DVD Shrink Test

johnmeyer wrote on 5/11/2006, 10:27 AM
There is a long discussion in this thread:

2 hour Brick Wall - Fave DVD Templates?

about the best way to fit lengthy video (2+ hours) onto a single-layer, single-sided DVD. While many advised using two DVDs or using dual layer, for those that insist on doing this, the two ways advised were to either us the appropriate bitrate and then use 2-pass encoding or, the other alternative, use the best possible encoding (8,000,000 CBR) and then take the result of this and use DVD Shrink to get it to the required size.

I decided to create a test to see what quality changes I might see. This test is highly dependent on the video chosen, and I may not have chosen the best thing. It happened to be something that was already on my system. I took 30 seconds from a dance I filmed in HDV, and then made the following three clips, all rendered down (using Best) to 720x480:

1. I rendered starting with the NTSC DVD Architect 4:3 template, but changing the quality to Best (because I was downsampling from HDV to SD) and changing the bitrate to 8,000,000 CBR.

2. I rendered exactly as #1, but this time changing the bitrate to 2-pass VBR, with the maximum set to 8,400,000; the average set to 3,800,000, and the minimum left at 192,000. I chose 3,800,000 because that is the bitrate that will let you fit exactly 2 hours 30 minutes on one single-sided, single-layer DVD.

3. I took the result from #1 above (the high quality clip) and created DVD files using DVD Architect 3.0. I then used DVD Shrink to open those files and set the Custom Ratio to 50%. The reason for using this ratio is that at 8,000,000 bps, you can fit almost exactly 1 hour 15 minutes on a single DVD. Since we want to create a DVD of exactly twice that length, I need to shrink by 50%. I then took the main VOB file from this shrunk file set, and used Womble MPEGVCR to extract the MPEG stream (Womble simply extracts the stream -- no re-compression or alteration of the video takes place).

I then uploaded these three streams to the links below (which will be good for seven days).

test (2-pass VBR 3,800,000) .mpg

test (8,000,000 CBR).mpg

test (DVD Shrink 50% from 8,000,000 CBR).mpg

After having done all this, I think that I didn't choose the video very wisely. The background is so plain that the encoder has an easy time keeping up, even at low bitrates. I cannot tell much difference between the clips, even when played on my monitor. If I had time, I'd re-do the test with some video that has lots of detail. My favorite in the past was basketball. All those lines in the court as the camera pans around give the MPEG encoder fits.


Comments

No comments yet - be the first to write a comment...