DVD-CAM MPEG-2 file to xvid

[garce]

I need help with xvid encoding setting
I have a Hitachi DZ-MV580 video camera that records directly on 8cm DVD-RAM discs. The files are MPEG-2 (NTSC 29.97 fps) with extension VRO. The camera has several output qualities • Approx. 20 minutes in Extra Fine Quality Mode (Single Side 704x480, 3-10Mbps VBR)
• Approx. 30 minutes in Fine Quality Mode (Single Side 704x480, 6Mbps CBR)
• Approx. 60 minutes in Standard Quality Mode (Single Side 352x480, 3Mbps CBR
The approx. duration in minutes shown is for one side DVD-RAM - 1.4G

I found that I can process the files with AutoGK/xvid.
I would like to know what type of bitrate/filesize should I use for each case. I usually hear that for movies 800+kbps is decent and there are some considerations depending on resolution bits/pixel. I was using an xvid calculator and was confussed because the file size will only be reduced by 1/2 (based on keeping same resolution)
My idea is to have some decent results, that I have a hard time telling the difference on a 27inches TV+Phillips DVP642 player between original and xvid.
Would somebody that have done the same thing have some recommendation for this particular case?

[Tommy B.]

Best solution would be leaving the material in MPEG2 and burning it on DVD.

If you want to recompress into XviD, you should compress the video in single pass mode, using quant 2 or 3.

The reason for this is the unpredictable compressibility of the video. You might want to encode to 1CD using two pass, but this could mess up the picture in case it can only be comressed using high quants.

Also you can not compare "movie" with "home video". While movies usually consist of scenes with blured background (depth of field) and little motion (using a tripod for example), they do not consume as much bitrate as a home video, which is overall very sharp (blurred areas are easier to compress than sharp ones), shaky (a lot of motion) and has a good amount of noise.

If you still want to stick to recompressing use the fixed quant. It will maintain good quality but you won't have any control over the final file size. Also denoise your material (with FFT3DGPU for example) with avisynth to make it more compressible.

[chilledoutuk]

Ok the quality of your cameras ccd is pretty poor which means that the video will have quite a large amount of noise plus the fact that there are going to be artefacts where the on the fly mpeg-2 encoding doesnt have enouggh bitrate in fast motion scenes.

All these things are going to make your video harder to compress than a reasonable quality dvd movie.

If you intend to manipulate your video then these camcorders are nowhere near as good as a standard minidv.

for a 60 minute minidv tape you have 13gb of video data.

with one of those little 8cm dvds you have 1.4gb not even using the best video compression in the world can compensate for that difference let alone a rubbish cbr mpeg2 encode.

This is exactly why i warned my sister off of these dvd camcorders they are made for convenience but will always be of lower quality due to the crap dataspace for recording the video to.

[garce]

Thanks for your comments. I am fairly new to this. Could you help me with this please:
1) How would I denoise the material with FFT3DGPU? I have never done anything like this or use this program
2) After that what will be a fair xvid videorate bitrate for autogk?

[SeeMoreDigital]

I am not all that familiar with MPEG-2 VRO files. Can you provide us with a short (10MB or so) sample?

One of our forum moderators, Richard Berg (http://forum.doom9.org/member.php?u=1704), has provided some temporary ftp space (ftp://tempclips%40avisynth.org:QfJY(86m@avisynth.org/) for you to up-load your sample if that helps!


Cheers