|
miniDV compressibility
Background
I am trying to encode my NTSC DV videos to xvid. I render to huffYUV (YUY2 mode) after editing as an intermediate step. My Avisynth script merely runs tomsmocomp to deinterlace the file. If the shoot took place indoors I use some form of noise reduction. I have a separate MPEG-2 encode for T.V. viewing, so I want the MPEG-4 encode for P.C. viewing to be deinterlaced.
The camcorder is a Panasonic PV-DV953. The xvid encoding options are: 2Mbps with two passes, qpel, gmc, default B-frame and quantizer options (but Soulhunters V5 matrix), VHS6, MSP6, VHQ1, chroma motion, turbo;-), chroma optimizer ... basically everything you need for a decent encode. The source is full frame.
Problem
Why do I have to go to bitrates as high as 2Mbps before I can get something barely acceptable? DVDs look great at half that bitrate.
My hypothesis is that camera shake is throwing off motion estimation. My work is hand held -- consider this is a constraint -- and I am using optical image stabilization as it is. Moreover, I suspect the camcorder is applying excess sharpening.
P.S. Is it better use huffYUV in RGB mode in this case?
Last edited by adoniscik; 15th June 2004 at 04:28.
|