cutting/joining x264 clips won't reencode with keyframes - a blurry mess!

[forgotpassword]

Would have said that I'm trying to use either VirtualDub or Avidemux but there wasn't enough room in the subject title.

I've been cutting and joining clips with VirtualDub in the Xvid and Divx formats for a couple of years. I just recently jumped on the x264 and Avidemux bandwagons.

At any rate, I have many clips in x264 that I'm appending and cutting parts out of. And when I go to save, I MAKE SURE that I am reencoding my work - not just copying the video streams directly. However, when I am playing the video and I get to a part where I cut something out, it acts like I just copied the video streams directly. That is, a frame will stay the same and blur where something moves before finally "snapping" back to what it's supposed to be. Or the colors will go crazy for several seconds.

It doesn't matter if I save it in x264, Xvid or uncompressed RGB. The video acts exactly the same no matter what format I saved it in. Although something interesting is that results are not the same between VirtualDub and Avidemux, even though I edited the clips exactly the same way. But the two are really similar nonetheless.

I suppose the obvious thing to do would be to batch reencode the clips to Xvid or Divx at a high bitrate so I can properly edit them without any noticeable quality. But I'd rather not do this as I've grown quite accustomed to the x264 format. Seems to be bit nicer than Xvid and Divx.

I'm wondering if there's a setting somewhere that I can change? Or something I can do differently or what?

Any suggestions would be greatly appreciated. Thanks!

[poisondeathray]

If you are appending & cutting but not on keyframes this will happen.

Since you are re-encoding anyway, do the editing in avisynth and you can cut & append anywhere. You can visually preview the script with avsp. Or if you use MeGUI it has a avs cutter

Uncompressed RGB should work, but only if you are editing that, not editing the h.264 footage then saving to RGB. But you uncessarily undergo colorspace conversions using RGB=>YV12 (=quality loss).

Another option is to use a lossless YV12 intermediate and edit that( like lagarith, huffyuv) where every frame is an I-frame and you can cut anywhere. The filesize will be huge , but smaller than uncompressed RGB24 (compared to avisynth which would only be a few kb.)