
> 0.92         0.97         0.92        0.94        0.87
> 0.94         0.88        0.94         0.87     0.94     0.94   0.87
> 0.93   0.97
> bridge_far_qcif        yuv420p         2101 78012        0.98
> 0.93         0.98         0.93        0.95        0.89
> 0.95         0.89        0.95          0.9     0.95     0.95   0.89
> 0.94   0.98
> carphone_qcif          yuv420p          382 14188        0.97
> 0.92         0.97         0.92        0.94        0.89
> 0.94         0.88        0.94         0.88     0.94     0.94   0.88
> 0.93   0.97
> claire_qcif            yuv420p          494 18344        0.96
> 0.91         0.96          0.9        0.92        0.88
> 0.92         0.84        0.92         0.86     0.92     0.92   0.84
> 0.91   0.96
> coastguard_qcif        yuv420p          300 11140        0.97
> 0.93         0.97         0.93        0.94        0.89
> 0.94         0.89        0.94         0.89     0.94     0.94   0.89
> 0.93   0.97
> container_qcif         yuv420p          300 11140        0.97
> 0.92         0.97         0.91        0.92        0.86
> 0.92         0.85        0.92         0.84     0.92     0.93   0.84
> 0.91   0.97
> deadline_qcif          yuv420p         1374 51020        0.97
> 0.91         0.97          0.9        0.92        0.85
> 0.92         0.83        0.92         0.82     0.92     0.92   0.82
> 0.9   0.97
> flower_garden_422_qcif yuv422p          360 17824        0.97
> 0.93         0.97         0.92        0.93        0.89
> 0.93         0.87        0.93         0.83     0.93     0.93   0.83
> 0.92   0.97
> football_422_qcif      yuv422p          360 17824        0.98
> 0.93         0.98         0.94        0.95         0.9
> 0.95          0.9        0.95          0.9     0.95     0.95    0.9
> 0.94   0.98
> min                                                      0.96
> 0.9         0.96         0.87        0.91        0.84
> 0.91         0.79        0.91         0.81     0.91     0.91
> avg                                                      0.97
> 0.92         0.97         0.91        0.93        0.88
> 0.93         0.86        0.93         0.86     0.93     0.93
> max                                                      0.97
> 0.92         0.97         0.91        0.93        0.88
> 0.93         0.86        0.93         0.86     0.93     0.93
> //----------------------------------
>
> I've done it for the QCIF resolutions, just because that was the fastest
> results to regenerate, as I had to change the format and contents of the
> CSVs created during the testsuite.
> I can now easily (and quite quickly) generate these results for any
> input videos desired.
>
> Something I've noticed is that e.g. enabling CRC produces larger files
> (of course), but as the Derf videos are very short, that size difference
> disappears when rounding to 2 decimal places.
> This might have a more visible difference with longer files.

Why you need to rereun all the time? What is wrong with approach to run
once and display results in many ways? I'm really disappointed with all
this situation. btw, where are nice graphs in colors?