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

[BJ_M]

Please explain...

"avoiding things like making a 720 x 480 mpeg2 from 720 x 486 cap"

I capture from DigiBeta to uncompressed 4:2:2. My SDI capture card, like yours, defaults to 720x486. How else can one make a DVD compliant MPEG2 unless it is resized to 720x480 while encoding?

if you resize it - with any type filter - it defaults to be resized by center positioning and because the ratio of differance of size is so large (or consider the size is so close) that the pixels will become staggered which causes smear .. you can see this if you blow up a uncompressed frame pre and post resize .. either you postion the source to the left and resize to left or you just crop 6 pixels from one side (is best way). i use eyone fusion and trim in those cases - but for the test above i mentioned i trimmed in v-dub

i learned this doing blow ups to 70mm film and high def on big screens (really big, but even d-beta is to low a resolution for final output, but good for tests) .. and it is a documented fact ,ive seen it mentioned on a couple NLE and the like sites also...

[MNale]

Sorry to sound stupid, but please explain...

"either you postion the source to the left and resize to left or you just crop 6 pixels from one side (is best way)"

720x486 is width and height. Why trim from the left side (720 = width) when it is already DVD compliant at this width? I can understand a six line vertical crop (maybe 3 from the top and 3 from the bottom) perhaps? I capture in Adobe Premiere and a 6 line crop of this sort would be easy. Is this what you mean?

At present, I use no filters and encode to .m2v with Premiere CCE plugin and it automatically sizes to 720x480 for me. If a crop within Premiere while encoding with CCE will produce better quality, I'll go that method. Thanks.

[BJ_M]

BJ_M:

Would you qualify this as too much NR. I have a noisy calbe feed ( probably average for most people ). I belive my NR is light to moderate as I find the smearing and ghosting from over NR to be more objectionable than the noise in the first place.

I cap blacks to 16 and lights to 235 no other leveling is necessary because I have calibrated my brightness and contrast to begin with. I do a 2x2 smoother to restore continuious tone areas ( it does not smooth edges ). Then I do a temporal filter that has a ramp based on luma channel ( ie the brighter it is the more leway it has to filter ). The temporal is a non-locking, if two pixels are close enough to be filtered it blends them together. At this point I do a bicibic resize from 720x480->480x480 and in some cases I do a very light sharpen to bring out the pop again. I only do sharpening on progressive materal since I don't have a good interlaced shapeneing filter.

If you want to see some before and after it really does make a huge diffrence in quality of output materal.

yes - it prob is not to much in the case of cable cap -- i dont honestly know as i dont cap off air ... and you would have a better feel of what is needed (though i would have black set to 0 myself) .. i was more refering to so called settings some people have written, often higher levels than you on sources such as dvd rips and DV ..

i think what KingJohn says here is correct for most people - use just tmpgenc and v-dub and default templates would fix a lot of problems some have (though believe me audio sync issues happen even in a studio -- specially with film conversions from special venue 30 and 48fps films) .

[BJ_M]

Sorry to sound stupid, but please explain...

"either you postion the source to the left and resize to left or you just crop 6 pixels from one side (is best way)"

720x486 is width and height. Why trim from the left side (720 = width) when it is already DVD compliant at this width? I can understand a six line vertical crop (maybe 3 from the top and 3 from the bottom) perhaps? I capture in Adobe Premiere and a 6 line crop of this sort would be easy. Is this what you mean?

At present, I use no filters and encode to .m2v with Premiere CCE plugin and it automatically sizes to 720x480 for me. If a crop within Premiere while encoding with CCE will produce better quality, I'll go that method. Thanks.

what the question is -- does CCE crop or resize the image ? if it resizes - pre crop in adobe .. tmpgenc and Main Concept resizes (unless you use the crop function) ..

yes - it is as you say top and bottom .. left resize refers to a type of resize and i didnt explain myself clearly and re-reading my post i dont understand it (best to accually be awake typing)...

or you can also - for the fact when you cap off d-beta -- you get a image all the way across the frame - even outside the safe area .. you can also just reduce the image to be just outside the safe area ( we reduce by 6%) and then you have full image on tv .. unless you get some of the newer toshiba and sony wide screen HDTV monitors for the home which allow to to set underscan (nice!) ... my cap card allows me to set a crop as its capping .. but its unstable enough sometimes that i dont use that feature ..
im shopping right now for a new card with component and sdi I/O and real time effect rendring..

[Thorsten Schumann]

Hi,

after reading the postings, here are some comments:

Large files problems:
Files larger then 2GByte which are imported by the AVI-Import failed because of the Windows-AVI restriction. OpenDML is done through DirectShow which has a lower priority in the current version. We will fix this problem in the next version.

Offset of the start time code:
There are two settings for this:
1. On the Video settings page for the GOP timecode
2. on the multiplexer settings page for the SCR/PTS/DTS timecodes.

Crash on video only encoding:
This could not be recreated here. Please give us some more details

Errors if selecting MPEG1 Layer2:
This happens if you switch from DVD-NTSC with PCM audio. It's a bug in the program and will be fixed on the next version.


The next version will be available in a few days.

Thorsten

[BJ_M]

Thorsten Schumann - thank you for your response:

some replies;

"Large files problems:
Files larger then 2GByte which are imported by the AVI-Import failed because of the Windows-AVI restriction. OpenDML is done through DirectShow which has a lower priority in the current version. We will fix this problem in the next version."

I get around this by just doing a frame serve which works very well with your product.


"Offset of the start time code:
There are two settings for this:
1. On the Video settings page for the GOP timecode
2. on the multiplexer settings page for the SCR/PTS/DTS timecodes. "

1. on the GOP panel it seemed this was a delayed start for encode - not a time code offset , though if it IS a time code code offset how does one input exact 00:00:00;00 format or is it frame or sec only based ..
2. on the multiplexer tab this is only start up delay padding for either video or audio for sync purposes -- which is not offset time code , which is encoded pre multiplex anyway ..




"Crash on video only encoding:
This could not be recreated here. Please give us some more details "

I didnt have this problem either ...



"Errors if selecting MPEG1 Layer2:
This happens if you switch from DVD-NTSC with PCM audio. It's a bug in the program and will be fixed on the next version. "

ok ...


other bugs or wishs:

- selecting 30fps in detailed setting and then hit adv setting and 29.97 is shown there.
- using a 30fpsNDF input file and enocde @ 29.97 will end encoding before really done.
- cant set level ID to high level , main profile -- common for HD (HD ability would be GREAT benifit)
- 15meg/s limit on data rate .... if you can get it to be same as TMPGEnc you would be more compititive (50meg/s +)
- DVD audio delay in mulltiplex should be 80 - 90 ms
- in VBV in custom settings you have to double the number you want - though TMPGEnc has the same isssue at higher bit rates..
- no ability to use frame seq. like targa or tiff or jpeg files ... would be GREAT benifit ...
- you can resize image but can not crop --so it seems, though an active area display size is hidden in adv controls ..
-saving settings and custom templates seem not working.

don't get me wrong - its a good product - if fact: CONGRATULATIONS on a great product ..

[FM]

Product IS good (as everything from Mainconcept though) but the price is too high for the individual. If the price would $100 I would buy it without thinking...

What about the support and upgrade? If version 2 comes out, will users of v1.1 have to buy it again with the full price?

[Lamont Cranston]

I discovered a problem with the MainConcept Meg Encoder V1.1. I encoded an NTSC DVD using the default values from a 720 x 480 DV Type 1 video. There was one section of the encoded video where the frame rate appeared to slow down but the audio stayed in sync. I switched the DVD player to step mode and viewed the suspect section step by step. As I stepped, the image would show movement for 2 successive steps and on the next step there would no image movement but just a change in image quality and/or artifacts. This sequence of steps continued for ~45 seconds and then went away. The original AVI and the mpeg 2 files I encoded on TMPGEnc, Ulead's MainConcept encoder in DVDws did not show this problem. The problem occurs every time.

I am running W2k Pro with all the service packs and patches applied.

[Lamont Cranston]

I discovered a problem with the MainConcept Meg Encoder V1.1. I encoded an NTSC DVD using the default values from a 720 x 480 DV Type 1 video. There was one section of the encoded video where the frame rate appeared to slow down but the audio stayed in sync. I switched the DVD player to step mode and viewed the suspect section step by step. As I stepped, the image would show movement for 2 successive steps and on the next step there would no image movement but just a change in image quality and/or artifacts. This sequence of steps continued for ~45 seconds and then went away. The original AVI and the mpeg 2 files I encoded on TMPGEnc, Ulead's MainConcept encoder in DVDws did not show this problem. The problem occurs every time.

I am running W2k Pro with all the service packs and patches applied.

Additional info: I transcoded the avi file from DV type 1 to DV type 2 and mpeg encoded the new file. The slowdown problem was not there. I transcoded the DV type 2 file back to a DV type 1 encoded this new file and the slowdown problem was there. I tried several avi DV type 1 clips with high action and encountered the same slowdown problem.

The slowdown is subtile and hard to detect with the untrained eye. Fortunately I had two visiting Phd EEs whose expertise is image processing and they detected the slowdown immediately.

[BJ_M]

I discovered a problem with the MainConcept Meg Encoder V1.1. I encoded an NTSC DVD using the default values from a 720 x 480 DV Type 1 video. There was one section of the encoded video where the frame rate appeared to slow down but the audio stayed in sync. I switched the DVD player to step mode and viewed the suspect section step by step. As I stepped, the image would show movement for 2 successive steps and on the next step there would no image movement but just a change in image quality and/or artifacts. This sequence of steps continued for ~45 seconds and then went away. The original AVI and the mpeg 2 files I encoded on TMPGEnc, Ulead's MainConcept encoder in DVDws did not show this problem. The problem occurs every time.

I am running W2k Pro with all the service packs and patches applied.

Additional info: I transcoded the avi file from DV type 1 to DV type 2 and mpeg encoded the new file. The slowdown problem was not there. I transcoded the DV type 2 file back to a DV type 1 encoded this new file and the slowdown problem was there. I tried several avi DV type 1 clips with high action and encountered the same slowdown problem.

The slowdown is subtile and hard to detect with the untrained eye. Fortunately I had two visiting Phd EEs whose expertise is image processing and they detected the slowdown immediately.

maybe its related to thier bug with priority on DirectShow (just a wild guess)

a slow down as you say is hard to explain other than a buffer empty out (VBV underflow) , peaky high bit rate on a gop - or something else ..

btw - the guy from main concept was right about time code, offset is located near the GOP setting (not the other though i suspected) and it works as advertised... maybe im the only one who uses time code on both ac3 and mpeg2's -- but in spruce it sure keeps them in sync if thier time code matches ..

[FM]

Heh... If file size is longer than 2Gb, the encoder shows it as negative. Also SpruceUp and ReelDVD refuse PAL files created by Mainconcept encoder with the errors:
- frame rate value 4 (reserved) is illegal -- ReelDVD
- resolution is invalid for DVD -- SpruceUp

Also preview cannot be switched off while already encoding even though menu command are available for that.

Small bugs, easy to fix but the product should not have them when it goes to public.

[Lamont Cranston]

I discovered a problem with the MainConcept Meg Encoder V1.1. I encoded an NTSC DVD using the default values from a 720 x 480 DV Type 1 video. There was one section of the encoded video where the frame rate appeared to slow down but the audio stayed in sync. I switched the DVD player to step mode and viewed the suspect section step by step. As I stepped, the image would show movement for 2 successive steps and on the next step there would no image movement but just a change in image quality and/or artifacts. This sequence of steps continued for ~45 seconds and then went away. The original AVI and the mpeg 2 files I encoded on TMPGEnc, Ulead's MainConcept encoder in DVDws did not show this problem. The problem occurs every time.

I am running W2k Pro with all the service packs and patches applied.

Additional info: I transcoded the avi file from DV type 1 to DV type 2 and mpeg encoded the new file. The slowdown problem was not there. I transcoded the DV type 2 file back to a DV type 1 encoded this new file and the slowdown problem was there. I tried several avi DV type 1 clips with high action and encountered the same slowdown problem.

The slowdown is subtile and hard to detect with the untrained eye. Fortunately I had two visiting Phd EEs whose expertise is image processing and they detected the slowdown immediately.

I loaded the MainConcept Meg2 file with the slow down problem into Ulead's Media Studio Pro and stepped through the clip frame by frame. The same problem can be found at the same points where there is image movement for two frames and on the next frame there would be no image movement only a change in image quality or artifacts.

[Lamont Cranston]

The MainConcept audio sync issue reported in this thread
http://vcdhelp.com/forum/viewtopic.php?t=114179
may be caused by the frame slowdown problem.

[Lamont Cranston]

I ran one last test to verify the frame rate slowdown In the MainConcept mpeg encoder. I captured a satellite newscast with a crawl across the bottom in NTSC 720 x 480 AVI DV Type 1 and encoded it with both MainConcept mpeg encoder V1.1 DVD default settings and TMPGEnc with similar settings. I loaded each Mpeg file into Ulead's Media Studio Pro V6.5 and stepped through the video frame by frame. When I stepped through the TMPGEnc mpeg video the crawl moved a small amount with each frame. With the MainConcept encoder the crawl did NOT move a small amount in each frame. I saw the same pattern where the crawl would move on two successive frames and on the next frame there would be no movement.

I like this encoder but it has too many bugs.

[BJ_M]

i saw this now also ... it doubled up on I frames once every while ... so odd - but only on certain files ..

well tmpgenc has the odd thing of spiking the data rate -- specially at the end of a file ..

[Lamont Cranston]

i saw this now also ... it doubled up on I frames once every while ... so odd - but only on certain files ..

I could not find the problem with DV Type 2 files, only type 1. Currently I don't have any tools to look at the GOP. I usually don't have any need. These added I frames could account for the audio sync problem reported in another thread.

[BJ_M]

i saw this now also ... it doubled up on I frames once every while ... so odd - but only on certain files ..

I could not find the problem with DV Type 2 files, only type 1. Currently I don't have any tools to look at the GOP. I usually don't have any need. These added I frames could account for the audio sync problem reported in another thread.

get yourself 'bitrate viewer' full version ... this is one great program ..

not only can you check the bitrate - but you can check all GOP's , change all mpeg header info (every detail including aspect ratio and bit rate with out havign to re-encode) .. read transport and program and elementary streams - even if mux'd with ac3 .... checks for other data as well ..
only thing it can not check on is VBV buffer under/over flows - but these is only a couple, programs that can .

also checks avi 's (i havnt tried it) ..

[Lamont Cranston]

get yourself 'bitrate viewer' full version ... this is one great program ..

not only can you check the bitrate - but you can check all GOP's , change all mpeg header info (every detail including aspect ratio and bit rate with out havign to re-encode) .. read transport and program and elementary streams - even if mux'd with ac3 .... checks for other data as well ..
only thing it can not check on is VBV buffer under/over flows - but these is only a couple, programs that can .

also checks avi 's (i havnt tried it) ..

I got the Bitrate Viewer and it works great. Thanks. I looked at the GOPs in the MainConcepts mpeg 2 files and did not find any extra or missing frames.

[Baldrick]

Great comparison, I added a link from the latest news on the start page.

BJ_M please include the new panasonic mpeg2 encoder also cause I said that you would test it....

Panasonic mpeg2 encoder demo http://pdn.securethink.co.jp/en/product/fset_J13.html

[Quietnewbie]

I tried to compare your svcd samples but unfortunatly it was not possible. Your use of a non-svcd encode as a svcd encode for cce invalidates this comparison. Additionaly, you did not compare the more reasonable uses of either the cce or tmpgenc tools. If you want to compare the two products then you should us a 4 pass process for cce and the cq setting for tmpgenc. Adjust the bitrates within both tools to achieve the same size files. You would then have a legitimate benchmark of each to compare. I would also suggest using tmpgenc defaults for the various settings, and as always, make sure your field order is set properly. Interlaced source tends to be more difficult to setup.

[BJ_M]

Quietnewbie -

the CCE encode for svcd is svcd specs except for that i didnt resize it to 480 x 480 ... it is a valid comparison for quality ..

".....you did not compare the more reasonable uses of either the cce or tmpgenc tools"

< -- what are more resonable uses than encoding dvd's , svcd and vcd's?

"......you should us a 4 pass process for cce and the cq setting for tmpgenc...."

< --- 4 pass proccess for CCE is a waste of time accually - even CCE says so . Comparing CQ against a VBR is not a valid comparison really , just for the main fact is that since i do this professionally daily I dont have time to do all the variations possable.

"....Adjust the bitrates within both tools to achieve the same size files..."

Well again the point was that i used the standard templates for the most part and "adjusting" would constitute NOT using standard templates ... plus the fact is using standard templates for the most part shows that at nearly equal encoder settings that output file size is different but quality is all about the same. I am limited by the fact anyway that there are rigid upper limits on VCD and SCVD for them to remain thus. if i slide the bit rate around to make the file sizes the same (like i have time to fool around with that) , obviously i would either be above the bite rate levels for valid svcd and vcd or i would have to make one or two of mpegs at a artificially low bit rate - which of course the quality would suffer. proving nothing except lower bit rates = lower quality.

"....make sure your field order is set properly..."

ah yes thank you for that advice - i realize the samples must be very bad , i always wondered what the heck i was doing several 100,s of thousands of $$ of software and hardware in just my office - not to mention our studio equipment. :roll:

[therick]

lol BJ_M- great comparison, I guess now we can see that there are two MPEG-2 encoders out there better and faster than TMPEG- one is outrageously priced and one is reasonably priced, at least for the majority of hobbyists on this forum...good work.

[mol3000]

how long does it take a full movie at 3 pass in cce 2.66 to encode in dvd ???????

[offline]

Can anyone suggest what I can do to get Main Concept
to accept picvideo MJPEG (non yu2 forced) .avi with pcm sound
to load without crashing in XP ?

I can load other codecs w/o problems but I tend
to capture in picvideo Quality=18 . stereo PCM
16Mhz, 16 bit audio (as VHS is the source)

thx

[LeeBear]

BJ_M: Nice comparison of the encoders. I downloaded the SVCD samples and was fairly impressed with the Main Concept encoder. I have a few questions. Was the sample clips extracted from a longer encode or was the 30 second clip all you encoded for each test? I ask because a I would think a short clip wouldn't be able to show the encoders true effeciency at distributing bits like a longer movie length encode would. Secondly I noticed after downloading the samples were all slightly different size (probably due to the VBR) CCE being the smallest at 3.65MB and Main Concept the largest at 4.23MB. Your rate the Main Concept as looking better and I would agree with you after looking at the samples, but I wonder if it's due to it being a better encoder or it's 16% larger file size? Why didn't you use a common SVCD bitrate like a CBR of 2524 kb/s to leave unequal bitrate out of the quality equation.

-LeeBear

[BJ_M]

Leebear - the clips were the full lenth encoded and were not trimmed form a longer section .. the lenth used should be enough - and kinda interpolates what ive seen on longer runs.

CBR at same setting on all would be a good comparison except that is not used by to many people here .. i may do it anyway if i get the chance ...


coming today: CCE 2.66, New Panasonic, Vitec DVD tools, Canopus ProCoder, blast from the past Mpeg Repair, and 2 pro hardware encode's (5000$ and 25,000$ mpeg encoders) and maybe some other software encode's ..

[Lamont Cranston]

I looked at your DVD test samples and for me to evaluate the results I downloaded and played each sample on a large screen TV. They still all looked 'good' and the only way I could detect any differences was in A/B tests and I did not detect any blocking. [My eyesight is not as good as it used to be ] I would like to 'suggest' that you include in your test video a series of clips that would stress the mpeg encoders to failure so we we could learn the limitations and characteristics of each mpeg encoder. You define scope of the test so it does not get out of hand. This is only a hobby for me but my experience has taught me that comparison of mpeg encoders is not black and white but trade offs. For example I have test clip where TMPGEnc has blocking in one scene and MainConcept does not and the reverse happens in the next scene same clip. Given a 'good' source I have had the Ligo LSX35 produce a 'good' results (DVD).

[choonsin]

how abt the encoding time taken? so among these 3, which is the fastest?

[BJ_M]

its pretty close between Main Concept and CCE

[choonsin]

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)