How good is MainConcept MPEG-2 codec?

chpol wrote on 8/29/2002, 7:19 AM

I've used Vegas Video's MPEG-2 encoder to do the encoding for some home movie DVDs. I used DVD Workshop to do the DVD authoring, but it presumably had no affect on the video quality because I made sure it didn't do any re-encoding.

To get the best possible quality, I encoded at a variable video bit rate with a maximum of 9.8 and and average of about 8.5 to 9.2 Mbps. I'd describe the video quality of the resulting DVDs to be fair, but not particularly good, but I am admittedly fussy. I often see slight to moderate MPEG artifacts, and some parts of the video flicker noticably.

My video is admittedly quite hard to compress, since it is almost all taken with a hand-held mini-DV camera, and has a lot of pans, zooms, and following shots. There are very few scenes where the camera remains static for any length of time.

I'm wondering how the MainConcept encoder compares to other encoders that are available at a realistic price (a few hundred dollars). I tried the latest evaluation version of TMPGEnc Plus, but it didn't look obviously better, even when I set all the quality setting to maximum. It was also many times slower.

I'd be interested to here other opinions.

Comments

kkolbo wrote on 8/29/2002, 8:03 AM

You probably did not get any advantage to such a high bitrate encode. I have actually have had artifact problems from bitrates that high. That is not caused by the encoder, but by the player that just could not perform at that high a rate, even though 9.8 is quote legal. Leave the max at 8,000,000. Average it below that or do a CBR at that rate.

Also remember, the better the video input, the better the encode. You camera compressed it the first time to DV and it may have left some artifacts as well. I was not surprised to hear that you had similar results from the other encodes.

The test of an encode for me is with a 4-4.5 mb average. Most do just fine at 6+. I tested the SF/MC encode against hardware encodes. On encodes where the clip was encoded in one shot, not picked apart scene by scene like a professional encode artist might do, SF/MC held its own and often produced a visually superior encode. I switched from using stand alone encode products and hardware encoding to using SF/MC.

I believe, although I do not know, that SF/MC is probably better that other packages that use MC. There was a lot of work by SF to tweak the pass off of the data stream between SF and the MC encoder. Vegas 3.0 vs. 3.0c has a noticable difference because of that work.

That is my opinion. If you can not get a good encode from the SF/MC then you probably will not get a good one without a specialty artist and far more money than I have.

K