FFMS / ffms2

An FFmpeg based source library and Avisynth/VapourSynth plugin for easy frame accurate access
Other
574 stars 103 forks source link

fpsden-fpsnum don't work properly on RC4 #443

Closed Hackerpcs closed 1 month ago

Hackerpcs commented 1 month ago

I am using StaxRip to mostly encode from lossless SD HuffYUV avi, capturing VHS from ATI 600 USB and JVC HR-S7600EU VCR using AmarecTV. I use "fpsnum=25, fpsden=1" to force PAL 25fps because AmarecTV naturally inserts some frames when needed (start/stop of the tape, empty tape at the end with static, etc) because otherwise the result is less than 25fps, causing audio sync issues.

Using ffms2-2.40, the result correctly is that the duration being adjusted up when using "fpsnum=25, fpsden=1" with the added frames as stated in the documentation, e.g. on this ~3h recording, a total duration that before was 186:52, after using "fpsnum=25, fpsden=1" was adjusted up to 186:55

2 4 fpsden off

2 4 fpsden on

In contrast, if I instead use 5.0-RC4, the arguments don't work properly and the duration isn't adjusted, staying at 186:52 with or without the arguments and the end result has audio sync issues

I can't upload the source file because of its size but here is sample where I intentionally start and stop the tape to make the result have enough inserted frames. Due to the smaller duration the difference isn't that drastic but it there, the frames before (using Staxrip's Preview feature to see the resulting avs script) are 1932 and after if "fpsnum=25, fpsden=1" is used are 1943 which are the correct as seen in the AmarecTV's log (line 3171, "VT=00:01:17.773s(1943f)")

https://mega.nz/folder/XE4UzAZA#5co2GPSZm-z-j19wv4feKg

Dendraspis commented 1 month ago

Just a few little additions to this report, that could help:

myrsloik commented 1 month ago

Fixed here too

Dendraspis commented 1 month ago

Thank you very much. 🙂