First lets set the stage :

Typically I encode with a minimum bit rate of 1.8, average 6.0 and max out at 8.0 Mbps which gives me around 1.5 hours on a DVD. If I want 2 hours I average at 5.0 and the percieved quality is usually indistinguishable from the original.

Now My kids recently got a SAMSUNG Hi-8 Camarder whch delivers a fairly good, but nosiy image. Personally I use a SONY Digital 8 and the image it delivers impresses me a great deal.

For sometime I have been encoding my Digital 8 DV material with the Cinemacraft 2.66 encoder and have been a happy camper. however I tried the same encoder with the Hi 8 material and was somewhat distressed by the amount of mpeg artifacts I was getting - blocks etc. - artifacts I was not getting with the DV material. Now it was time to test!

The encoders:

Cinemacraft 2.66, Latest Mainconcept version, Latest TMPEG version as of Jan 2005. Bitrate settings as per the above, number of passes CCE - 3, Mainconcept - 2 and TMPEG - 1. I tried the encodings with and witthout filters and used Virtualdub to do a frame by frame comparison, I, P and B, with the original. I attempted to keep the GOP structures the same and closed all GOPS to make sure editors would have no drama with the video. Finally, I used a datavideo analog to DV converter to grab the Hi8 material.

The results :

In a word Mainconcept won. It produced video that was only a hair different from the original albiet with a minor colour shift, with the noise filter set at 6 most of the chroma noise in the digital 8 image was removed and detail seemed to be unaffected. In fact I would archive material with this encoder and have little qualms as I really could not tell the mpeg file from the original AVI file. The only cavet is that in low light this encoder seems to produce a fair amount of mosquito noise, although setting the noise filter on reduces this issue.

CCE 2.66 produced video that was darker and more smoothed with a loss of fine detail. It certainly reduced mosquito nose in low light and macro blocks but that came at a cost.

TMPEG was excellent, quite similar to Mainconcept, but quite prone to produce blocks - particularly on moving water images and for interlaced home material it would be my last choice. It was also abominally slow!

Conclusion:

All 3 encoders are good and on a 68CM Sony Wega produce an excellent image. However, Mainconcept produces the image most faithful to the orginal and I will use it in all future encoding efforts. I think I have been misguided in using CCE 2.66 and assuming it is the best encoder out there but I cannot comment on newer versions which may more closely match Mainconcept in output.

A final note:

Editing Mpeg 2 DV quality on Ulead Video studio does not seem to be a major problem and the reencode of the joined clips around an effect, if performed at around 8 Mbps, is not noticable. I think archiving to Mpeg 2 and editing is an achievable goal based on practical experience.

I will close by saying the Sony Wega is an excellent TV for finding problems with mpeg files as it seems to show mosquito noise and macro-blocking exceptionally well, much better than the other 68 CM TV I have in the house which masks artifacts rather well. Incidentally I live in PAL land, now any thoughts or comments particularly as regards the colour shifting that MPEG encoders tend to produce!?