Encoder comparison

[onisama]

I'm sure people have tried this before, but I was wondering myself what kind of results I could get from each different encoder. I tried CCE 2.5, Canopus Procoder 1.01.35, and TMPG 2.5. The AVS was the same for all three:

Telecide(guide=1,gthresh=50,chroma=true,threshold=30)
Decimate(cycle=5)
TemporalSmoother(2,1)
BicubicResize(480,480,0.0,0.6)

And the settings for each are (everything else is default):

CCE:
4 pass VBR, Image qulaity - 26, Noise reduction - 6

Procoder:
2 pass VBR, mastering quality

TMPG:
2 pass VBR, soften block noise - 40 & 40


The video length was 48 seconds.
-- CCE took 8 min, 57 sec
-- Procoder took 9 min, 34 sec
-- TMPG took 6 min, 11 sec (that doesn't seem right, it should be slower than the other two?)

If you extrapolate that to a 25 minute video, you get (if extrapolation/conversion is accurate in comparison to a real 25 minute video encoding session):
-- CCE: 4 hrs. 40 min
-- Procoder: 4 hrs. 59 min
-- TMPG: 3 hrs. 13 min

My past experience with tmpg has been different than this test, usually tmpg it took almost 1.5 - 2 times as long as cce.

Here is a portion from a screenshot of each:

cce - http://meisenheimerphoto.com/stuff/video%20samples/cce.jpg
procoder - http://meisenheimerphoto.com/stuff/video%20samples/procoder.jpg
tmpg - http://meisenheimerphoto.com/stuff/video%20samples/tmpg.jpg

You'll notice tmpg looks terrible (lots of block noise), but cce and procoder are comparable. I actually think cce looks a little better.

Maybe this will help someone to decide what encoder to use, I think I'll stick with cce.

[Kika]

48 Seconds of Video are too short for a true comparsion for Multipass-Encoding. That's because it's not enough room for spreading (allocate, share? What's right? Help me to brush up my english) the Bitrate. And TMPGEnc and ProCoder do need longer Parts, because they have a different Methode for Bitrade-Spreading.
The Speed of TMPGEnc and ProCoder depants on the Settings and of the Video. And please remember: TMPGEnc offers two different Modes of 2pass, the old one and the new one. The new one has the same Problem like 1pass Constant Quality - too small B-Frames. This can be solved by setting B-Picture Spoilage to 0 - 5.

Soften Block Noise is often used the wrong way: It does NOT prevent Blockyness! It's a Macroblock Based denoiser. If you set it too high, you will eventually PRODUCE Blockyness.

TMPGEnc general speed: It's from 1.5 up to 6 times slower than CCE and a little bit faster than ProCoder. In 2pass Mode, the second pass is a lot faster than the first one, in some cases faster than CCE (on smaler Formats like Half-D1).

As i can see, you tried to encode an Anime or something like that. A Small tip: I wrote a special Matrix for Animes.