panzerboy
Fractal Lover
Posts: 242
|
|
« Reply #1 on: December 05, 2011, 12:26:24 PM » |
|
I strongly suggest you use two pass. Unless your videos have constant complexity and motion, a fixed bitrate will hit the wall and not provide enough bits for the fast moving complex scenes and waste bits in the slow simple scenes.
Firstly in virtual dub drop down the the video menu set the fast recompress option. This avoids an unnecessary colorspace conversion and degradation, unless you're using filters. Again in the video menu select compression, select your Xvid codec and click configure. If your doing HD you may want to change the profile @ level to unlimited, otherwise you may be limited to the bitrate you can select. Select the encoding type to be "Twopass - 1st pass". Ok, Ok back to virtualdub and from the file menu "Run video analysis pass". There should be an Xvid status, make a note of the average bitrate that it shows when the video analysis is done. Back to video menu, compression, configure and select encoding type of twopass - 2nd pass. If need be click the Target size (kbytes) to change it to Target Bitrate. Enter in the average bitrate from the Xvid status and I guarantee you'll see no compression artifacts. Often you can get away with half of the average bitrate, especially if the average bitrate was crazy high, like 30,000Kbs. ok, ok back to virtualdub and now you can save as avi.
The bitrate you need will depend on the complexity and swiftness of movement. For my mandelbrot zooms I've used from 15,000Kbps to 7,000Kbps for 720p 30frame/sec videos. I once had an extremely detailed, fluff all over the screen, mandelbrot zoom that had an average bitrate of 50,000Kbps. I tried slowing it down to really boring speed but still couldnt get the bitrate under 35,000Kbps, so I abandoned it. Fractals can be a worse case scenario for video codecs. They assume there will be variation in detail and motion so they can steal some bits from low bitrate scenes to improve the appearance of the high bitrate scenes. So you might give a thought to not trying for the most complex detailed fractal you possibly can.
My other suggestion is to try H.264, theres an open source version called X264. You can get similar results to Xvid with half the bitrate. It may quadruple your encoding time, its much harder on the CPU. It doesnt have the nice status of Xvid to tell you the bitrate so you may have to experiment. Remember you only have to run the 1st pass once, you can run the 2nd pass as many times as you like. My poor old Pentium D can't quite serve the frames up fast enough in Windows media player and stutters quite bad, though VLC plays much better. I use H.264 now because the file sizes to upload to YouTube are 1/2 to 1/4 the size of my Xvid videos. I believe H.264 is YouTube's native format so they may mess (extra comprssion) with your video less if your bitrate is suitably low (5,000Kbps?).
That being said, with XVID you have the option of editing in virtualdub with direct stream copy setting, as long as you cut to key frames. This means you don't have to recompress and suffer a generation degradation. You just can't do that with H.264 because it puts keyframes in seemingly random places. It does this so it can reference previous keyframes again, think of a dialogue scene between two actors. The camera alternates between the two people so when the scene returns to a actor it can use the previous keyframe again. clever!
Of course the best way to do editing is with uncompressed video, and that's another story.
|