[FFmpeg-user] scaling algorithms

[Paul B Mahol]

Dana 17. 10. 2015. 13:29 osoba "Michael Koch" <astroelectronic at t-online.de>
napisala je:
>
> Hi all,
>
> I have some questions about the scaling algorithms in FFmpeg.
> https://www.ffmpeg.org/ffmpeg-scaler.html#sws_005fflags
>
> 1. Which is the default algorithm?
>
> 2. I have many pictures with resolution 5472x3648, which are combined to
make a video with resolution 1800x1200. Any pixel in the video must be
calculated from 3x3 pixels in the input pictures.
> The input pictures contain stars in the night sky which are slowly
moving. Most stars are so small that they are just one pixel.
> If I use a "nearest neighbor" algorithm, then the star will blink on-off
in the video.
> If I use any averaging algorithm, then the contrast will go down because
the star is averaged with the dark neighbor pixels.
> Which is the best algorithm in this case? I need something like the
maximum value out of the 3x3 neighborhood.

That looks like the job for the filter, see inflate, deflate, erosion and
dilation filters.

>
> 3. Please improve the documentation. For example: ‘experimental’ Select
experimental scaling algorithm.
> Wow, that sounds interesting, but what does it do?
>
> Thanks,
> Michael
>
>
>    --    **********************************************
>       ASTRO ELECTRONIC   Dipl.-Ing. Michael Koch
>            Raabestr. 43   37412 Herzberg
>               www.astro-electronic.de
>       Tel. +49 5521 854265   Fax +49 5521 854266
>    **********************************************
>
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user