Frame Accurate Chapters

A quick note about MPEG2 compression
One of the main ways that the MPEG2 compression works to fit your video onto a DVD is by using "Motion Redundancy". Essentially, what this method does is to look at every frame of video and try to remember what parts of a scene have not changed and not save those parts of the scene, but simply refer back to the picture of them that it had moments before. Since NTSC video has 30 frames per second of video, this means that your tape is sending out 30 complete pictures every second that it wants to put on a DVD. If we were to save every single bit of every picture, we would need 5 DVDs just to record one 2 hour production. Instead, by looking backward and forward from frame to frame, the MPEG2 Encoder (the device that takes your video and "squeezes" it down for DVD) only saves the parts of a frame that have changed from one frame to the next.

In the case of a video of a person seated and talking, it is very likely that the only movement in the entire picture would be the person's lips, or at a maximum, their face. Since most of the scene (walls, desk, fixtures) are still, the Encoder can essentially tell each frame the following: "Ignore the room and the fixtures, because I've got a copy of them and just save the person's face". If the person's face only takes up 20% of the screen, then you might assume that 80% of the scene could be "compressed".

I-frames
With most professional video decks you can view each individual frame of your uncompressed video, but with MPEG2 you can only view the frames that contain 100% of the picture. These frames are called the "I-frames". The example below shows how MPEG2 video compresses each frame of your video. The DVD specifications only allow for chapters to be placed on I-frames. When you submit your frame accurate timecode to us for your chapter points we default the chapter point to the previous I-frame before your exact frame. This ensures that your video won't come in after the chapter has started. Because there are two I-frames for every second of your video, your DVD chapter point will be within 15 frames of your exact frame accurate point. Before we burn your DVD, we first preview your chapters to make sure they are in the best possible location.

In the original video above (your tape), the subject is only moving the red parts of their body. For the sake of this example, assume that all other things in the picture remain motionless. Since every second contains 30 frames of video, this example covers approximately 1/4 of a second.
	I-frame						I-frame
When the MPEG2 Encoder reads the stream of video, it grabs an initial frame that contains 100% of the video. For the next group of frames (generally 14 but only 5 for this example), the Encoder only stores the changes since the last full frame. Since only arms and legs are moving from one frame to the next and the rest of the scene is motionless, the only data stored for those frames is for the moving parts.