I have recently done a VCD for a relative. But the quality of the mpeg-1 was noticingly inferior to the perious ones I made. I am wondering why and hope experts here can render a helping hand.

My Workflow- 1)Started from Video Capture DV capture using premiere with settings as follow; Mode Microsoft DV AVI Compressor Microsoft DV (PAL) Frame size 720 X 576 Frame rate 25fps Pixel Aspect ratio D1/DV PAL (1.067)

2)Breakup scene using Scenanlyzer (initially preview cannot be seen, remedied by installing Canopus playback DV codec)

3)Edit using premiere with following similar settings as Video Capture settings (as above)

4)FrameServe TMPEnc 12a using AVIsynth Export settings- Link to Avisynth 1.0Beta31 Compressor- (uncompressed) Frame Size- 352*288 Frame Rate- 25fps Pixel Aspect Ratio- D1/DV PAL (1.067) Field- No field

5)Under TMPEnc settings Motion search accuracy- Highest quality (very slow) Video Source type- Non-interlaced Field order- Odd field first Source Aspect Ratio- 4:3 625 line (PAL) Image positioning method- preserve aspect ratio 2) Noise reduction- spatial 40 Range 2 Temporal 40 High-quality mo checked Enable filter checked Basic color correction- Brightness 15 Gamma 12 Enable filter checked GOP structure standard Quantizer matrices Output YUV data as Basic YcbCr instead of CCIR601 checked Use floating point DCT checked

-- Foo Chek Wee (chekwee@hotmail.com), May 14, 2001

Answers

You chose video source type as non-interlaced. If you didn't have any special settings on your DV or D8 camcorder, the DV AVI files it produces, like any other conventional video source, is INTERLACED, which is what you should choose as video source type in TMPGenc. It could be possible you have a top-notch camcorder that CAN capture progressive; are you aware what your camcorder is capable of and what settings you used when you took your shoots?? The one other thing is to let TMPGenc read the field order correctly; this is done by double- clicking deinterlace, then choosing even-odd (field) then previewing the video. If there is judder and thin horizontal lines and the video doesn't proceed smoothly frame by frame you get out of there and choose the correct field order. Then you choose deinterlace>even field adaptive. To be fair I took this from www.flexion.org, which primarily talked about DVD ripping, but in the DVD2AVI method all aspects of deinterlacing in TMPGenc is equally helpful for any video situation. Three things were done here: 1) correct field order reading was determined and done; 2) a dominant field (even) is chosen); 3) a unique field for your MPEG-1 stream is created out of both fields through smart interpolation. Some of the other encoders, in contrast, merely use one field and discard the other, which is why they're fast, and TMPGenc is slow. TMPGenc is really powerful and excellent, but needs some effort to tweak and get acquainted with. Lastly, you frameserve in the original resolution 720x576. TMPGenc sees this resolution as the source, and does its own truncation/resizing down to 352x288. TMPGenc performs this much better than any competition, including Premiere.

-- Mehmet Tekdemir (turk690@yahoo.com), May 14, 2001.