[FFmpeg-user] ffprobe -show_frames and coded_picture_number
jiachielee at live.com
Thu Aug 15 04:35:58 CEST 2013
As I said, it’s all about the perceptional difference and the actual
difference. As long as you or any other users can accept the result, it’s
fine with me. The duplicating of the “wave patterns”?
Strictly speaking, a periodic function (wave) can be represented by Fourier
series, which is an infinite series. Infinite series is important in many
aspects of mathematics and Physics; as many of the common functions or
mathematical constants are defined by different infinite series. When it
comes to actual computation, do you think we can actually compute “infinite”
number of terms? We merely calculate the approximated answers, up to the
necessary accuracy; of course, we can only determine necessary accuracy from
empirical results. Mathematical periodic functions do exist; but when we
actually do the calculation, we always use approximation.
In the case of converting of analog signal (sound wave) to digital signals
(0 and 1), the only possible empirical results we can get are subjective
perceptional tests, which differ from one person to another. Technically
speaking, it’s impossible for any calculator or computer to compute the
actual “wave patterns”, so to speak. We only use good approximation. LPCM
audio data does not have the actual “wave patterns” from the analog signal,
which is the main reasoning behind the analog vs. digital debate. Through
resampling, we get digital data, which we later use to re-create the
approximated “wave patterns”, i.e. audio signal.
One way to increase accuracy of the approximated results is to avoid
rounding-off during the calculation. For computer, however, using floating
point or decimal number is inevitable but we can increase the number of
floating points to reduce errors; just like the producers use twice the
sampling rate during the production stage.
View this message in context: http://ffmpeg-users.933282.n4.nabble.com/ffprobe-show-frames-and-coded-picture-number-tp4660602p4660727.html
Sent from the FFmpeg-users mailing list archive at Nabble.com.
More information about the ffmpeg-user