Comparison of TMPGEnc 2.56 vs. CCE 2.64 vs. Main Concept

[BJ_M]

its pretty close between Main Concept and CCE

wow thanks BJ_M fast reply!

i m more interested in vcd-mpeg1 only, not mpeg2.

if i wwere to hear ur view, tmpgenc or mainconcept gives better vcd output? ( i m excluding CCE here as mentioned in ur posting that vcd quality by CCE sucks)

main concept - except that certain sources and some other problems people come across cause Main concept to not work .. see others posting here on problems.

[LeeBear]

BJ_M: I understand that encoding a long full lenght movie would be time consuming and I'll take your 30 second clip as a valid comparison test. I also understand why you choose to encode using VBR instead of CBR since I myself also use that mode to encode my movies. I do not even doubt your words when you say the samples interpolate what you've seen on longer real life encodes you've done. But isn't the purpose of putting up samples is so people can judge themselves what they consider to be the best quality? And when it comes to MPEG encoding quality and bitrate are directly related, a 16% increase in bitrate should give a noticeably better result. When doing a test of any kind one should keep all controllable variables the same so one can exclude that variable as affecting the result. Surely a min, average, and max bitrate is controllable in all the encoders.

I know you've taking your time to do all these test, and I and I'm sure others appreciate it, but tests, especially comparison tests should be as fair as possible. Most people are going to read your post, download the samples and conclude "yeah main concept is the best encoder" when in fact the only thing I (or anyone) can conclude from the samples is "with 16% more bitrate the main concept encoder gave better results then CCE or TMPG".

-LeeBear

[BJ_M]

well i pointed out that it only gave better results on mpeg1 VCD in fact..

though i agree with you in principle. if you would like a CBR comparison i dont mind .. not sure where you get the 16% increase though, in fact i though i did post CBR results as well as VBR ..
a whole lenth movie or a longer segment is not any more time consuming to me - but bandwidth is an issue ... 15000 downloads of 400-500 meg clips is going to cost some some coin at my end..

i've been up all night working on a project - so im not going to think about it right now anyway..

BJ-M

12. Panasonic SVCD (CBR) default settings -- i cant get it to do VBR , may be because in demo mode ?

I think I read in the help,or maybe on there site , 2pass VBR is only available for interlaced sources.-no idea why?

I tried some samples using shaft(DVD is interlaced) as the source, 2-pass VBR worked fine, only the colours were all washed out in comparision to Tmpgenc and damm slow.

[choonsin]

Hi BJ_M, i just read thru some of the postings and found that
there is a audio sync issue lying in mainconcept encoder.

how does this affect the normal dvd to vcd conversion ?

also, will this encoder take .d2v created by DVD2AVI?

[BJ_M]

BJ-M

12. Panasonic SVCD (CBR) default settings -- i cant get it to do VBR , may be because in demo mode ?

I think I read in the help,or maybe on there site , 2pass VBR is only available for interlaced sources.-no idea why?

I tried some samples using shaft(DVD is interlaced) as the source, 2-pass VBR worked fine, only the colours were all washed out in comparision to Tmpgenc and damm slow.

i also used a interlaced source and no mater what VBR was blocked to me in any profile ... weirdness is all i can say .

demo mode only works 10 times btw

[WarpEnterprises]

The stupid settings which are not available (2pass only interlaced,...) can be used by saving as project and then editing this textfile, some settings can only be used with a video system file (m2v).
10x testing is VERY few (HKEY_LOCAL_MACHINE\Software\PDNCO).

I tested it at low CBR bitrates, progressive:

REALLY BAD

(slow, mosquito noise, no scene change detection, bad letterbox borders) - worse than TMPG and ProCoder (my favourite)

Also occasional crashes, mocks about too low bitrates,...

[The village idiot]

Rumor has it that Pure Motion's Edit Studio/mpegXS is using the Mainconcept encoder SDK. I must admit, it is fast and looks pretty good. Just wish it would do 352x480. Limited to 720x480 for DVD.

[choonsin]

i am still confused.

issit WORTH to invest in buying a copy of this mainconcept [USD149!!!]?

i m too used to using DVD2AVI which produces .d2v file to be processed in tmpgenc 2.56 which is until now my great fav. but the speed of tmpgenc is a turn-off.

i tried out the demo version of mainconcept but could not get it accept the d2v file( did i miss out something important?).

i need more outputs fr you experienced guys out there.

[ajp]

choonsin,

wrap the .d2v in an AVISynth .avs file and it will accept it. This way you will have acces to additional filters etc if you need it, well worth the effort (check www.doom9.net). Never tried .d2v directly....

I actually find the Panasonic encoder very good, especially for less than perfect sources like TV anyway. The filters and noise reduction are very good. It IS slow as hell though but not as slow as it might seem in the beginning...
Converting DV doesn't produce any blocks or anything for me, very nice quality?

[Lamont Cranston]

If you would like to get more information on the setup of MainConCept's stand-alone encoder you can look here:

http://files1.sonicfoundry.com/manuals/mpeg_overview.pdf

The info is for the plug-in but the mechanics is the same.

Here is another tip to get around the MainConcept 4G files size limit:

Using stand-alone and only been playing with the Demo so far. Was going to splash out on it but also came across the same problems you did regarding >4Gbyte files on 1.1. You can fix this by going back to the DirectShow driver buy using Regedit:

Need to go to [HKEY_CURRENT_USER\Software\MainConcept\MCMPEGEnc]

and change the value under AVI Import to 0. The program will then load up using the DirectShow drivers as it did in version 1.0.

[ajp]

That did not work for me, there's still something wrong with the large AVI file support. At least for type 2 DV. The encoding stops after a while and I get som strange message. I've been told this will be fixed in the next version coming real soon...

I normally capture to large type-2 DV (AVI 2.0) so the next version will make things even simpler for me.

Right now I have these choices:

1. Wrap the file in an AVISynth .avs file and frameserve. Speed isn't affected, at least not for me. Use AVISource for type-2 DV and Directshowsource for type-1 DV.

2. Capturing to multiple 2GB type-2 files or 4GB type-1 files and batch encoding them to one mpeg file..

[Lamont Cranston]

That did not work for me, there's still something wrong with the large AVI file support. At least for type 2 DV. The encoding stops after a while and I get som strange message. I've been told this will be fixed in the next version coming real soon...

My large avi files (23 G) are NTSC DV type 2 openDML compressed with MainConcept DV codec and the MainConcept encoder with the registry patch works fine until MainConcept corrects the problem.

I checked again and encoded a NTSC DVD mpeg 2 file using the default settings (with the aformentioned registry patch) from a NTSC AVI DV Type 2 OpenDML 7.77G file withtout a problem.

There are two MainConcept encoder problems reported that puts my buy on hold.

The appearance of frame rate slow down reported in this thread and
http://www.wwug.com/forums/uleadmediastudio/020723181220.htm. Note by the date this problem has been around for a while. The MainConcept Encoder evaluations here are an interesting read.

Audio/video out of sync is reported http://vcdhelp.com/forum/viewtopic.php?t=114179&highlight=
Personally, I have encoded a 1 hour video and found no audio/video sync problems.

[ajp]

I changed the registry setting and now a 3.5GB type-2 file encoded without any problems (imported as directshow source). Weird, didn't work before or maybe I did something wrong. I will run some more tests on larger type-1 and type-2 with the changed registry settings....

Who knows, maybe the next version will be out this week. It sounded like it was coming real soon when I talked to them...

[LeeBear]

BJ_M: I was only comparing the SVCD templates like I said in my previous message. And since that time I've noticed you have edited the original post so the results are different now. Originally you said Main Concept quality was better then CCE by 3 points, followed by TMPG. Now you say TMPG is the best followed by Main Concept and then CCE. What you conclude isn't that important because I'm sure the people will make up there own mind after looking at the samples. Some people may not like the soft look of TMPG, especially noticeable in the white text on black and rate it differently. But the results can still not be compare fairly since the three SVCD samples have different sizes. 3.65 MB for CCE, 3.84 MB for TMPG, and 4.23 MB for Main Concept. These file sizes are what Windows reports on the sample files, that's how I got the 16% more bitrate, 4.23 MB is about 16% larger then the 3.65 MB of CCE.

-LeeBear

[Lamont Cranston]

BJ_M: What product and model Snell and Wilcox analyzer are you using? I would like to look at the specs. Thanks.

[BJ_M]

BJ_M: I was only comparing the SVCD templates like I said in my previous message. And since that time I've noticed you have edited the original post so the results are different now. Originally you said Main Concept quality was better then CCE by 3 points, followed by TMPG. Now you say TMPG is the best followed by Main Concept and then CCE. What you conclude isn't that important because I'm sure the people will make up there own mind after looking at the samples. Some people may not like the soft look of TMPG, especially noticeable in the white text on black and rate it differently. But the results can still not be compare fairly since the three SVCD samples have different sizes. 3.65 MB for CCE, 3.84 MB for TMPG, and 4.23 MB for Main Concept. These file sizes are what Windows reports on the sample files, that's how I got the 16% more bitrate, 4.23 MB is about 16% larger then the 3.65 MB of CCE.

-LeeBear

? i didnt edit the post re: outcome of the results ... really , honest!

I suggest simply that people judge for themselves.

yes - they are different sizes -- i am glad to know what makes a smaller size at the same bit rate ... a lot of that is in the GOP structure - more B frames equals smaller files .. but if the quality is consistent with a lot more B frames - that is a more effecent encoding (spelling = me bad).

I have been playing with CCE 2.66 now and it gives different results than 2.64

[BJ_M]

BJ_M: What product and model Snell and Wilcox analyzer are you using? I would like to look at the specs. Thanks.

Mosalina - > you can get a demo copy here ( i was using a demo): http://www.snellwilcox.com/

direct link because its hard to find is here :

http://www.snellwilcox.com/products/mosalina/content/welcome.html

AND Links here : http://www.snellwilcox.com/products/mosalina/sidebar.html

[adam]

yes - they are different sizes -- i am glad to know what makes a smaller size at the same bit rate ... a lot of that is in the GOP structure - more B frames equals smaller files .. but if the quality is consistent with a lot more B frames - that is a more effecent encoding (spelling = me bad).

The Gop structure has no effect on filesize. The number of I, P, and B frames only determines how the bitrate is distributed per frame. It is the amount of bitrate every second that determines filesize and this is something that can always be controlled in multipass encoding or of course CBR, even from one encoder to the next.

The average bitrate is the sole determinant of filesize.

[LeeBear]

BJ_M: I totally agree with you that people should judge for themselves but if the samples are not encoded fairly how can one determine if it's the 'higher bitrate' or the actual 'encoder effceciency' that's giving the better result?

Whether the encoder decides to use more B frames in the GOP structure has nothing to do with the bitrate since you're using VBR encodes. The encoders should still obey the min, avg, max settings you put in. And all the MPEG samples should be fairly close to the same size. So the only explaination for the different file size is 1) the encoders didn't obey the min, avg, max bitrate setting or 2) you set the min, avg, max bitrate setting different for each encoder. If the latter is the case then it's clearly not a fair comparison. Secondly it's also misleading because on the sample links you put the size of the Main Concept SVCD encode as 3.8 MB when it's closer to 4.3 MB in reality. If tests aren't done fairly I can put up VCD's sample encoded with CCE that will look better then one's using TMPG... and no one in there right mind would say that CCE produces better MPEG1 encodes then TMPG.

-LeeBear

[BJ_M]

yes - they are different sizes -- i am glad to know what makes a smaller size at the same bit rate ... a lot of that is in the GOP structure - more B frames equals smaller files .. but if the quality is consistent with a lot more B frames - that is a more effecent encoding (spelling = me bad).

The Gop structure has no effect on filesize. The number of I, P, and B frames only determines how the bitrate is distributed per frame. It is the amount of bitrate every second that determines filesize and this is something that can always be controlled in multipass encoding or of course CBR, even from one encoder to the next.

The average bitrate is the sole determinant of filesize.

yes you are right - i am think of all I frame editing where there is no gop and it larger than when its compressed down .. other that that -- yes , its data rate which is determed per GOP ...

sorry for the wrong nfo .

[BJ_M]

ok leebear -- i dont disagree with you, you are correct , the anomaly is that even at the same data rate - different products produce file sizes that are different -- i just used the encoders templates and i as i mentioned - the templates that come with them do a fair good job .. so it boils down to a comparison between standard settings or a comparison between same data rate - both have a valid place ...

here next are same date rate, same gop , same matrix even

[adam]

I've noticed that later versions of TMPGenc do not respect your avg bitrate setting completely and often allocate more bitrate than you tell it, resulting in a larger file. My guess is that MainConcept is doing this as well, and even moreso which is why you have three different filesizes even though the same avg was set for all. As for CCE, regardless of what you think of the quality it is utterly and completely predictable and accurate. If you specify a certain avg you can be sure that is what you get.

Assuming the encoders are doing their job correctly, in this regard, you should get nearly the exact same file size at the same average every time. Personally to make a fair comparison I think you would need to adjust your settings to create files of roughly equal size otherwise you are giving some encoders an edge as a result of their own inaccuracies.

A MB or two difference may seem inconsequential but you can imagine how much bigger the discrepancy could be if using a 2hr encode.

[BJ_M]

yes -- in fact cce is very reliable .. if you look at the pics in the first post i put there from bit rate viewer -- panasonic is none to good and tmpgenc is only slightly better (tmpgenc always spikes at the end and some times in other locations - often substancially over your max - causing skips and such on dvd playback) -- cce is straight as a arrow - notice all those pics are from CBR encodes at the same data rates ...

A question about Main concept

You mention in the first post

8. Main Concept SDVD mpeg2 - default sdvd setting , VBR , no changes to any settings. stand alone encoder , on its own MC changes its matrix for svcd. Fastest encoder - beating CCE by almost twice the speed but only doing a 2 pass vbr. smaller file size. same quality as tmpgenc - slightly better rating for quality than cce.

Does mainconcept really do 2-pass VBR, I could not find any referance to this, maybe I´m blind or something?

Or is 2-pass not available in the full version?

[ajp]

Mainconcept does NOT use 2-pass AFAIK. I think it can best be described as Constant Quality mode with the ability to set max/avg/min bitrate, which in my book gives the best of both worlds. Multipass sometimes sticks to close to the avg bitrate and doesn't always give the best quality. Comparing CCE's 3-pass with Mainconcept maybe isn't that fair, especially when comparing the ability to stick to the average.

I hear version 1.2 is coming out soon, will interesting to see what's new...

Anyway, just my 2 eurocents...

ajp- exactly what I was thinking that it only has a form of CQ-vbr.
Have yet to do a real comparision of the quality with Tmpgenc, but the 2 encodes I have tried with main concept were damm fast.

What would be interesting would be to make 3 samples, 1 from each encoder (Tmpgenc, CCE & mainconcept) using its best settings(e.g. for VBR.)

The only limit to the test the Average bitrate must be constrained to 2500 for example. eg CCE can have as many passes as it likes!

Post these 3 samples make a Poll and let members of this forum vote which is the best.(without identifying which sample comes from which encoder. Then after a few weeks identify which sample comes from which encoder.)

[Lamont Cranston]

I believe that the MainConcept Encoder has a serious problem when encoding NTSC DV type 1 avi files. THe problem is subtile and difficult to see and only a few people have detected it. Here is a way you can see it for yourself. I would post the clips but I can't.

You can easly create a test clip that will demonstrate the problem. In a video editor create a title that crawls from left to right with a duration of 30 seconds or more. Frwom this title create a NTSC 720 x 480 AVI DV type 1 video file. Mpeg encode the avi file the NTSC DVD with the MainConcept mpeg encoder v1.1 using the default settings. When you play the mpeg 2 file the title will move erratically for the first few seconds and then move smoothly. The same avi encoded with TMPGEnc moves smoothly for the entire duration.

[ajp]

I think a better comparison would be to use the CQ mode of all the encoders, maybe that's only because I mostly use that myself. For me, multipass is only good for low bitrate stuff and some special cases. Since I only encode to DVD bitrates it doesn't matter that much to me, I find CQ modes gives me better quality and most of all best speed. I can totally understand why multipass is important to those making low bitrare SVCD though... (with high bitrate SVCD close to 2500 I think CBR is better though)

I guess comparing max quality using any setting and not caring about how long it takes is a valid test as well, all depends on what is important to you..

[BJ_M]

Mainconcept does NOT use 2-pass AFAIK. I think it can best be described as Constant Quality mode with the ability to set max/avg/min bitrate, which in my book gives the best of both worlds. Multipass sometimes sticks to close to the avg bitrate and doesn't always give the best quality. Comparing CCE's 3-pass with Mainconcept maybe isn't that fair, especially when comparing the ability to stick to the average.

I hear version 1.2 is coming out soon, will interesting to see what's new...

Anyway, just my 2 eurocents...

i was told that it does a 2 pass in the same way that a hardware encoder does a two pass - that is that it goes forward several frames before encoding them - you can see this frame serving with ccee for example where the frames server may be 10 frames ahead of the encoder ..

i am not postive MC does 2 pass - but was told that it does ... (in the method as above)