We seek to the exact position of the keyframe for direct stream/remuxing, but FFmpeg seeks to the previous keyframe when the exact time is provided as input. To work around this, add a 0.5 second offset to the seeking time.
Signed-off-by: gnattu <gnattuoc@me.com>
The library will also be updated when the number of children remains the same before and after scanning. For example, if one video file is removed and replaced with another version, we still need to remove the old one in this case.
Signed-off-by: gnattu <gnattuoc@me.com>
* Use more accurate rounding in GetFixedOutputSize
Signed-off-by: gnattu <gnattuoc@me.com>
* Force trickplay thumbnails to have even width
Signed-off-by: gnattu <gnattuoc@me.com>
* Use Convert.ToInt32
Signed-off-by: gnattu <gnattuoc@me.com>
* Force video size as thumbnail size if the trickplay width setting is larger
This will fix an issue when the trickplay setting contains a very huge width, but the video has a lower resolution than that setting. Our scaling filter logic will not do any upscale, and we have to force to use the video width
Signed-off-by: gnattu <gnattuoc@me.com>
---------
Signed-off-by: gnattu <gnattuoc@me.com>
* Move calculation of LUFS to a scheduled task as it's pretty slow
* Correctly calculate album LUFS
* Don't try to convert replaygain tags to LUFS values
This is an alternate approach which is more proper, but changes all parts that uses/overrides the original ValidateChildren method
Signed-off-by: gnattu <gnattuoc@me.com>
Bond_009
gnattu
Jellyfin Release Bot
Joshua M. Boniface
Shadowghost
Cody Robibero
Tim Eisele
gnattu
Nick Ganter
Niels van Velzen
Claus Vium
Nyanmisaka