MainConcept vs TMPGenc... wow

[johneboy]

So i wanted to test speed difference between TMPGenc and MainConcept..
I really couldnt believe the results.. But here they are..

oh yea, this is on a 500Mhz AMD K6-2 with 160MB RAM

Movie : Black Hawk Down
Runtime: 144 min

TMPGenc
using VBR 9100/4500/2500
Motion -> High

Total Time To Encode
125 Hours

Frames Per Second Est. 0.57

--------------------------


MainConcept 1.3.1
using VBR 9100/4500/2500
Motion -> 8 (7 is the middle option, so i would assume 7 is equal to TMPGenc's "Normal" Choice.

Total Time To Encode
18 Hours

Frames Per Second Est. 4.0


---------------------------

Do the math, that is 6.9 times faster

At that speed, if i could encode a 100min VCD in 8 hours on TMPGenc, i should be able to do the same in MainConcept at almost real time.. I will definately have to try that out.

[incognito]

A few random thoughts:

1) your bit rates seem to be different for the two different runs.

2) did you actually let your computer run 125 hours encoding a movie?
If I saw a 125 hour eta I'd cancel the job and try something else!

3) I'm not certain that the results of a 500mhz k6-2 will scale to
a 2GHZ P4.

[Gazorgan]

Ditto:

I have found when I start TMPG it says 3 hours...and when I'm done it was 10 (2-pass...etc etc) or the opposite...says 12 hours and takes 4 ( usually CBR...go figure?).

How about 5 minutes of video and a side-by-side comparrison?

[oddbod77]

your bit rates seem to be different for the two different runs.

Yeah I noticed that too, I think its a typo rather than a difference.

I'm not certain that the results of a 500mhz k6-2 will scale to
a 2GHZ P4.

Well no since the K6 doesn't support hyperthreading (btw I have no idea how much difference this makes in practice)

There is also a slight price difference between the two, with Mainconcept costing more than a good hardware encoder, I think I'll stick with Tsunami.

Obvious question but how does the quality compare?

[johneboy]

A few random thoughts:

1) your bit rates seem to be different for the two different runs.

The bit rates were a typo, they were the same

2) did you actually let your computer run 125 hours encoding a movie?
If I saw a 125 hour eta I'd cancel the job and try something else!

yea, trust me, i thought about it.. but im kinda used to that kinda ridiculous time frame

3) I'm not certain that the results of a 500mhz k6-2 will scale to
a 2GHZ P4.

Maybe not.. All i know is that im MORE THEN HAPPY with 18 hours for a DVD->DVD-R re-encode on a 500Mhz

[johneboy]

Ditto:

I have found when I start TMPG it says 3 hours...and when I'm done it was 10 (2-pass...etc etc) or the opposite...says 12 hours and takes 4 ( usually CBR...go figure?).

How about 5 minutes of video and a side-by-side comparrison?

125 hours was the final time.. When i first started it, it told me it would be 80.. but it lied.

As for the 5 Min video clip.. i'll hafta try that out

[dcsos]

Main Concept works much faster on my 2.0GHZ system as well.
Easily 5 times the speed of TMPG
What confuses me is the same job runs much faster on Sound Forge VEGAS 3.0
than
on ADOBE PREMIERE 6.5
both include the MainConcept MPEG engine...what gives?
Sorry I don't have actual figures, but
2 x faster seems possible!

Shuttle AB-30
1GIG SDRAM(133)

[BJ_M]

upgrade adobe 6.5 mpeg encoder and the speed should be about the same -- but always a little slower through premiere excpet if you have dual or hyperthreaded system (update to premiere also adds hyperthreading/multiproccessor support)

[sammie]

MainConcept is much faster. I don't know about 6.9x, but easily 4x on a 2.0 and 2.4 P4. What I care about most though is quality and reliability. With my set up both MainConcept and TMPEGenc run circles around CCE on stability. CCE is still the leader in quality of output, but in my opinion that gap is narrowing rapidly.

I've been fooling around with MainConcept a lot lately because of the speed and I'm starting to get 1 pass results that are very equivalent to the quality of multipass TMPEGenc, without using anysort of unusually high bitrates, and without claiming to be an expert in maximizing quality for MPEG encoders! Assuming they don't break anything when they implement multipass for Mainconcept (supposedly with v.1.4 this month), I'm betting that the quality will pass TMPEGenc if not give CCE a run for the title. I'd love to see if some of the same collective expertise that is focusing on TMPEGenc would try optimizing MainConcept what the results would be.

[johneboy]

OK, ran yet another test with TMPGenc and MainConcept

This test was converting the trailer for Hackers off the DVD to standard VCD.

Total Frames : 4974



TMPGenc
Total time: 1567 seconds (00:26:07)
Time spent per frame: 0.315 seconds
3.17 Frames Encoded per Second
File Size : 28195KB




MainConcept 1.3.1
Total time: 776 seconds (00:12:56)
Time spent per frame: 0.156 seconds
6.41 Frames Encoded per Second
File Size : 28169KB


As for quality.. here is some pictures to compare...


MainConcept 6:1 Zoom


TMPGenc 6:1 Zoom



here is the screenshots at 352x240
TMPGenc
MainConcept

[bbb]

From a distance, looking at your pics, I thought you were encoding the movie Mask (with Cher).

[RabidDog]

I like the idea but those single screen shots tell us nothing really... I mean there are no obvious distortions..I cant believe main concept CANT do multi pass... a bit bloody lacking!


BTW 125 hours .. cost of leccy = £18.90 perhaps?? or was this done at work or your pc is connected to the street lights

[THE RIPPER]

From a distance, looking at your pics, I thought you were encoding the movie Mask (with Cher).

yeah rocky dennis,

[johneboy]

I like the idea but those single screen shots tell us nothing really... I mean there are no obvious distortions..

I too see no difference, cept the MC pic looking a tab bit softer.. but it was just a comparison to see what kind of quality per frame you'd get.

BTW 125 hours .. cost of leccy = £18.90 perhaps?? or was this done at work or your pc is connected to the street lights

i assume leccy = electricity (im american, forgive me )
in anycase, my computer is never off anyways, so it wouldnt make a difference to me

[vitualis]

oh yea, this is on a 500Mhz AMD K6-2 with 160MB RAM

Remember, this means that your processor only has MMX instructions.

If these programs are run on a CPU with SSE and SSE2 as well (e.g., P3, Athlon, P4), the benchmarks may be completely different.

Regards.