better looking VCD's

[Anonymous65]

Oversampling is a known method of getting better looking VCD's (352x240), but it might be hard to capture at 704x480 mode on many PC's without dropping frames. Vertical reduction, 352x480 capture is one way of gaining detail, but it also introduces some blur when combining the interlaced fields. Another trick is to capture 704x240, which the simple BTxxx chip can do, and then reduce it to 352x240 during mpeg-1 encoding. This gives a sharper image and less edge noise comparred to regular 352x240 captures. I belive this is the process used for producing commercial VCD's.

[Ronin2]

skittelsen,
I read one of your previous posts and currently I'm capturing with a Pinnacle PCTV card at 352x576 (PAL)and 2:1 vertical cubic reduction using Virtual Dub. The resulting VCD's are great. I'll try this other method at 704x288 and see how it goes. Do I have to do any horizontal reduction during the capture, or is it taken care of when frameserving to TMPGenc.

Thanks for your advice,
Ronin2

[Anonymous65]

Hi, the reduction is done by Tmpeg, but it can also be done by Vdub (after capture). I have not tried to see if there is a difference in quality between letting Tmpeg or Vdub do the 2:1 horizontal reduction. Letting Tmpeg do the reduction does increase the time it takes to encode, especially if using noise reduction and/or sharpness filters. I'm not sure at what stage the 2:1 reduction takes place, before or after the filters. The nice part is that the 2:1 horizontal reduction also increases the persived vertical resolution.