How do I get rid of those STINKIN' BLOCK's in the XSVCD's!!

[Spicuzza]

Kinneera, thank you for your opinion. Those are great suggestions.
Between this thread and the other one previously mentioned I think most people after reading both would have a good understanding of the complex nature of digtal video editing.

One more thing......, the Blond and a Brunette!?

Gary

[kinneera]

I ask you to find an encoder, any encoder, that will produce a video of better quality at 352x480 @2,000Kbps MPEG-2 better than TMPEG 2.53 @2,000Kbps at the same resolution.

Easy - CCE.

And my last test, which I guess you didn't have the chance to download ( sorry, the files were deleted from the site ) were a side by side comparison of an MPEG-2 ( "The Mummy Returns" ) created with TMPEG at a constant bit rate of 2,520Kbps, standard SVCD template...The MPEG-1 at a variable bit rate of 300Kbps to 2,300Kbps was equal in quality. No visible advantage.

You compared CBR MPEG2 to VBR MPEG1? Sorry, no dice, unless I'm reading that wrong. Also, it has been admitted by most that TMPGEnc is in fact better at MPEG1 anyway, so again, an encoder discrepancy.

Now I don't know if it's the MPEG encoders, decoders, or what. The point is, and I sustain my facts by so many tests, that MPEG-1 below 2,000Kbps looks better, or is better ( anyway you want to look at it ), than MPEG-2.

Try CCE 3 or more passes with MPEG2 at 1500Kbps avg bitrate from a DVD source at 352x480 (for academic purposes, who cares about the logo). You will find that it easily matches an MPEG1 produced using 2-pass VBR in TMPGEnc.

[kinneera]

One more thing......, the Blond and a Brunette!?

If only I was fortunate enough to have them to spare...

[Mavrick]

OK... Still not working right. Here's what I'm doing now, based on what yall have been saying...

1. Capturing new Hi-8 video in VirtualDub (cause I cant capture uncompressed AVI in VS5 or 6) at 720x480 sound is at 41.x stero.

2. Taking the uncompressed AVI's into MediaStudio Pro and editing it.

3A. Making MPEG2 with a CBR@4000 In TMPGEnc with the filters (noise reduction, sharpen image and soften block noise - at defaults) on. Setting image to 480x480 with 4:3 ratio.
When this is played the video is crystal clear but is very jerky - bad during fast moving camera shots.

OR

3B. Using Media Studio Pro and converting to MPEG2. If I try to set 4:3 aspect ratio it chages the 480x480 (or any) settings. WTF??? So I haven't set the 4:3 ratio...
If I do it at 640x480, then when played on TV. I get a vertical blue line about an inch wide on the right hand side of the screen. but looks good other wise.
If I keep it at 720x480 and encode the picture is smooth and clear but somtimes I get "jittering Ghosted" images during fast moving images or camera moves.


I'm buring in NERO with SVCD template with Compliant SVCD turned off...
tried burning at slower rates of 8x... dont help...

I have just ordered a new 120GB Western Digital SE HDD, So I can better handle the uncompress video files.

I SWEAR - IF I EVER GET THIS TO WORK RIGHT. I WILL MAKE A STEP BY STEP GUIDE ON HOW TO DO IT!!!!

ALL I WANT IS SHORT Hi Quality XSVCD's WITH Hi-8 or VHS Quality.. No Blocking, Ghosting, Jittering, Lines, etc...

Im gonna capture at 480x480 and see how that goes. (using the same methods above)

[adam]

Not that this has anything to do with this thread but I'd like to clear this up. There is an unbelievable wealth of documentation comparing mpeg1 and mpeg2 and they all draw the same conclusion, that at the same settings mpeg1 and mpeg2 are virtually identical in quality unless your source in interlaced. To understand this you simply have to determine why mpeg2 was invented in the first place. Read here: http://bmrc.berkeley.edu/research/mpeg/faq/mpeg2-v38/faq_v38.html#tag62

Mpeg2 was developed for the purpose of being used for broadcast television. Mpeg1 was not acceptable since it is extremely inefficient to implement interlacing in it. The sole purpose of creating mpeg2 was to support interlacing in an effective manner, period. Yes the document states that mpeg1 is designed to be optimal at 1.5Mbits/sec but that is because it was intended to be used at lower resolutions. Optimal bitrate levels for mpeg1 and and mpeg2 are identical, they both are determined by bits per pixel, which is independant of resoultion. I'll quote from that document, "The MPEG sweet spot is about 1.2 bits/pel Intra and 0.35 bits/pixel inter" Again, this applies to both mpeg1 and mpeg2.

Now there are some quality benefits that mpeg2 inherantly has over mpeg1 such as greater DC coefficient precision, non-linear mquant, intra VLC, etc, all of which are explained in the document I linked to BUT the quality improventments of these are minimal and in real world applications are probably not noticable, unless of course your encoding interlaced material.

kwag I don't doubt the validity of your tests but they are inherantly subjective, understandably so, and involve far too many variables. Mpeg encoders can never be completely consistent so you cannot compare the output of mpeg1 and mpeg2 from the same encoder, and it would be even more variable to test the output from two separate encoders.

[Spicuzza]

Mavrick, I personally don't use VirtualDub or TMPGEnc, however, I do use VideoStudio 6.0 and on your last post you said,.... "(cause I cant capture uncompressed AVI in VS5 or 6)...."

In VideoStudio 6>new project>pick any template>in capture step across from Format: change to AVI>click options>video format>select the frame size you want to capture>go back to options>input source>select between composite/tuner/s-video.

You should now be able to capture AVI at whatever frame size you selected. Please understand that the video that plays in the preview screen is just that, a preview of the captured video based on the template frame size not necessarily what the actual captured video looks like.

If you go to the drive/folder that you captured the video you'll see a file name like uvs020429-001, this is your actual captured file and simply play this in Windows Media Player if you would like to see the quality of the capture before you proceed with editng and encoding your final movie.

[Mavrick]

Still Crackin at it...

Welp, Here is whats going on now...

Captured in Virtual Dub 720x480 uncompressed AVI

Converting to 352 x 480 in TMPGEnc at 3000 CBR...


Here goes another CD

[kwag]

I ask you to find an encoder, any encoder, that will produce a video of better quality at 352x480 @2,000Kbps MPEG-2 better than TMPEG 2.53 @2,000Kbps at the same resolution.

Easy - CCE.

And my last test, which I guess you didn't have the chance to download ( sorry, the files were deleted from the site ) were a side by side comparison of an MPEG-2 ( "The Mummy Returns" ) created with TMPEG at a constant bit rate of 2,520Kbps, standard SVCD template...The MPEG-1 at a variable bit rate of 300Kbps to 2,300Kbps was equal in quality. No visible advantage.

You compared CBR MPEG2 to VBR MPEG1? Sorry, no dice, unless I'm reading that wrong. Also, it has been admitted by most that TMPGEnc is in fact better at MPEG1 anyway, so again, an encoder discrepancy.

Now I don't know if it's the MPEG encoders, decoders, or what. The point is, and I sustain my facts by so many tests, that MPEG-1 below 2,000Kbps looks better, or is better ( anyway you want to look at it ), than MPEG-2.

Try CCE 3 or more passes with MPEG2 at 1500Kbps avg bitrate from a DVD source at 352x480 (for academic purposes, who cares about the logo). You will find that it easily matches an MPEG1 produced using 2-pass VBR in TMPGEnc.

@kinneera:

CCE MPEG-1 . Sorry! Been there, done that. CCE MPEG-2 below 2,000 is worse than TMPEG's MPEG-2. TMPEG 2.53 wins.

Tried CCE MPEG-2 at 3 passes. TMPEG CQ beats it in quality. Again, sorry, but you should do more tests.

For the record, I did "Mission to Mars" three times over a month ago:

(1) TMPGENC 2.53 Plus - 2-pass VBR ( New 2-pass mode in Plus version )
(2) TMPGENC 2.53 Plus - CQ=74
(3) CCE - 3-pass VBR

Final file sizes were about 815MB for each.
And the winner was TMPEG 2.53 CQ=74. And I mean FAR better quality than CCE and 2-pass own TMPEG.

And to educate you a little more, here are some links from the pro's:

--------------------------------
At low data rates (1.15 million bits per second), MPEG 2 is unrealistic,while half-resolution MPEG 2 (also called "half-D1") fares only slightlybetter.........
Read more here: http://www.bergen.org/ATM/ATMATM/Background_Information_1.html
-------------------------------

This is related to Video on Demand. But the same principles apply.

MPEG2 does not look very good until you get to 4.5 mbit/sec delivery (See Figure 11., Chart of MPEG)
MPEG1, although marginal at the standard 1.5 mbit/s rate, can be encoded at higher bit rates to improve quality. MPEG1 at 3 mbits looks better than MPEG2 at 3 mbits.
Read more here: http://www.dvsystems.com/products/white.html#mpeg1vmpeg2
-----------------------------------------


And for the grand finale

-----------------------
VII. What's the difference between MPEG1 and MPEG2?
Standardized in 1992, MPEG1 was intended for VHS-quality signal transmission primarily for the then-nascent digital video market and is still considered an efficient use of bandwidth and storage space. MPEG2 was created as the standard for digital broadcasting to provide higher levels of bandwidth transmission needed by, amongst others, direct satellite service (DSS) providers. MPEG1 has an average compression rate of about 1.5 megabits per second (Mbps); the largest possible compression rate for MPEG1 is slightly more than 5 Mbps. MPEG2 bitrates fall between about 3Mbps and about 15Mbps. Interestingly, at bitrates below 3 Mbps, MPEG1 actually performs better than MPEG2. This is because the higher level of precision built into the MPEG2 algorithm requires more processing than MPEG1. At lower bitrates, the percentage difference is great enough to cause significant digital artifacts to appear in MPEG2 encoding that do not appear in MPEG1. MPEG2 should not be used at bitrates lower than 3 Mbps.
Read it here: http://www.d-co.com/digicaster-faq.html

And I could go on, and on, and .....

Good (mpeg-2) night! :P
kwag

[kinneera]

@kinneera:

CCE MPEG-1 . Sorry! Been there, done that. CCE MPEG-2 below 2,000 is worse than TMPEG's MPEG-2. TMPEG 2.53 wins.

Tried CCE MPEG-2 at 3 passes. TMPEG CQ beats it in quality. Again, sorry, but you should do more tests.

This is a subjective opinion that you are entitled to, but it does not establish any technical basis for concluding that MPEG2 is inherently inferior. Refer to adam's comments above. I also suspect you don't really know how to use CCE. I have created SVCDs where people were surpised when I got up to switch the discs. Oddly enough, you're not even comparing to MPEG1 at this point, so what exactly is your point?

And to educate you a little more, here are some links from the pro's:
At low data rates (1.15 million bits per second), MPEG 2 is unrealistic,while half-resolution MPEG 2 (also called "half-D1") fares only slightlybetter.........
Read more here: http://www.bergen.org/ATM/ATMATM/Background_Information_1.html

You really need to develop the skill of critical reading. The comment about half-D1 clearly indicates that the author of this document has made the same mistake you repeatedly commit: comparing the typical full-D1 (720x480) application of MPEG2 to SIF (352x240) MPEG1. Encoding four times the resolution MPEG2 at the same bitrate as VCD is inherently going to make MPEG1 look superior, but only because the basis for comparison is inherently flawed. If I cannot get it into your head that the MPEG2 must be encoded at 352x240 for the comparison to be fair (and at this rate, it seems probable that you will never grasp this concept), then there is no point in continuing this discussion. I would also add that this link is hardly professional, as the author doesn't even know the correct resolution of broadcast television and no credentials are provided.

MPEG2 does not look very good until you get to 4.5 mbit/sec delivery (See Figure 11., Chart of MPEG)
MPEG1, although marginal at the standard 1.5 mbit/s rate, can be encoded at higher bit rates to improve quality. MPEG1 at 3 mbits looks better than MPEG2 at 3 mbits.
Read more here: http://www.dvsystems.com/products/white.html#mpeg1vmpeg2

I will repeat it again, in the hopes that it will finally sink in. RESOLUTION!! This article is clearly comparing the DVD application of MPEG2 - at FULL D1 to the videoCD application of MPEG1 at SIF. So again, no shit MPEG1 looks better at 1.15 Mbps if you are encoding 1/4 the number of pixels. They are correct that 720x480 only begins to look good at ~4.5Mbps, but thats a resolution issue, not an MPEG1v2 issue.

MPEG1 has an average compression rate of about 1.5 megabits per second (Mbps); the largest possible compression rate for MPEG1 is slightly more than 5 Mbps. MPEG2 bitrates fall between about 3Mbps and about 15Mbps. Interestingly, at bitrates below 3 Mbps, MPEG1 actually performs better than MPEG2. This is because the higher level of precision built into the MPEG2 algorithm requires more processing than MPEG1.

I've critiqued this inferior document before. First off, the author clearly hasn't done his own research, or he would know that the MPEG1 compression limit is not 5 Mbps, but 100Mbps. The "more processing" argument is just ridiculous - if it takes more processing to create the MPEG, this will only slow down the encoding, not make the final product lower quality. Likewise, if it takes more processing power to decode, it would be universal to MPEG2 (i.e. at any bitrate), and you can be sure any specifically designed hardware decoder (read: DVD player) will have the sufficient processing power - it will not affect quality. The analogy would be trying to play MPEG2 video on a 386 - yeah, it will look like shit, but not because MPEG2 is inferior, but because the decoder is inferior. In fact, the logical counter-argument is that the extra precision/processing is actually netting you better quality. Again, you need to think critically when you read...

[Mavrick]

I DID IT!!!!!! YEAH!!!!!!!

Got my XSVCD CLEAR - CRISP - SMOOTH!!!!!

Here is what I did...

1. Captured in VDUB at 480x480 uncompressed AVI 44.x dv sound

2. Pulled AVI into MediaStudio Pro edited and output it to a MPEG2 with a variable bit rate of 4000 @ 480x480.

NO TMPGne, no other utils besides these two. Best part is MediaStudio can render the MPEG file in a fraction of the time! It's FAST!

Thanks to EVERYONE HERE!!!!

[injunpana]

Wow, this is fun! Can't wait to find out who turns out to be correct or more correct, is it Kwag or is it kinneera? Kwag's research and trial and error method or Kineerra's supposed vast knowledge and expertise in the field of video, or whatever this field is they are talking about.

I could be wrong but with Kwag, what you see is what you get. He is obviously not an expert, just a guy who read anything that would help him make the best vcds he can. He is a self-taught guy on how to make the best vcds out there through sheer doggedness in research and testing it in actual. Everything he says is the result of his experimenting. Maybe his methods were wrong, or erroneous .... i dunno, but still it did produce results that others are raving about.

Kinneera is an unknown. Is he an expert or an authority in this field such that he is able to critique a book or a technical paper and say the author was wrong by just reading it and running it through his brain, even without doing some testing or maybe some calculations? I wonder what his qualifications are? His profession? His field of expertise? Maybe he could enlighten us more about him so we can decide and would know if we are to believe him.

I would surmise that authors write their books have colleagues check or edit it before publication, that they tried their best to research for it especially if the book is a technical one. They do this because their reputation is on the line. And here's Kinneera who reads it, think on it, analize it, and just like that he can say that the author is wrong. I'm in awe of this guy, he must be the top in the field to be able to do this. Maybe he really is the all-knowing that he claims to be.

I'm staying tuned ....

[kinneera]

Is he an expert or an authority in this field such that he is able to critique a book or a technical paper and say the author was wrong by just reading it and running it through his brain, even without doing some testing or maybe some calculations?

I've never claimed to be an expert. However, none of the sources Kwag likes to quote are published in books, with proper editing and peer review. It's also generally a good rule of thumb that if they state something that is flat out factually wrong (as in the case of MPEG1's maximum bitrate, for example), which can be verified in a reputable technical document (MPEG spec. itself), then it is probably not a credible source. And just because something is published - in any medium - does not necessarily make it correct or immune to logical review. Do you believe everything you read? If so I've got a purple unicorn I'm looking to sell.

What I can tell you is that given equal parameters, I have done MPEG2 video at avg. bitrates well below 2000Kbps that looked just as good as what I could achieve with MPEG1. Even with TMPGEnc, no less.

All that I would like to see is for Kwag to stop making blanket statements about MPEG2 being inferior at low bitrates based only on his own incompetence in using encoders or setting up proper comparisons. People, such as myself, regularly do MPEG2 at sub-2000Kbps (2-disc SVCDs, for example) that look quite good. Don't think for a second I (and others) haven't done loads of trial-and-error experimentation to achieve these results.

[kwag]

@injunpana

Most of the MPEG stuff I've learned has been by reading almost the complete book "Video Demystified" by the author Keith Jack.

I can't praise this book enough.

I read most of that book specifically to learn about the internals of MPEG encoding, so that I could understand the technical aspects of MPEG-1 and MPEG-2, and try to fix some of the bugs in programs like DVDx and others.

Really this (X)VCD stuff is my hobby, but it connects very tightly with my profession, which has been Software Development ( C, Python, Forth ) in the communications, networking and electronics field.

So the only things that I had to learn was the MPEG stuff. Everything else just came natural. Piece of cake!

I hold an FCC General Class Radiotelephone license with ship radar endorsement for the past 20 years, and I'm also an Extra class ham radio operator for the last 10 years.
So experimenting with TCP/IP and digital transmissions via radio (AX.25 Packet ) has been one of my favorite hobbies for a while.

So I guess MPEG stuff fits perfectly in my "Hobbies Box!" now.

And you're right, I'm not an expert in MPEG ( Yet! ). I just know what works and what doesn't.

But, I like to present facts so that "What I See ( and many people have seen and agreed! ) Is What You Get"

kwag

[injunpana]

Okay, thanks to both of you for responding.

It's just that both of you are so sure of what you are saying and when they conflict with each other I don't know whom to believe. Being a newbie, I don't have enough knowledge to discern who is correct. I guess i may have to do some digging too to rectify that. I just wish that other experts come in too and prove or disprove what was said.

Anyway, all in all, I learned a lot from reading both your posts. Kodus to both of you.

[kinneera]

I wish you the best of luck...playing around with things on your own is still the best way to learn! And of course, we'll be glad to attempt to help you with specific troubles. Don't fear MPEG2, it can be your friend!!

[Mavrick]

I have made my FIRST HIGH QUALITY Final CD with Menus!

I do have one last question on this, if yall know the answer....

Why does 480x480 res WORK???

It's not 4:3 ratio... 720x480 etc... All other res. make "since" it's wider than it's tall. Just like TV.

How come 640x480 put a 1" blue strip down the right side of the screen?

For now I'll chalk it up to MAGIC....

Thanks to ALL!!!

[kinneera]

MPEG has a field in the file header that keeps track of what the aspect ratio is supposed to be (display aspect ratio). Thus, to MPEG the aspect ratio is independent of the resolution. If you select 640x480 and 1:1 aspect ratio, since 640x480 corresponds to square pixels (computer screens), and TV "pixels" are slightly rectangular, there will be a difference in width.

[Spicuzza]

injunpana, both kinnerra and kwag are WRONG! WWWIII coming!

Let me explain by using a non-mpeg analogy. If I have a post card quality clean crisp 35mm photo that I scan into the computer twice, one saved as jpeg and the other saved as bitmap, we can go on and on ad nauseam debating whether the jpeg looks better than the bitmap and vice versa based on the scanned dpi setting.

This pedantic debate is foolish regarding MPEG-1 vs MPEG-2, just like it would be foolish to debate jpeg vs bitmap. They are file formats and compression technologies. Which is BETTER a Blonde or a Brunette?!

Why? Because the quality of the bitmap, MPEG-2, jpeg, MPEG-1, are going to be ultimately determined by the quality of the SOURCE material! i.e. the scanned photo/captured video!!

I DID IT!!!!!! YEAH!!!!!!!

Got my XSVCD CLEAR - CRISP - SMOOTH!!!!!

Here is what I did...

1. Captured in VDUB at 480x480 uncompressed AVI 44.x dv sound

2. Pulled AVI into MediaStudio Pro edited and output it to a MPEG2 with a variable bit rate of 4000 @ 480x480.

NO TMPGne, no other utils besides these two. Best part is MediaStudio can render the MPEG file in a fraction of the time! It's FAST!44.x dv sound

2. Pulled AVI into MediaStudio Pro edited and output it to a MPEG2 with a variable bit rate of 4000 @ 480x480.

NO TMPGne, no other utils besides these two. Best part is MediaStudio can render the MPEG file in a fraction of the time! It's FAST!

So, what did Mavrick do? He finally captured his video 480x480 uncompressed AVI, edited the AVI, then encoded his final movie to MPEG-2.

Now I will bet dollars to donuts that if Mavrick did nothing more than go back into MediaStudioPro and only changed the output to MPEG-1 nobody would be able to see a difference in visual quality.

Best part is MediaStudio can render the MPEG file in a fraction of the time! It's FAST!

Mavrick, not exactly, you can't give Ulead credit for that one per se' however, they do recommend that you capture/edit/encode all using the same frame size. They should KNOW, they wrote the software. The single thing that is going to slow down the encoding time is vast difference from the captured frame size to the encoded frame size.

Kinnerra's not going to like this one but as you found out for yourself capturing VHS tape and Hi8 tape at 720 x 480 is an absolute waste. It's all that did for you was produce a huge captured file size and take forever to encode.

The point of my post here is the absolute critical importance of the quality of the video capture is what will first and formost determine the quality of your MPEG. 1-2, Blonde-Brunette, Mp3-Wav

Gary Spicuzza
Holiday, FL

[kinneera]

injunpana, both kinnerra and kwag are WRONG! WWWIII coming!

Perhaps a little overly apocalyptic. Let me begin by saying that I agree with what you're saying, although I must point out we're not technically "wrong", just talking something different. For clarity purposes, there are a couple of things worth mentioning, though.

This pedantic debate is foolish regarding MPEG-1 vs MPEG-2, just like it would be foolish to debate jpeg vs bitmap. They are file formats and compression technologies. Which is BETTER a Blonde or a Brunette?!

Well, BMP is better. That's actually two very different digital representation methods, one compressed and the other not. That comparison is analogous to uncompressed AVI vs. MPEG! MPEG1 and MPEG2 however use the same underlying compression method.

Why? Because the quality of the bitmap, MPEG-2, jpeg, MPEG-1, are going to be ultimately determined by the quality of the SOURCE material! i.e. the scanned photo/captured video!!

This is of course, absolutely correct. It's one of the "golden rules" of encoding, in a sense. Of course, the uncompressed BMP probably will look better.

Now I will bet dollars to donuts that if Mavrick did nothing more than go back into MediaStudioPro and only changed the output to MPEG-1 nobody would be able to see a difference in visual quality.

My point all along! See, we can live peacefully.

Kinnerra's not going to like this one but as you found out for yourself capturing VHS tape and Hi8 tape at 720 x 480 is an absolute waste.

Like I mentioned before, downsizing (esp. to an irregular size if you are accounting for overscan) can function as a convenient noise reduction. In this sense it may be useful, and I would hold that you always want to at least capture 480 vertical so you retain both fields. That said, you're right, there aren't many circumstances where full-D1 capture is useful, and I rarely do it myself.

P.S. I think I prefer brunette, but it's a close call.

[Spicuzza]

To injunpana:

Ok, it appears two out of three agree that it's the quality of the captured SOURCE material that will actually determine the final quality of the MPEG.

Kwag, what say you based on this single issue?

Gary

[kwag]

To injunpana:

Ok, it appears two out of three agree that it's the quality of the captured SOURCE material that will actually determine the final quality of the MPEG.

Kwag, what say you based on this single issue?

Gary

Hi Gary:

Here's my answer to everyone reading this post. And to confirm that MPEG-1 is better than MPEG-2 below 2,000Kbps.

The following screenshot details three views from three different MPEG files.

Here are the conditions and parameters that I set, to be fair with the encoders.

All samples were encoded at 900Kbps, for test purpose, and to use a low bitrate so that anyone can see the visible artifacts and compare all three files.

I encoded a 5 second clip from the same source.
A small DVD rip, processed through AviSynth, with TemporalSmoother filter to clean even more the original material.

The top screenshot is TMPEG 352x240 @900Kbps MPEG-1CBR.

The middle shot is also TMPEG 352x240 @900Kbps MPEG-2 CBR ( But to be unfair to MPEG-1, I increased DCT to 10 bits 8) ). So the quality is even higher than at 8 bits.

And the bottom shot is CCE 2.5 352x240 @900Kbps MPEG-2 CBR. ( With quality slider all the way to the left to Complex. The best quality that CCE can produce. )

Now look carefully. The screenshot at the top is MPEG-1, and you can see the quantization peaks at 3.13.

Shot number 2, MPEG-2, Q level is 7.17 Oh, what is this! I hear you!. Yes everyone reading this. The higher the Q level, the more compression, the WORSE the quality. This is your MPEG-2 !.

Shot number 3, CinemaCrapt 2.5 MPEG-2. Has a slightly better Q level than TMPEG!. Of course, for MPEG-2, CCE is better than TMPEG. But still far cry from the MPEG-1.

But back to the original topic. The MPEG-1 generated by TMPEG has a Q factor of more than TWICE ( almost THREE times ) better than it own MPEG-2 or CCE's MPEG-2.

And here are the samples, so that you can download them and see for yourself.

The samples and this screenshot will be up for a couple of days, then I'll remove this post, to conserve bandwidth in my site.

Hopefully someone will take the mpeg's and screenshot and post it somewhere else.

Even prior tests I've done at 2,000Kbps, I CAN see artifacts in MPEG-2 in my HDTV. No matter if I use CCE, TMPEG, BBMPEG, Etc, Etc...
At the same bit rate, MPEG-1 artifacts are way way lower. Almost not visible.

Above 3,000Kbps, MPEG-2 starts to shine, and outperforms MPEG-1.

The following is a full screenshot about 400K, so it's big. And as they say, " A picture is worth a thousand words", here is your picture:

http://www.kvcd.net/mpegshot.jpg

Please download it from the link. I didn't want to include it as an in-line picture because it's pretty big.

The sample files are here. They are only video and they're about 500K each. No audio to save space:

http://www.kvcd.net/tmpeg-mpeg1.mpg
http://www.kvcd.net/tmpeg-mpeg2.mpg
http://www.kvcd.net/cce-mpeg2.mpg

I've yet to find an encoder, any encoder, that can produce MPEG-2 with a quality curve of MPEG-1 as shown above at bit rates below 2,000Kbps.

Enjoy,
kwag

[kinneera]

This is better, but there are still some issues...

The middle shot is also TMPEG 352x240 @900Kbps MPEG-2 CBR ( But to be unfair to MPEG-1, I increased DCT to 10 bits 8) ). So the quality is even higher than at 8 bits.

At extremely low bitrates, DC precision greater than 8 is actually worse not better. So that disadvantaged MPEG2, not MPEG1.

( With quality slider all the way to the left to Complex. The best quality that CCE can produce. )

All the way to the left is not necessarily the best quality CCE can produce. It simply prioritizes the reduction of Gibb's artifacts (mosquitos, edge noise) at the expense of more solid-area artifacts. It's actually more of a subjective tradeoff as to which artifacts bother you more or what balance you prefer.

Shot number 2, MPEG-2, Q level is 7.17 Oh, what is this! I hear you!. Yes everyone reading this. The higher the Q level, the more compression, the WORSE the quality. This is your MPEG-2 !.

This is TMPGEnc, this is not "your MPEG2".

CinemaCrapt 2.5 MPEG-2. Has a slightly better Q level than TMPEG!...But still far cry from the MPEG-1.

By putting the quality priority in CCE all the way to 0, you demanded that it attempt to minimize quantization in complex scenes as much as possible. This will force it to compromise everywhere, resulting in a worse average quantization. Of course the real issue is that in CBR encoding, each frame is going to get the same number of bits no matter what. The only question is how well the encoder utilizes those bits, which is a variable not related to the MPEG specification itself. It strikes me that what you are proving is that the programmers of encoders have optimized CBR MPEG1 much more since there is a standard based on it, and almost certainly committed more time to optimizing VBR, especially multi-pass VBR, for MPEG2, since DVD and SVCD are based on that. But again, it all points back to too many variables, primarily in the encoders, and not the MPEG specifications. I encourage you to review adam's previous comments on this.

Using multi-pass encoding, I can produce a 60 min. SVCD that looks just the same as the comparable XVCD. But it would just be a matter of my techniques might vary from yours in some way. The problem with the whole discussion is that it is inherently unresolveable, since the MPEG specification only standardizes decoding, not encoding. But the bottom line is that the DCT and motion-prediction techniques that constitute the bulk of both specifications are nearly identical.

I would really be happy if you would merely allow that given the proper time and commitment, either format can be made to look just as good. I've done both, and I've been happy with both, and I would like for others to know that it is possible.

[thxkid]

You wanna get rid of the blocks at 720x480 you need to bump up the bitrate to 6000, had the same problem with my Daewoo5700, 4000 bitrate just had to many blocks, looks perfect at 6000 bit, tried 7000 and 8000 made no difference so I stuck with 6000 which gets me about 15-16mins on 80min. CD-R's and 20min. on the 99min. CD-R's.

[Spicuzza]

This thread should be locked as Mavrick has emerged the clear WINNER , by his intelligent use of the genius of the collective mind!

[Eyepop]

I am rather new to VCDHELP but I must say about 75% of the material here seems way beyond my degree of knowledge and beyond my degree of caring but it is very good information I'm sure. Most importantly about this thread is that this is probably and not only the most helpfull but also the MOST FREAKING FUNNY DEBATE EVER! Thanks for all of the mindless banter to keep me interested and thanks to Mavrick for being the only person here to express things in laymens terms.
I only have one question and it's for Mavrick->How many CD's and hours do you estimate you spent trying to use this thread to your ultimate success?

[Mavrick]

I know what your saying...

I spent 3 weeks (approx 2-5hrs a day) and went through 25 or so CDs.

I also learned that after making the MPEG2 Files, using Ulead's DVD Movie Factory ($50US) to burn the CD's, fixed my fast forward and rewind problems. It also makes better menus than Nero.

[holistic]

Now I don't know if it's the MPEG encoders, decoders, or what. The point is, and I sustain my facts by so many tests, that MPEG-1 below 2,000Kbps looks better, or is better ( anyway you want to look at it ), than MPEG-2.

kwag

Will agree with you on this kwag . Much playing around with low bitrate encodes showed me similar results. I mentioned this point some time ago and was shotdown.

It dosen't matter now as i have adopted a new paradigm.

Mavrick : you are capturing from Hi8 camcorder - home videos ????

Consider the following - If it is home family movies your are going to want to preserve the content. My research has shown me that even 'standards' become oboslete (can you say 8 track/78 record) imaging how hard it would be to recover/remove non-stardard data from a disk in say ...30 years.
I therefore decided to stick with an industry standard of 1/2 D1 or 352*480 (for the NTSC system) and encode 20 (+_ 2 minute) CD at a bitrate of about 4500 and used 48KHz audio.
It will now be easier to transfer my Mpeg2's to DVD when/if i get a DVD-RW (that FMD drive looks sweet if it ever comes to production)
*note : i have read some DVD authouring programs like to see seperate audio / video streams - as i haven't got that far i cannot comment - but am not worried as it can always be split.

[Darr1960]

Maverick, I am doing the samething as you I am using
VideoStudio 5 and converting my 20 mins of home DV shot with my Sony Trv 330 to UleadMpeg2 high setting(4000kps) to play on my RCA 5240P dvd Player..

I have been messing around with the settings of
tempgenc, and even tried trex SVCD , and I still find
that my XSVCD mpeg files created with Ulead mpeg2 high
quality look better than the SVCD mpeg2 files I created with Tempgenc
and Trex.. The XSVCDs look really good to me..I have authored using the
Ulead DVD moviefactory creating XSVCDs and I have created using Nero with the SVCD plug-in(turning off non-complaints)..I use these settings because they are for my viewing only.. I create standard whitebook VCD when I share... If you like using Uleads Mpeg2 high setting(720x480@4000
stick with it...

[Kdiddy]

Having been through this pissing contest many of times b4, the bottom line boils down to this....

There are too many variables (source video, encoding software in mostly inexperience users hands, decoding chips on DVD players, TVs/monitors, & lastly EYES) in the whole encoding/decoding process in making both mpeg1/mpeg2 video to make blanket statements of good or bad about either side. In the end, it is going to be up "you the newbie" to use/trial and error method to see which works best for your setup. I use to be concern about misinforming the "newbies' with blanket misstatements. But if you are the type of person to assume that is only one way to accomplish a goal because thats someone told you, the you deserve to in end up with a mind-numbing headache that reading all these ranting posts can give you.

- Use the guides as an outline.
- Learn what options do what on the encoder, and tweak away.
- It is not an overnight process, so dont get upset if you are still tweaking a month later.
- If still confused, search forum first, if no luck then asked concise questions about your problem. But please stay away from asking blanking questions because you end up with this type of post here.

[Akai Rounin]

OK, well one thing that has come of this war of words between kwag and kinnera is that I have learned something. I had tried to convert video to MPEG-2 for SVCD in times past and the result was ALWAYS horrible. At the same, or even LOWER bitrates, ordinary MPEG-1 looked better. Reading that the bottom of the barrel limit for SVCD is around 3000 bits/sec explained why I was having such trouble. I had always used bitrates that were under 2000 for SVCD, that's why they looked so bad. This is cool, now I know I won't have to bother with MPEG-2 until I get a DVD burner. Thanks a lot guys!

HUN-YA!

Red Ronin