MPEG-2 Encoder Shoot-out



Interested to find out the best MPEG-2 encoding solution for DVD?
As a community we can find out together…
Check out this initiative:


That sounds very interesting. Will the panel be judging the encoded video based on what looks best to them or on how closely it matches the original?


What looks best to them. The original material is the same for all.



what encoders do you have in mind?

cce, canopus, tmpgenc…but also freeenc and quenc?


I don’t think it’s really necessary to include both FreeEnc and QuEnc, I’m led to believe that FreeEnc is basically just QuEnc with adjustable parameters, the philosophy behind QuEnc is to cut out the less important options to prevent people new to encoding from making stupid changes, getting a bad encode and then blaming it on the encoder. As a result I think QuEnc is going to be more widely used and supported than FreeEnc.

Another encoder that should definitely be included is HC , which you can read about here and here. Its quality rivals CCE and possibly betters it at low bitrates and it’s freeware.


yeah i know the doom9 threads…i visit it more then cdfreaks lately (sorry).
Quenc and freeenc are tuned differently…but as we want to know which one is best…i guess we want to try both.
HC for sure is great…but honestly …i think it will only make a diference in low bitrates movies.

Besides those mentioned…we also got Honestech and mainconcept…even ffmpeg (think quenc and freeenc are based on it)

the prob with video quality is… it is a visual thing. People never agree on it.
You only can judge about file size and some other stuff.
Also quality can be influenced by different matrixes.


The encoders used is the choice of the participants, so we’ll see in the submissions.
And yes, it is a “visual thing” therefor we are not planning to judge on mathematical analysis, but purely visual.


I can see a couple flaws in this plan. Firstly you’re just relying on the word of the participant that they have encoded the video the way they claim, a few rules might be a good idea too, i.e. no filtering or custom matrixes and the test would be more complete if you took account of the time spent encoding but of course that would require a test machine that would remain the same for each encode, which is impossible with the present setup. I not trying to sound negative but I think if you made a few changes this could be a very valuable test.