Extreme quality : CCE or TMPGenc ?

[jamepa]

I know that some pro's have been posting about this few minutes ago, but I want your opinion about this .

My video source is DVD2AVI + VFAPI, I load a *.avs in CCE and encodes it.
The output file 2 hours later, is MPEG1 with a bitrate avg 2400, min 2000 and max 3000 ), 3 passes, about 38 minutes long, and the resolution is 352 x 288.

I found that the .mpv is about 400 MB. and found with surprise that the same settings, with the same source, encoding with MPEG2 option in CCE is 700Mb

I found that the mpv dont have a good quality, it have "pixelation", specially in scenes with a lot of motion.

The same avi, same settings on TMPGENC, 4,5 hours later have better video.

This say me that CCE is very fast , but TMPGenc have better quality.

What do yo think ? thanks.

[Chris X]

I have not used CCE so I cannot tell you which is better. But I have found TMPGenc to produce very high quality results on my VCD's.

I hear that CCE is better for MPEG2 Encoding, then TMPGenc. But TMPGEnc better for MPEG1 encoding, over CCE.

Hope this helps

[wkc]

I've also read somewhere posting CCE and TMPEGenc quality side by side and TMPEGenc has better quality. I think it was on VCD and SVCD format.

Ever since I read about that, I've been using TMPEGenc all the time. For quick jobs, I just used the CBR and "Fast" setting. When I need quality, I uses VBR 2 pass and "High" setting. I also remember that the person who wrote the guide mentioned that he couldn't tell the diff between "High" and the "Highest" setting. So, I've also been just using the "High" setting. The result is amazingly good.

[adam]

If the filesizes were different than you did something wrong. The encoder has nothing to do with the filesize, the avg bitrate is what dictates size.

As far as quality, Custom Technology themselves even admit that CCE's primary focus is on mpeg2 encoding and that, as such, their mpeg1 encoding quality is sub par. Perhaps this has changed in the newest release but generally speaking CCE is not very good at mpeg1 encoding. For mpeg2 encoding, however, it is regarded by many as the highest quality encoder on the planet, and I tend to agree.

[Martyn1980]

Im testing Version 2.64 of CCE right now, and so far, there does seem an improvement on Mpeg1 encoding, but it doesn't appear to support multipass VBR mpeg1 anymore though.

The mpeg2 encoding has also improved slightly, though not by a massive amount, but the matrices featur, which has 5 presets, and the ability to save your own, is interesting to experiment with, and seem to improve low bitrate encodes.

Tmpeg is still best for VCD as far as i can see, and CCE for SVCD, if you look at the pirate movie scene right now, then you can see they share this opinion, with most VCD releases being encoded with Tmpeg and most SVCD releases being 3-5pass VBR.

Most people serious about encoding prefer CCE because the quality is better, not because its quicker, even Tmpeg is quicker than a 5 pass VBR on CCE, because even on a fast machine its still going to take 8-10 hours for an average length film.

I hope this isnt going to move into another encoder war, but i had to have my say on the matter anyway.

[jamepa]

Thanks for your posts.
I see that a lot of people agree that CCE is better for mpeg2, so i´m going to make some test to compare tmp and cce only with mpeg2 video.

Some more questions, oriented only to mpeg 2 :

Really is an advantage make 5 passes in CCE 2.5 ?. I obtain good video with tmpgenc and 2 vbr pass. Thinking that in cce 2.6x we haven't this number of passes I see that Custom Technology thinks that it´s not neccesary.

Setting high bitrates (avg 2400, min 2000, max 3000) in cce or tmpgenc, always give me problems at multiplex time. I must set higher mux rates on bbmpeg to obtain a mpg without desyn....

To finifh I don't war a encoder war, like Martyn, Im only looking for the best metod to do my vcd, svcd, cvd.

Thanks a lot.

Just to add another encoder into the pot, I´ve recently been trying out Pro-coder over the last week, and its giving some very impressive results, especially with DV sources to mpeg 2. Today i started frameserving a DVD rip, will be interesting to see the result when finished.

cheers

[adam]

jamepa: According to CCE, no there is no advantage to even doing 5 passes. The CCE manual states that the maximum quality benefit or additional passes is reached between 3 and 4 passes, and I agree. Doing more than 2 passes can definitely increase quality, however. Try doing a 2 pass encode, 3 pass, and 4 pass encode in CCE and see if you can tell any difference. If not than just stick to 2 passes. CCE is so ridiculously fast that you can do 5 passes in the time it takes TMPGenc to do 2, so why not just do the extra passes? In any case be aware that the number of passes setting in CCE does not include the creation of the .vaf file, which constitutes as the first pass. So if you set CCE to do 2 passes its really doing 3, 3 pass is really 4 etc.. CCE 2.6x still supports up to 9 passes just like the previous versions. Are you maybe using a crippled demo version?

Well those settings you listed arent compliant. You can't multiplex a non-svcd comliant stream as a svcd, you'll get underflows which cause desync. Either keep your encodes within the specs or you'll have to change your mux rate, no way around it.

[jamepa]

Thanks to all for ur posts, im going to apply this knowledge.

thanks again

[banjazzer]

I see that a lot of people agree that CCE is better for mpeg2, so i´m going to make some test to compare tmp and cce only with mpeg2 video.

And quite a few people don't agree with this. 8) If you want to compare CCE with TMPG I suggest you do some tests using TMPG's CQ mode with min bitrate=300 and max bitrate=2520 (enable padding ticked) and quality setting between 75 and 85. If you find the results inferior to CCE (given approximate final filesizes) I will be very surprised.

[jamepa]

Opps,. I can make the test that you mentioned, but I prefer your opinion, which is welcome. Why do u think that tmpgenc is better for mpeg2 ?

[Martyn1980]

The reason i do 5 passes is, i usually do my encoding when i go to bed, one half of the film at a time, so with all films its done before i get up, there might be no difference between 2 and 5 but since i will be sleeping about 6-7 hours i might as well do the best i can.

Banjazzer, if Tmpeg is so much better for Mpeg2, then how come most VCD release groups never use it, for anything other than Standard VCD, are you saying you could do a better encode of Star Wars SVCD, using Tmpeg's CQ.

[banjazzer]

Why do u think that tmpgenc is better for mpeg2 ?

I did not say it was better for mpeg2. I said I would be surprised if TMPG using CQ mode was inferior to CCE. Frequently people compare CCE with TMPG's 2-pass VBR, which is not necessarily its best mode. 8)

I can make the test that you mentioned,

If you do, set Motion search precision to *high quality*. And notch up the quality setting until you get roughly equivalent filesizes.

Banjazzer, if Tmpeg is so much better for Mpeg2, then how come most VCD release groups never use it, for anything other than Standard VCD, are you saying you could do a better encode of Star Wars SVCD, using Tmpeg's CQ

I did not say TMPG was better for mpeg2. But I don't agree that TMPG, used properly using its best mode, will give results inferior to CCE. There are reasons why people choose CCE, apart from the misguided notion that because it is ridiculously priced it must be the best. Using TMPG's 2-pass VBR mode is very slow. CQ is much faster, but you need to do some sample encodes and use some basic maths to work out the final filesize. Even then, you can get it wrong. CCE does not have the limitations of slowness and unpredictability. On the other hand, it has it's own limitations of user-unfriendliness, bugginess and lack of features.

Given a DVD source for Star Wars I am confident I could produce something equally as good as the current SVCD, using TMPG's CQ mode. However, I don't have one at present, so it is a moot point.