2009-02-05, 08:10
I've been fighting XBMC/boxee for the last couple of months related to choppy, stuttery, or basically not smooth playback of video. I was seeing choppy playback, mostly noticable on slow panning scenes (video would sort of go smooth, stutter, smooth, stutter in a very random way). I was also seeing no frame drops on the media info screen and extra idle CPU, so it wasn't bandwidth or processing power.
I finally came to a revelation when looking through my log file for any random issues in playback when I noticed this...
25fps is weird to me since I'm running 24fps video into a 60hz set, so I thought maybe my encodes are wonky. I have only been using Handbrake (I run a mac, not much else) and pretty much always with the framerate set to "Same as source". I went back into handbrake and ran a new encode with a forced framerate of 23.976, played it in XBMC and it looks smooth.
It appears as though XBMC has some issues with frame timing on handbrake encoded videos with "same as source" fps. The ATV plays these back perfectly...
I'm running some more encodes tonight so I'll update this thread with some additional results. So far it looks promising...
I finally came to a revelation when looking through my log file for any random issues in playback when I noticed this...
Code:
ERROR: CDVDPlayerVideo::OpenStream - Invalid framerate 48000, using forced 25fps and just trust timestamps
...
NOTICE: fps: 25.000000, pwidth: 720, pheight: 480, dwidth: 720, dheight: 404
25fps is weird to me since I'm running 24fps video into a 60hz set, so I thought maybe my encodes are wonky. I have only been using Handbrake (I run a mac, not much else) and pretty much always with the framerate set to "Same as source". I went back into handbrake and ran a new encode with a forced framerate of 23.976, played it in XBMC and it looks smooth.
It appears as though XBMC has some issues with frame timing on handbrake encoded videos with "same as source" fps. The ATV plays these back perfectly...
I'm running some more encodes tonight so I'll update this thread with some additional results. So far it looks promising...