I have been making VCDs for quite some time. I normally capture at 352x24)x29.97 frames/sec when I am capturing to avoid the need for deinterlacing and resizing when capturing video of long durations. I also was having problems with my iomega buz (dropped frames) when capturing full frame video and now that I changed to IWILL motherboard I can capture 720 x 480 at 6 meg/sec. This leads to me my question about deinterlacing both fields before MPEG encoding. I have done extensive reading in the net about it and still remained confused. I would like to hear some advice about the following:

1- using smart deinterlace with virtdub to combine fields before sending AVI to the mPEG encoder

2- MSP pro or Premiere can also deinterlace but it appears that it only drops one field and does not combine them so capturing at 240 would do the same. This would lead to movement artifacts in rapidly moving scenes. Is this true and if is is, one could also use the panasonic MPEG encoder to only encode one field.

3- I have problems with the panasonic encorder with color washout making the output file loosing some of its color vividness. I have now tried both BBMPEG as well as the Japanese freeware encoder with the english patch (TMGPencorder). Both of these encoders have excellent color rendition and little blockiness. The freeware update of AVIMPG now call AVIVCD is quick but has significant blockiness. BBMPEG as well as TMCPencoder also have setting to deinterlace. Does anyone know how these work - do they combined the fields or just drop one field. I have tested them with and without the delinterlace setting and see little changes on the computer but I haven't burn them to a CDR and played them on my standalone DVD player

4- Is it really worth the effort to deinterlace and should I just capture at 352 x240 only.

Thanks

-- klebsiella (klebsiella@visto.com), May 10, 2000

Answers

Hmm... depends on the source. I've foung that 24p source -> 29.97 broadcast gets muh better results just doing a bicubis resize. If you are comming from true 29.97 source I'de use VDub with deinterlace ( new rathersmart deinterlace ) and the resize.

As for the encoders... thankyou. I have too noticed that bbMPEG produces much smmother color rendidtion. Tmpgenc looks good, but the output still dosn't seem to do better the bbMPEG ( with enhanced motion settings ).

As for capture size... YES, if you want quality output capture at the larger frame size. I always cap at 720x480. I find that although slower to down convert, the added bits help me do do things like noise reduction, level, hue, and deinterlacing without damaging the picture quality. It's amazing what a few filters can do to make broadcast video look damm good.

One thing to check is to make sure that your card it ACTUALLY capturing both fields. This should be very easy to spot, if it is interlaced you are, if not don't bother with caps over 240.

Cap @ 720x480 Vdub filters Dynamic Noise Reduction (6-10) * careful too high will leave ghosting *rather smart deinterlace ( optional ) resize ( persise bicubic ) 352x240 or 480x480(SVCD) Dynamic Noise Reduction (2-4) Level Hue/saturation

Happy encoding

-- eric (eric@snowmoon.com), May 10, 2000.