[Ffmpeg-devel] [PATCH] tunable vb_strategy=1

[Corey Hickey]

Jeff Clagg wrote:
> On Mon, May 29, 2006 at 06:47:11PM -0700, Corey Hickey wrote:
>> vb_strategy=1:b_sensitivity=n
>>  n   psnr  %I  %P  %B      q    qI    qP    qB    t1   t2
>> ---------------------------------------------------------
>>  1  42.95   2  31  66   4.75  2.89  3.18  5.55   228  387
>>  5  42.92   2  41  56   4.74  3.15  3.44  5.74   218  363
>> 10  42.90   1  47  50   4.71  3.35  3.61  5.81   213  350
>> 15  42.88   1  51  47   4.70  3.26  3.68  5.86   211  344
>> 20  42.86   1  53  44   4.68  3.44  3.73  5.88   210  340
>> 25  42.85   1  55  42   4.66  3.31  3.76  5.88   208  336
>> 30  42.84   1  56  41   4.65  3.35  3.79  5.90   207  335
>> 35  42.84   1  58  39   4.63  3.42  3.80  5.89   206  332
>> 40  42.83   2  58  39   4.63  3.38  3.82  5.91   206  338  *default*
>> 45  42.82   1  60  37   4.62  3.37  3.84  5.92   209  336
>> 50  42.82   2  61  36   4.60  3.52  3.86  5.92   209  332
>> 55  42.82   2  61  36   4.60  3.52  3.87  5.91   208  334
>> 60  42.82   2  62  35   4.60  3.56  3.88  5.92   208  334
>> 65  42.81   2  63  34   4.59  3.44  3.89  5.92   207  327
>> 70  42.81   2  63  34   4.58  3.67  3.90  5.92   204  326
>> 75  42.81   2  63  34   4.58  3.67  3.90  5.92   204  324
>> 80  42.80   2  64  33   4.58  3.57  3.92  5.92   204  322
> 
> I get the idea this is sort of a low motion clip you're testing with
> here. Did you verify that you don't get excessive B-frames, and a PSNR
> drop, when lowering n on a higher motion clip?

No, not really. There's a pretty good mixture of both. It's 24000 frames
starting at 1350 seconds into The Matrix. Pretty early-on is a part
where lightning makes the whole scene go light-dark-light-dark,
alternating once per frame. That's a very easy place to see where
B-frames are harmful. Beside that, there's rain, more lightning, a scary
insectoid machine, Neo sliding through tubes...

I wasn't trying for a section with lots of motion -- I just wanted a
representative section with a good variety where I could do 17 tests
without taking forever. To fulfill your curiosity, though, here are the
numeric results from a 100-second section at 2929 seconds: Morpheus and
Neo flailing at each other. :)


 n   psnr  %I  %P  %B      q    qI    qP    qB    t1   t2
---------------------------------------------------------
 1  37.68   2  30  66   10.3  6.98  8.06  11.4    31  152
 5  37.73   2  74  22   9.37  6.90  8.86  11.2    25  127
10  37.73   2  82  15   9.21  6.76  8.97  10.8    25  123
15  37.73   2  86  11   9.11  6.85  9.02  10.3    25  123
20  37.73   2  88   9   9.06  6.77  9.03  9.88    24  121
25  37.72   2  89   8   9.05  6.90  9.03  9.90    24  120
30  37.71   2  90   6   9.02  6.88  9.04  9.55    24  119
35  37.71   2  91   6   9.02  7.12  9.02  9.65    24  119
40  37.72   2  91   6   8.99  6.94  9.01  9.45    24  118


...so, yes, you do have a point there, but I think it only comes into
effect when there's a lot of motion in the clip and only when
b_sensitivity is very low. Encoding a clip with this high a "motion to
length ratio" at that bitrate looks pretty rotten anyway, and 37.73 dB
PSNR is low. Still, moderately low b_sensitivity values handle this one
fine as well. Visibly, I can't really see any difference.

-Corey