[FFmpeg-user] missing a selected frame

[Ulf Zibis]

Am 19.07.19 um 11:22 schrieb Stéphane Chauveau:
> On 7/19/19 9:18 AM, Moritz Barsnick wrote:
>> On Fri, Jul 19, 2019 at 00:21:53 +0200, Ulf Zibis wrote:
>>> ./ffmpeg -y -v warning -i debug/CYD_1.5m_x264.mp4 -vf
>>> select='eq(t\,10.16)+eq(t\,10.2)+eq(t\,10.24)+eq(t\,10.28)+eq(t\,10.32)+eq(t\,10.36)+eq(t\,10.4)+eq(t\,10.44)+eq(t\,17.52)+eq(t\,47.96)+eq(t\,49.08)+eq(t\,49.2)+eq(t\,55.72)+eq(t\,83.0)'
>>>
>>> -q 5 -c:a copy -vsync vfr debug/outsides/CYD_1.5m_x264_%02d.png
>> I'm not sure how ffmpeg's expression evaluation calculates 't' (means I
>> couldn't find it in the source), but seeing that your av_ts2minutestr()
>> certainly does it differently, I may point out: floating point equality
>> is a bitch. ;-) I'm saying: 10.2 may not be equal to 10.2,
>> computer-mathematically speaking.
>
> Using eq() on floating points is indeed a bad idea because of floating
> point rounding.
>
> It is probably better to use the frame number or the pts but if you
> really want to use time then you need to test an interval of time
> using the between() function.
>
> The size of that interval and the number of digits of the reference
> time must be carefully chosen.
>
> Here, the times 17.52, 49.08, ... appear to be given with a precision
> of 1/100s which is probably quite close to the framerate of your video.
>
> If that is possible you should try to add one digit to get a precision
> of 1ms.
>
> The size of the interval of time should be smaller than the framerate
> (so no more than 1 frame per interval) but larger than the precision
> used on the time values to insure that the rounded value remains in
> the interval.
>
> Assuming  time values with 2 decimals (so 10ms precision), a good
> interval could be +/- 15ms so a total interval of 30ms and a maximum
> framerate of 33 FPS which is not a lot.
>
> With 3 decimals (so 1ms precisions), you could use an interval of +/-
> 2ms so a total interval of 4ms and a maximum framerate of 250 FPS
> which should be good enough.
>
> So my advice is to add a 3rd digit to your time values and to replace,
> for instance,  eq(t\,47.96) by something like
> between(47.963\,t-0.002\,t+0.002).
>
> You can easily experiment with different intervals by creating two
> variables. For example, with your current times with only 2 decimals,
> you could try something like
>
> -vf 
> select='st(0\,t-0.015)\;st(1\,t+0.015)\;between(10.16\,ld(0)\,ld(1))+between(10.2\,ld(0)\,ld(1))+....'

Much thanks for your big effort. Before doing that, I think, I better
use "pts" than "t", even that is less human readable and more numbers to
type or copy.

-Ulf