Sample HD MPEG-1 Clip... [Archive]

Shinobu

22nd June 2005, 13:36

Kika

22nd June 2005, 13:48

i have better result in encoding mpeg1 with quenc than with tmpgenc (and it's free).

Did you ever try CQ-Mode with P- and B-Spoilage setting of 0?
Works good with 1280x720 and 960x720

digigami

8th July 2005, 10:50

this sample is very dark and in a 2.35 AR => hight compressibility.
it as also a lots a block in some aeras.

i've tested and used a lot mpeg1 and mpeg2, with hdtv here's MY visual impression:
hdtv mpeg2 > 10 mbps looks good.
hdtv mpeg1 > 15 mbps looks good also.
hdtv mpeg2 < 6 mbps looks like crap.
hdtv mpeg1 < 9 mbps looks like crap.

with a very good mpeg1 encoder with custom matrix, and 3 or more passes, i think it can do good looking HDTV quality around 9-10mbps, less than that is a dream.

ps: i have better result in encoding mpeg1 with quenc than with tmpgenc (and it's free).
ps2: if it's for hdtv archiving on DVD media, just use mpeg2, you will always have a better quality, and all graphics card (since around 1999-2000) have full mpeg2 hardware decoding

I'm curious about the assertions you make in your post here, especially regarding the differences between MPEG-1 and MPEG-2 bitrates. MPEG-2 adds many new features over MPEG-1, but for 24p (ie. feature film) encoding, there are few. Instead, most of the new features in MPEG-2 are related to the coding of interlaced video. These features include, dct_type, alternate_scan, field_picture, field motion vectors (4-per macroblock) etc.

For standard 24p encoding, there are few gains to be had, at least, that is my experience as the developer of MegaPEG and MPressionist.

Also, you refer to block artifacts in our MPEG-1 HD sample. This sample is designed to illustrate the use of MPEG-1 as a HD streaming/delivery format, not as an archival format. I am curious to know what decoder you are using to evaluate the stream. The diagnostic tool I use primarily is MPressionist, and it includes both highly-accurate (reference) and fairly-accurate (commonly-implemented) IDCT algorithms. My guess is that if you are seeing visible artifacts then the IDCT implementation of your decoder software is not conforming to the MPEG standard. I've reviewed our HD stream in a number of players, including Apple's MPEG-2 decoder (MPEG-2 playback component), MPressionist, VideoLAN and Windows Media Player, and no artifacts are visible under normal viewing conditions, which is to say, the same conditions one would use when watching TV.

Could you describe the playback environment you are using? Thanks.

Gen Kiyooka
--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

easyfab

22nd December 2005, 21:35

for info

http://www.digigami.com/index.html

Mpeg2 better than h264 ?
What H264 encoder is used for the comparaison?

bobololo

23rd December 2005, 03:00

For standard 24p encoding, there are few gains to be had, at least, that is my experience as the developer of MegaPEG and MPressionist.

Gen Kiyooka
Digigami

I've read with interest your press release claiming the very best performance of your VBR MPEG-2 HD encoder. Being myself involved in codecs development, I've to admit that I'm simply amazed by such assertions. With some exaggerations, it's like saying "tractors run faster than formula1 race cars" and you should be very brave to affirm this ;)

Hence, as you can guess I'm really wondering what kind of nitro you load into your tractors to make them so good ;) Obviously I'm not asking you to unveil your secrets but maybe you could at least enlight us with some technical clarifications with regard to your claims : encoding conditions (resolutions, bitrate, rate control constraints, etc.), which H.264 did you compare to and how it was set, how did you compare, and what makes you say which result is better than the other (objective metrics, subjective assessment ?), etc.

Thanks !

CruNcher

23rd December 2005, 06:43

@easyfab and bobololo
looking here http://digigami.in-long-beach-ca.com/megapeg/HD-MPEG-1-VBR/commandN/
it seems he compares with *cough* apple quicktime h.264 :P (no comment)

Sagittaire

23rd December 2005, 13:11

IMO digigami encoder isn't better than for libavcodec for MPEG2 ... lol

JohnV

23rd December 2005, 14:35

@easyfab and bobololo
looking here http://digigami.in-long-beach-ca.com/megapeg/HD-MPEG-1-VBR/commandN/
it seems he compares with *cough* apple quicktime h.264 :P (no comment)
Well, looks like normal marketing to me.. ;) Of course they must know that their claims are not true if they use a good h.264 encoder for comparison.

digigami

23rd December 2005, 19:24

Well, looks like normal marketing to me.. ;) Of course they must know that their claims are not true if they use a good h.264 encoder for comparison.

Hi -

Thanks for the responses. Currently, the primary marketing thrust of H.264 is that it can compress HD feature films down into the same bitrates as standard definition DVD. Our marketing is aimed at demonstrating that MPEG-2 VBR is a viable alternative to H.264 for this purpose. As I stated in my blog, I am willing to do comparative analysis. I am not interested in comparing 5s test clips of random camcorder footage, but rather professionally produced feature film/motion graphics/TV content.

It is not strictly marketing. Both our MPEG-1 and MPEG-2 VBR encoders are producing HD bitstreams with average bitrates well below 9mbs/s. If you look about 1 year ago we were promoting our MPEG-1 encoder as an alternative to Real/Flash for streaming video. You will also find a 1/2 D1 NTSC MPEG-2 sample for DVD on our web site at 0.5mbs/s. There's a consistent theme here - not an isolated incident.

In the last 24 hours I produced a 1280x720x50 fps version of the Mako Shark Club Jacket sample to HD-DVD specs, average bitrate 11 mbits/s and the quality is essentially perfect.

In order for MPEG-2 VBR to be a viable alternative to H.264, it does not have to be better, it merely needs to be comparable.

As far as the quality goes, the clips on our site speak for themselves.

Gen

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 19:32

Well, looks like normal marketing to me.. ;) Of course they must know that their claims are not true if they use a good h.264 encoder for comparison.

If you look in the CommandN folder, you will find the H.264 QuickTime version. Go ahead and compress it using Nero. According to H.264 marketing you will be able to get it down to 1/2 the size of my MPEG-1/MPEG-2 file. So around 85MB. Complete the test and I will post the results and write about it in my blog.

My personal belief is that there are fundamental lower limits (called the entropy of a clip) which can be approached by any good encoder technology but reducing further one runs into the law of diminishing returns.

Also, if you have a sample you'd also like to compare with, I prefer to start with a high quality source.

Gen

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 19:41

technical clarifications with regard to your claims : encoding conditions (resolutions, bitrate, rate control constraints, etc.)
Thanks !

Generally, the encoding target(s) are standard HD profiles used in DTV, HD-DVD etc. Strictly speaking, I think most people are interested in feature film content at HD resolutions, so 24/25 fps at 1920x1080 or 1280x720. Bitrate constraints derived from the standards. 18 mbits/s (720p ATSC HDTV ceiling). 25 mbits/s (1080 ATSC HDTV). 27 mbits/s (HD-DVD).

For H.264 samples I use what I can find. If you have suggestions, I'd like to hear them.

BTW, as I am not an H.264 expert, what features does H.264 have for coding interlaced video?
--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 19:51

how did you compare, and what makes you say which result is better than the other (objective metrics, subjective assessment ?), etc.

This is how I compare things:
http://www.digigami.com/mpressionist/screen.php?screen=comparesplit

I wrote this tool specifically so I can have these kinds of discussions without the dialogue descending into mud-slinging and religious fervour.

I can compare side by side with split screen:
Source/MPEG-2
Source/H.264
MPEG-2/H.264
MPEG-2/MPEG-2

Also, can compare pixel residue (actual per-pixel frame differences) which is useful for picking up differences too fine for the eyes to see.

How do you compare?

Gen

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sagittaire

23rd December 2005, 19:53

Hi -

Thanks for the responses. Currently, the primary marketing thrust of H.264 is that it can compress HD feature films down into the same bitrates as standard definition DVD. Our marketing is aimed at demonstrating that MPEG-2 VBR is a viable alternative to H.264 for this purpose. As I stated in my blog, I am willing to do comparative analysis. I am not interested in comparing 5s test clips of random camcorder footage, but rather professionally produced feature film/motion graphics/TV content.

It is not strictly marketing. Both our MPEG-1 and MPEG-2 VBR encoders are producing HD bitstreams with average bitrates well below 9mbs/s. If you look about 1 year ago we were promoting our MPEG-1 encoder as an alternative to Real/Flash for streaming video. You will also find a 1/2 D1 NTSC MPEG-2 sample for DVD on our web site at 0.5mbs/s. There's a consistent theme here - not an isolated incident.

In the last 24 hours I produced a 1280x720x50 fps version of the Mako Shark Club Jacket sample to HD-DVD specs, average bitrate 11 mbits/s and the quality is essentially perfect.

In order for MPEG-2 VBR to be a viable alternative to H.264, it does not have to be better, it merely needs to be comparable.

As far as the quality goes, the clips on our site speak for themselves.

Gen

Well 11 Mbps for 1280*720*50 or 1920*1088*25 is very high bitrate for H264. Apple QT7 AVC done very good result with 1920*1088*25 trailers sources (very uncompressible sources) and it isn't by far the best AVC encoder. Try with 1.1 Mbps and with good H264 codec and see the result ... !!!

Here little H264 1280*720 (http://multimediacom.free.fr/Download/x264HP_IceAge_720p_1250.mp4) encoding at 1250 Mbps ... try to make that with MPEG1/2 ...

I test your demo encoder and it isn't visually very better than the other popular MPEG2 encoder. Say "MPEG2 is better than H264" (source? bitrate? average quant? subjective? objective?) is totaly ridiculous ...

I have little challenge for you : try to prove that your MPEG2 encoder is the best for metric ...
http://forum.doom9.org/showthread.php?t=98633

digigami

23rd December 2005, 20:35

I have little challenge for you : try to prove that your MPEG2 encoder is the best for metric ...
http://forum.doom9.org/showthread.php?t=98633

At SD resolutions, the smaller DCT block size in H.264 is definitely an advantage. Not so much of an advantage for 720p and definitely not an advantage for 1080p.

I will look at it. From what source did you obtain the trailer? DVD?

- Gen

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 20:36

little H264 1280*720 (http://multimediacom.free.fr/Download/x264HP_IceAge_720p_1250.mp4) encoding at 1250 Mbps ... try to make that with MPEG1/2 ...

This file plays as garbage in QuickTime. I also tried VideoLAN.

What playback system do you recommend.

Also, what source was used in the compression?

Gen
--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sagittaire

23rd December 2005, 20:46

At SD resolutions, the smaller DCT block size in H.264 is definitely an advantage. Not so much of an advantage for 720p and definitely not an advantage for 1080p.

I will look at it. From what source did you obtain the trailer? DVD?

- Gen

It's just a little challenge only for MPEG2 encoder (libavcodec, TMPGEnc, CCE, Procoder, HCEnc, QEnc).

First I think That you must prove that your MPEG2 encoder is better and by far than the other encoder.

My source is an HPII DVD trailler at 5 Mbps.

Sagittaire

23rd December 2005, 20:54

This file plays as garbage in QuickTime. I also tried VideoLAN.

What playback system do you recommend.

Also, what source was used in the compression?

Gen

H264 High Profil. Play fine on Sempron 2.1 Ghz (win OS) with:
- VLC player 0.8.2
- Media Classic Player (ffdshow or nero filter)
- Nero ShowTime

Source are here
http://www.apple.com/trailers/

Make with these source (MPEG4 ASP full HD)
http://forum.surdvd.com/viewtopic.php?t=43651

digigami

23rd December 2005, 20:54

I have little challenge for you : try to prove that your MPEG2 encoder is the best for metric ...
http://forum.doom9.org/showthread.php?t=98633

OK, I downloaded your 'source'. Let me get this straight. You want to do a codec comparison using a DVD VOB as the source material?

Trailers are the lowest class citizens in the DVD compression universe. They are generally quite filtered and encoded at a lower bitrate than titles.

If you are going to run your encoding challenge using compressed material, I suggest using a segment from a Superbit encoded DVD title, and I suggest avoiding CG oriented material as it tends to lack the noise characteristics which make encoding actual photography difficult.

http://www.digigami.com/mpressionist/blog/2005_06_19_archives.html

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sagittaire

23rd December 2005, 21:01

OK, I downloaded your 'source'. Let me get this straight. You want to do a codec comparison using a DVD VOB as the source material?

It's high quality source 5 Mbps MP@ML

The little challenge is to make the highest possible quality for reencoding with MPEG2 MP@ML with only 1.8 Mbps

Sagittaire

23rd December 2005, 21:04

and if you want another challenge test your encoder with this hardest sample
http://forum.doom9.org/showthread.php?t=104484

JohnV

23rd December 2005, 21:13

digigami: I'm downloading your QT source, but your server is very slow atm. Gives only like 3-7 kB/s, so it takes a very long time. I'm not a fan of any marketing claims, but I think it can be agreed that the lower the bitrate used, the higher the advantage for h.264 (in case of a good h.264 encoder, not Apple's...).

In the meantime take Sagittaire's challenge and host your MPEG2 1250kbps and 2500kbps (twice the h.264 bitrate) versions of Sagittaire's clip. :)

Sharktooth

23rd December 2005, 21:21

digigami: Mpeg-1/2 will NEVER and i repeat NEVER reach MPEG4-ASP or AVC efficiency.
Quicktime (or CrapTime as i call it...) is a ridicolous encoder and doesnt even exploit half of he capabilities of h.264.
Get a serious encoder (x264), learn how to set it up and compare your MPEG1/2 encoders with it.
There's only one guarantee, mpeg1/2 will be trashed.

digigami

23rd December 2005, 21:32

digigami: I'm downloading your QT source but your server is very slow atm. Gives only like 3-7 kB/s, so it takes a very long time.
Yes that server is likely under heavy load right now.

I'm not a fan of any marketing claims, but I think it can be agreed that the lower the bitrate used, the higher the advantage for h.264 (in case of a good h.264 encoder, not Apple's...).
The advantages of H.264 are smaller DCT block size, support for framerates less than 24fps and hindsight. There is no question it has the edge. My point is that under the conditions for HD-DVD and Blu-Ray, MPEG-2 VBR is a capable and viable choice for putting feature films on optical discs.

In the meantime take Sagittaire's challenge and host your MPEG2 1250kbps and 2500kbps (twice the h.264 bitrate) versions of Sagittaire's clip. :)
I am transcoding the 1080p Ice Age trailer. However, I will say, that the AVC version of this that he provided does not play on my Mac on either QT or VideoLAN. You're a H.264 person - who is to blame for this?

As far as transcoding the Harry Potter trailer, this has nothing to do with our press release which is about HD frame sizes, bitrates and HD-DVD and Blu-Ray applications. SD is over. MPEG-2 won.

I think it is interesting that this thread died when I asked for technical details regarding the 'blockiness' complaints over the first HD MPEG-1 clip.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 21:35

Update your VLC to a nightly version (0.85) and dont even try the QuickTime decoder/player coz it's b0rked as HELL.
As i said QT does not come even close the half of what h.264 can do...

EDIT: However you're telling us something like a 20 years old overclocked CPU can beat or be comparable to the actual ones... please...

digigami

23rd December 2005, 21:47

As i said QT does not come even close the half of what h.264 can do...

That's good, because both our MPEG-1 and MPEG-2 VBR can reduce most QT H.264 to about half the size.

Like I said, we're in the ballpark. We don't have to beat H.264, just remain competitive.

Any our MPEG-1 VBR files play on PIII machines without dropping frames. They play on Macs. They play on Linux machines. They play on Windows.
File size and bitrate are not the only factors.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 21:49

EDIT: However you're telling us something like a 20 years old overclocked CPU can beat or be comparable to the actual ones... please...

It is unfortunate, but I'm afraid your analogy is incorrect. To determine the best possible job that MPEG-1 can do you must do a mathematically analysis on the smallest possible representation of each frame possible with MPEG-1 bitstream syntax.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 21:50

CPU can beat or be comparable to the actual ones... please...

In actuality, the increase in processor performance is actually quite useful to everyone.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

JohnV

23rd December 2005, 21:53

I am transcoding the 1080p Ice Age trailer. However, I will say, that the AVC version of this that he provided does not play on my Mac on either QT or VideoLAN. You're a H.264 person - who is to blame for this?
You.. :) It plays ok at least with VideoLan 0.8.4a available from its main site.
And forget QuickTime and its AVC already. At least change to x264 and VideoLan or something on your MAC..

digigami

23rd December 2005, 21:54

Update your VLC to a nightly version (0.85)

So let me get this straight: in order to even play his H.264 Ice Age file I have to go to a developer-only prerelease build that was just released last week?

Interesting. If you install Windows NT 4.0 on a P3 with a decent video card, I'm pretty sure my MPEG-1 files will play without dropping frames.

If you go and look at the H.264 chipsets that they are considering for HD-DVD and Blu-Ray, I'd be surprised if they were using H.264 features that aggressively.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 21:55

ok... post a 1000kbps encoding (without any filtering etc) with your mpeg1 or 2 encoder of this clip: http://multimediacom.free.fr/Download/lossless.avi

we'll see what your encoder can do...

JohnV

23rd December 2005, 21:56

So let me get this straight: in order to even play his H.264 Ice Age file I have to go to a developer-only prerelease build that was just released last week?

Interesting.
No, it plays on old VLC 0.8.2, and it plays on VLC 0.8.4a currently available on the main site.. You don't need any nightly builds.

Sharktooth

23rd December 2005, 21:56

So let me get this straight: in order to even play his H.264 Ice Age file I have to go to a developer-only prerelease build that was just released last week?

Interesting. If you install Windows NT 4.0 on a P3 with a decent video card, I'm pretty sure my MPEG-1 files will play without dropping frames.

If you go and look at the H.264 chipsets that they are considering for HD-DVD and Blu-Ray, I'd be surprised if they were using H.264 features that aggressively.
no... the stable release is able to play it back, but since you're running into problems i suggested to use a nightly build.
also the chips for avc playback are not a problem... they're in production phase.
i'm sorry but it's ppl like you that kills any technology advantage with false claims...
produce that 1000kbps encode (some posts above) and prove your encoder can be on par with AVC otherwise get back in your dreams of presumption.

digigami

23rd December 2005, 22:02

At least change to x264 and VideoLan or something on your MAC..

I tried it in VideoLAN. It didn't even open a video window.

As far as forgetting about QuickTime/Apple, in case you didn't notice, they are about to take over the consumer video space with H.264, so I'm not sure ignoring them is a good idea.

BTW, how many profiles for HD encoding are there? Do you happen to known which H.264 profiles are spec'ed for HD-DVD?

Gen
--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:05

AFAIK AVC High Profile was implemented to be used with HD-DVD and HD-DVB (broadcast)...
im still waiting your encode...

digigami

23rd December 2005, 22:05

produce that 1000kbps encode (some posts above) and prove your encoder can be on par with AVC otherwise get back in your dreams of presumption.

Go ahead and produce an uncompressed HD source 720p or 1080p and we can have a discussion. Right now we are using QuickTime H.264 material as the sources? Pretty interesting considering the comments about QT on these boards.

I'm compressing my samples for our website from Pixlet 720p on lossless quality.

Have you even looked at any of our samples? I'm not interested in religious discussions. Either do some analysis or find another thread to rant on.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:07

gimme a sample and i'll prove the falseness of your statements.
btw why dont you try encoding that sample?
scared by the inefficiency of mpeg1/2?

Sagittaire

23rd December 2005, 22:09

The advantages of H.264 are smaller DCT block size, support for framerates less than 24fps and hindsight. There is no question it has the edge. My point is that under the conditions for HD-DVD and Blu-Ray, MPEG-2 VBR is a capable and viable choice for putting feature films on optical discs.

Yes ... certainely but say "MPEG2 is a good choice for BR/HD-DVD" is not "MPEG2 is more efficient than H264 for BR/HD-DVD".

I agree with first affirmation but certainely not with the second ... :D

And H264 advantage is not only smaller DCT block size ... ;-)

Sharktooth

23rd December 2005, 22:10

he doesnt even know the h.264 specs...
he just based his assumption on QuckTime (in)efficiency...

digigami

23rd December 2005, 22:15

Yes ... certainely but say "MPEG2 is a good choice for BR/HD-DVD" is not "MPEG2 is more efficient than H264 for BR/HD-DVD".

I am saying the former, and so we agree. I am also stating that under some conditions our encoder can produce smaller files. Our release also states that for material with high color saturation there is no question, H.264 is superior.

Nowhere have I stated that MPEG-2 is superior to H.264 in any respect. I have simply stated that with a good HD encoder it is a viable and useful alternative.

BTW, what is the minimum CPU/Clock for your AVC sample compressed from Ice Age. Also, which trailer is it?

My first pass on MPEG-1 VBR is 51MB. So you are 4x more efficient with a codec that is 10 years newer. However, my average bitrate is only 3mbs/s which is perfectly in line with our claims that our encoder can do HD at DVD bitrates.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:16

well... maybe someone has something to say...
http://www.avsforum.com/avs-vb/showthread.php?p=6773621&&#post6773621
now... please wake up :)

Sagittaire

23rd December 2005, 22:20

digigami

23rd December 2005, 22:20

I agree with first affirmation but certainely not with the second ...

So now let's go over the advantages of doing HD file sharing in MPEG-1 or MPEG-2 VBR. It is clear that our encoder can produce the streams at well under 10 mbits.

So the advantages are:
Play without dropping frames on lower class of machines.
No special downloads required to view files. Windows Media, QT and VLC all play them fine.
Largest installed base of players.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 22:25

well... maybe someone has something to say...
http://www.avsforum.com/avs-vb/showthread.php?p=6773621&&#post6773621
now... please wake up :)

In June 05, I showed Ben Waggoner our HD MPEG-1 VBR samples and he told me that it looked good and if our MPEG-2 worked with DVDSP HD-DVD it would be valuable:

Gen,
Get that working in MPEG-2 with 1080i60 w/ 3:2 pulldown that's DVDSP4 compatible for HD DVD titles, and you'd have something really valuable. Apple's H.264 is only 720p50/60 right now, and their HD MPEG-2 has a minimum 10 Mbps average bitrate.
-Ben

On Jun 18, 2005, at 1:13 PM, digigami wrote:
Check out our MPEG-1 VBR for mass HD distribution:
http://www.digigami.com/megapeg/samples.php

The main benefit / difference between our MPEG-1 VBR and H.264 is that we can reach Linux and Windows desktops (and pre-QT6 desktops) that don't have QuickTime installed. The bitrates and quality are comparable to the HD trailers from the QuickTime HD gallery.

And, like you said, it's HD at standard DVD bitrates.

Gen

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:29

Advantages of h.264:
Completely new and much more efficient technology.
6x lower bitrates for the same Mpeg2 quality.
DVD quality at DSL bitrates.
HD at DSL bitrates.
More than 5 hours (and easily even more) of recording on a single standard DVD at DVD quality.
Coupled with AAC audio the best and most efficient Audio/Video tech in the world.
VERY MUCH HIGHER encoding speed than digigami encoder :p
Estabilished solution in industry standards (chips are already in production and software players for PCs and MACs already exists).
etc...
should i continue?

also im not intereste in already encoded samples... only in lossless stuff so we can compare.

digigami

23rd December 2005, 22:30

I make HD test actually ... try High Detail level source like that
http://www.apple.com/trailers/dreamworks/overthehedge/hd/
I think that quality can be very good with H264 1080p at 2 Mbps and H264 720p at 1.5 Mbps. Try to make MPEG2 encoding with these bitrate.

Those are very good rates, and is probably below what is possible for us right now, but it may be possible in the future.

At those rates have you measured the loss of high-frequency detail? Because if you take a SD frame and upsample it to HD it is equivalent to throwing away 1-2 pixels of detail through filtering.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 22:31

Coupled with AAC audio the best and most efficient Audio/Video tech in the world.

Yes, AAC is an MPEG-2 technology, and it is excellent. Read the specs. ISO/IEC 13818.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:33

LOL!!!!!!!!!
AAC+ (aka AAC-HE V2) is not mpeg2...

digigami

23rd December 2005, 22:33

HD at DSL bitrates.

Can you make me a 720p sample based on a movie trailer from the Apple site (or any other HD source we both can obtain) that does HD at 500 kbps?

Thanks.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:34

well... 500kbps is not DSL bitrate...
i dont know where you live but we (in italy) have DSL up to 8mbps...
and however what i meant at dsl bitrate was 1200kbps.
however i can easily produce 500kbps streams at near DVD quality:p

http://www.live24.it/ (it's only in italian for now).

digigami

23rd December 2005, 22:39

LOL!!!!!!!!!
AAC+ (aka AAC-HE V2) is not mpeg2...

http://www.vialicensing.com/products/mpeg2aac/standard.html

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 22:42

and however what i meant at dsl bitrate was 1200kbps.

OK, that sounds fine to me. Take any 720p trailer from the Apple site and make 1200 kbps HD H.264 file from it.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sagittaire

23rd December 2005, 22:44

So now let's go over the advantages of doing HD file sharing in MPEG-1 or MPEG-2 VBR. It is clear that our encoder can produce the streams at well under 10 mbits.

1) Certainely with some source/situation but not with all
2) With the same bitrate H264 will produce very better quality
3) All the MPEG2 encoder can make that and perhaps with better quality

Sharktooth

23rd December 2005, 22:46

http://www.vialicensing.com/products/mpeg2aac/standard.html
It's not AAC+ (aka AAC-HE V2)...

Sharktooth

23rd December 2005, 22:47

OK, that sounds fine to me. Take any 720p trailer from the Apple site and make 1200 kbps HD H.264 file from it.
That is already AVC encoded material. As i said im interested in LOSSLESS stuff. The same sources as used for creating DVD movies...

digigami

23rd December 2005, 22:51

That is already AVC encoded material. As i said im interested in LOSSLESS stuff.

Ok, pick any lossless source you want. Preferably something with organic cinematography (the CG trailers like Ice Age are actually quite easy to compress).

As i said im interested in LOSSLESS stuff. The same sources as used for creating DVD movies...
Since that's what you're interested in, let's use one of your samples. I actually prefer to work from uncompressed myself. "The Least Likely" from our web site is made from uncompressed targa files.

Just so we're clear: I actually expect our encoder to be beaten by H.264, but also enjoy the circumstances when it does not. Or does so by a narrow margin. Because if the margin is narrow, the advantage is not as clear.

BTW, I think LAME MP3 encoder is actually quite comparable in quality to AAC, but other MP3 encoders are not.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:55

Just a question... do you know AVC supports 8x8 dct like mpeg2 too? (it's in High Profile...)
And for the last time AAC+ is not mpeg2 AAC...
However i supposed you had some lossless suorces (the ones you used to do your site sample encodings)...

digigami

23rd December 2005, 22:58

1) Certainely with some source/situation but not with all
2) With the same bitrate H264 will produce very better quality
3) All the MPEG2 encoder can make that and perhaps with better quality

Actually, for 720p, on easy material we run around 3 mbits/s and on difficult material between 7-11 mbits/s. The peak bitrate is higher than standard DVD, but the average is not. As I said earlier, highly saturated color is difficult with 4:2:0 chroma.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 22:59

Uhm... Mpeg2 @ 3mbps for 720p? how much blocks? lol...
hoever i'm compressing one of your samples (which is already compressed and already has visual banding problems at 14mbps...).
the sample in the first post is even blocky and flashing...

Sharktooth

23rd December 2005, 23:07

however "someone" asks if you currently know about the tests that were carried out by the BDA in 2004 that showed AVC HP to be around 3x more efficient than mpeg-2...
http://ftp3.itu.ch/av-arch/jvt-site/2004_07_Redmond/JVT-L033.doc

digigami

23rd December 2005, 23:07

I make HD test actually ... try High Detail level source like that
http://www.apple.com/trailers/dreamworks/overthehedge/hd/

I think that quality can be very good with H264 1080p at 2 Mbps and H264 720p at 1.5 Mbps. Try to make MPEG2 encoding with these bitrate.

According to this post:
http://forum.doom9.org/showthread.php?t=98247

The recommendations for DSL video are 640x480 at 15fps. I want to see 1280x720x24 or 1280x720x25fps, because that is the minimum of what HD means to everybody else. 1200 kbps is what was promised.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 23:11

the sample in the first post is even blocky and flashing...

You must be using a decoder with a IDCT that is mathematically weak. I look through our samples with our analysis tool before posting, using both 32-bit floating point and 16.16 fixed point IDCT, and there are no blocks.

If you want to make statements about the quality, you should qualify them with the specs of your playback system. In this case, what is the name of the IDCT algorithm and what is the mathematical precision of it, fixed point or floating point. I will say that the MPEG-1 IDCT in Windows Media Player is quite awful.

MPEG IDCT analysis (http://www.digigami.com/mpressionist/blog/2004_11_18_archives.html).

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

Sharktooth

23rd December 2005, 23:11

it's going on... but take into consideration i have no lossless material and im encoding one of your already encoded samples that shows blocks and banding effects...
the IDCT is the reference IDCT.

digigami

23rd December 2005, 23:13

your already encoded samples that shows blocks and banding effects...

If you're talking about Mako Shark Club, the banding effects are part of the source material.

Tell me what the linear frame#s are that have block artifacts and I will look at them.

Are you talking about a sample from here:
HD MPEG-2 VBR (http://www.digigami.com/megapeg/samples.php)

I find it strange that you are only interested in encoding from LOSSLESS sources but do not have any to work from.

--------------------------------------------------
Digigami
MegaPEG MPEG-2 Encoder (http://www.digigami.com/megapeg/) - people disagree as to whether it's any good.
MPressionist MPEG-2 Analyzer (http://www.digigami.com/mpressionist/) - no disagreements with this one, everybody likes it.

digigami

23rd December 2005, 23:17

Just a question... do you know AVC supports 8x8 dct like mpeg2 too? (it's in High Profile...)

Of course it does. It has the most advanced feature set around.
8x8 DCT goes back into the 70s.

Sharktooth

23rd December 2005, 23:19

im talking about the sample in the first post the marko shark club sample and all the other encodes.
a magnified example...
http://www.webalice.it/f.corriga/temp/blocks_and_rings.png
blocks, rings, artifacts in all the scenes...
maybe you need to check your encodes with a good LCD monitor before posting them...

I find it strange that you are only interested in encoding from LOSSLESS sources but do not have any to work from.Licensing problems.

Sagittaire

23rd December 2005, 23:26

Here sample H264 High Profil

King Kong (http://multimediacom.free.fr/Download/x264HP_KingKong_720p_1500.mp4)
1280*720*25
1500 Kbps

Harry potter IV (http://multimediacom.free.fr/Download/x264HP_HPIV_720p_1250.mp4)
1280*720*25
1250 Kbps

Ice Age 2 (http://multimediacom.free.fr/Download/x264HP_IceAge_720p_1250.mp4)
1280*720*25
1250 Kbps

All the QT AVC are uncompressible trailer (like generaly all trailer)
You can make that with all trailers (720p 1-1.5 Mbps and 1080p 1.5-2 Mbps)

I can make the same MPEG2 encoding in 1080p with libavcodec with 5 Mbps I think (same metric). IMO libavcodec is the best MPEG2 encoder. I think that if you can prove that your encoder is better than libavcodec could be a good publicity for your encoder ...

Sharktooth

23rd December 2005, 23:30

i can hear the silence...

digigami

23rd December 2005, 23:34

it's going on... but take into consideration i have no lossless material and im encoding one of your already encoded samples that shows blocks and banding effects...
the IDCT is the reference IDCT.

OK I'm going to assume 32bit floating point. Can you send me the frame #s where you are seeing the block artifacts? I'd like to check them here with my MPEG-2 IDCT analyzer (http://www.digigami.com/mpressionist/screenshots_hdtv.php). I can inspect the IDCT coefficients and look at what has happened to the block during compression and we can discuss it.

Maybe you can do a frame grab and use photoshop to encircle the blocks and that way I will know where to look.

Sharktooth

23rd December 2005, 23:36

blocking is EVERYWHERE in EVERY SAMPLE...
however look at the sagittaire's post... and see the HD at low bitrates MPEG2 can only dream of...

digigami

23rd December 2005, 23:39

All the QT AVC are uncompressible trailer (like generaly all trailer)
You can make than with all trailers with 720p 1-1.5 Mbps and with 1080p 1.5-2 Mbps I think ... ;-)

Mon francais est terrible. Mais... Your english is much better. But I did not understand what you said.

I can make the same MPEG2 encoding in 1080p with libavcodec with 5 Mbps I think (same metric). IMO libavcodec is the best MPEG2 encoder. I think that if you can prove that your encoder is better than libavcodec could be a good publicity for your encoder ...

Ok, did you mean that your examples are created from the sources on the Apple H.264 site?

Also, what do I need to play them? Minimum CPU and playback.

digigami

23rd December 2005, 23:44

im talking about the sample in the first post the marko shark club sample and all the other encodes.
a magnified example...

Thanks. Can you give me the frame#s please. Also, is this grab from a I, P or B-frame.

Gen

Sharktooth

23rd December 2005, 23:46

a good CPU or a Ati x1000 series videocard (soon even nvidia video card will be able to play back h.264 almost totally in hardware) installed.
mplayer, VLC or any other player based capable to decode FULL h.264 features.
however... i did some calculation and we can put 8 hours of 1280x720@25fps in a standard 4.3GB DVD...
We can place all the Lord Of The Rings EE in High Definition + extras in 2 DVDs...

i can smell something like 12 (out of 14) DVD less in the package, lowering the distribution cost just a bit...

Are we still talking of advantages?

digigami

23rd December 2005, 23:47

blocks, rings, artifacts in all the scenes...
maybe you need to check your encodes with a good LCD monitor before posting them...

Ok, now recommend which H.264 encoder I can use to encode this same clip, and what settings I should use. The source material is 1080p uncompressed. I have it as a series of .tga files and also as a Pixlet .MOV file on maximum/lossless quality. So your H.264 encoder would need to read one of those or I can convert it to an AVI.

digigami

23rd December 2005, 23:53

a good CPU or a Ati x1000 series videocard (soon even nvidia video card will be able to play back h.264 almost totally in hardware) installed.

Thanks. Still waiting for the frame#s of your complaints. If you don't have the #s, I'll take the frame times. HH:MM:SS:ff

OK?

Sharktooth

23rd December 2005, 23:55

Ok, now recommend which H.264 encoder I can use to encode this same clip, and what settings I should use. The source material is 1080p uncompressed. I have it as a series of .tga files and also as a Pixlet .MOV file on maximum/lossless quality. So your H.264 encoder would need to read one of those or I can convert it to an AVI.
x264... obviously...
head to the AVC forum and you'll find there all the info you need... but i guess you better get a PC or you wont be able to use the most poverfull frameserver application in the world: avisynth.
however, ever tried ffmpeg as mpeg2 encoder? im sure it works better than yours...
Thanks. Still waiting for the frame#s of your complaints. If you don't have the #s, I'll take the frame times. HH:MM:SS:ff

OK?
I'm not complaining about ONE frame.. but ALL frames!!! They all have clearly visible and annoying artifacts

digigami

23rd December 2005, 23:55

I can make the same MPEG2 encoding in 1080p with libavcodec with 5 Mbps I think (same metric).

If you can do this, then we agree that MPEG-2 VBR can do HD at standard def DVD bitrates. If you read our press release, that is exactly what we said.

digigami

23rd December 2005, 23:57

What PC Software can read these H.264 files and write them out as uncompressed AVI? I want to compare them to the Apple sources.

Sagittaire

23rd December 2005, 23:59

Ok, did you mean that your examples are created from the sources on the Apple H.264 site?

yes ... there are very good source because 8-10 Mbps is high bitrate for 1080p H264 AVC.

Also, what do I need to play them? Minimum CPU and playback.

My sempron 2.1 ghz can play my encoding but for HD-DVD/BR CPU playback it's not important ...

Sharktooth

24th December 2005, 00:05

What PC Software can read these H.264 files and write them out as uncompressed AVI? I want to compare them to the Apple sources.
... you definatly need a PC ...
install a h.264 directshow decoder (ffdshow, nero, etc...), install avisynth, create a script to load the .mp4, load the .avs into virtual dub and save uncompressed frames into avi.
or use some other opensource software to do the job as well on Mac or Linux.

digigami

24th December 2005, 00:07

digigami

24th December 2005, 00:08

... you definatly need a PC ...
install a h.264 directshow decoder, install avisynth, load the .avs into virtual dub and save uncompressed frames into avi.
or use some other opensource software to do the job as well on Mac or Linux.

What's a good H.264 direct show decoder that will render H.264 in the highest quality, rather than the fastest playback?

BTW, the IDCT on the video cards is not 32bit float reference, unless you are thinking of a particular card.

Sharktooth

24th December 2005, 00:10

What's a good H.264 direct show decoder that will render H.264 in the highest quality, rather than the fastest playback?

BTW, the IDCT on the video cards is not 32bit float reference, unless you are thinking of a particular card.
every h.264 decoders (except the crippled ones like QT) will decode at the same quality.
however you can get ffdshow for free here: http://mirror01.x264.nl/x264/ffdshow-20051221-gcc4.0.2-sse-x264.nl.exe

digigami

24th December 2005, 00:12

I'm not complaining about ONE frame.. but ALL frames!!! They all have clearly visible and annoying artifacts

Yes I heard you. What playback software (Build#) are you using to audition the files?

digigami

24th December 2005, 00:14

every h.264 decoders (except the crippled ones like QT) will decode at the same quality.

What aspects of H.264 AVC are not implemented by QuickTime. If you know the exact features. BTW, if you were to encode H.264 using x264 in a way that QuickTime will play, what settings do you target?

Thanks.

Sharktooth

24th December 2005, 00:15

Yes I heard you. What playback software (Build#) are you using to audition the files?
tried with quicktime, MS Windows MPEG Decoder, libavcodec, dgmpeg avisynth mpeg plugin, Media Player Classic mpeg decoder...
but the artifacts are always there...

Sharktooth

24th December 2005, 00:17

What aspects of H.264 AVC are not implemented by QuickTime. If you know the exact features. BTW, if you were to encode H.264 using x264 in a way that QuickTime will play, what settings do you target?

Thanks.
Doing that will seriously cripple the x264 encoder capabilities since QT simply is not compatible with the major h.264 features.
I already created some profiles for QuickTime compatibility but, compression and quality are seriously compromised.

digigami

24th December 2005, 00:17

I'm not complaining about ONE frame.. but ALL frames!!! They all have clearly visible and annoying artifacts

In comparing what you describe to what I see on my Apple Cinema using both MPressionist 32-bit float and 16.16 fixed point IDCT, I would say that your MPEG-1/MPEG-2 playback is just as crippled as you describe QT to be.

Windows Media Player has horrible IDCT.

VideoLAN on the Mac seems to be doing a good job. I'll have to look at it on my PC.

digigami

24th December 2005, 00:18

I already created some profiles for QuickTime compatibility but, compression and quality are seriously compromised.

That's cool, I am interested for other reasons. Where do you describe your QuickTime profiles.

Also, can you tell me what H.264 AVC features are not implemented by QT?

Sharktooth

24th December 2005, 00:19

In comparing what you describe to what I see on my Apple Cinema using both MPressionist 32-bit float and 16.16 fixed point IDCT, I would say that your MPEG-1/MPEG-2 playback is just as crippled as you describe QT to be.

Windows Media Player has horrible IDCT.

VideoLAN on the Mac seems to be doing a good job. I'll have to look at it on my PC.
I didnt use Windows Media Player... but Media Player Classic (it's open source and not by Microsoft).
However be aware, MACs have a higher Gamma than PC with Windows or Linux and that tends to "mask" the artifacts.
That's cool, I am interested for other reasons. Where do you describe your QuickTime profiles.

Also, can you tell me what H.264 AVC features are not implemented by QT?
The profiles are included in my x264 builds for windows (the sticky thread in the AVC forum) and there's also a discussion in the AVC forum.
the unsupported features are the whole "high profile" (8x8 dct, custom scale matrices, 8x8 intra blocks, lossless encoding/decoding...), mixed reference frames, multiple b-frames, weighted prediction and some other stuff i dont recall.
However i have to go now. You can reach me on irc (#x264 on irc.freenode.net) or MSN or this forum.

digigami

24th December 2005, 00:23

frameserver application in the world: avisynth.
however, ever tried ffmpeg as mpeg2 encoder? im sure it works better than yours...

Yes avisynth is a nice little utility. I read the sources about 7 years ago.

Regarding ffmpeg, I hope they fixed the bugs in the quantizer. They had some serious problems in the logic of handling clipped DCT values.

digigami

24th December 2005, 00:31

However be aware, MACs have a higher Gamma than PC with Windows or Linux and that tends to "mask" the artifacts.

I understand what you're saying, but if it is a IDCT mismatch problem then the 'artifacts' you are seeing is the result of the frame being reconstructed poorly. When you go for low bitrates the situation is exacerbated. If the DC coefficient in a motion compensated non-intra block is off by one during the deconstruction, the visual effect is quite large.

Did you read my blog entry about IDCT precision in MPEG-2 DVDs? Even well-encoded DVD discs exhibit the kind of blocking you are describing when auditioned on poor DCT. Switch to the reference 32-bit float and it looks perfect.

However i have to go now.
Thanks for all your help. I want to learn about H.264 and I think I came to the right place! I will definitely be checking out your x264.

Sharktooth

24th December 2005, 00:34

CruNcher

24th December 2005, 00:36

digigami

24th December 2005, 00:40

Digigami i think you are on OSX ;) so you most probably use an outdated version of Videolan (you have to keep up with development) strange that you don't know that as a developer of video solutions anyway try this http://nightlies.videolan.org/build/macosx/trunk-20051223-0114.dmg

OK thanks. I like to try and keep up, but if you look at how much software I am maintaining (GUI + codec) I sometimes fall behind. Also, after writing altivec for everything I now have to port everything to SSE2.

BTW, the disc image did not mount.

digigami

24th December 2005, 00:41

digigami

24th December 2005, 01:04

yes ... there are very good source because 8-10 Mbps is high bitrate for 1080p H264 AVC.

Yes, I think if you consider that Apple's audience is broadcast professionals, they have chosen to use a higher bitrate for those reasons.

BTW, can you encode the Ice 1080p trailer with your favorite MPEG-2 encoder (5 mbits/s is what you said) and post it along with the settings/config used?

Thanks.

digigami

24th December 2005, 01:13

last thing...
i was forgetting to post the encode :)
http://www.webalice.it/f.corriga/temp/mako.mp4
it was already of poor quality (noise, banding, blocks and other artifacts) but i applied no filters... encoded at 1200kbps without any special settings.

I just looked at the sample you are referring to on my Dell XPS with VideoLAN 7.2 and Windows Media Player. In WiMP, there are blocks. In VideoLAN it looks exactly like it does on my Mac, which looks great to my eyes. I think your MPEG-1/2 decoding chain could be improved.

Gen

digigami

24th December 2005, 01:23

Harry potter IV (http://multimediacom.free.fr/Download/x264HP_HPIV_720p_1250.mp4)
1280*720*25
1250 Kbps

Harry Potter is nice, but there are some blocks, but they are well hidden in darker areas and not in the region of interest, which is exactly what MPEG encoders are supposed to do.

I auditioned the file with VLC 0.8.4 on the Mac.

You tell me where the blocks are and we'll compare notes.

CruNcher

24th December 2005, 01:33

yeah i can understand that the MacTel stuff make you problems now, must be hard times for you @ the moment :) but you can't hide from the fact that the only future Mpeg-2 has for now is Blue-Ray HD-DVD decided for compression Sony didn't and we aren't have the problems in terms of SOCs that we had with Mpeg-4 SP/ASP it won't give different Generations of Chips all Chips that come out support High Profile (Frext). http://www.pacemicro.com/Corporate/documents/specs/DS810_spec.pdf
see this it's the same Level that also HD-DVD supports, everything will be interoperable not the mess we had/have with Mpeg-4 SP/ASP at the time.

And if we test stuff i would say lets test uncompressed stuff im no fan of testing stuff that was allready compressed

ftp://ftp.ldv.e-technik.tu-muenchen.de/pub/test_sequences/1080p/tractor.yuv

im for this sample 1080p @ 25 fps sequence :)

digigami

24th December 2005, 01:57

but you can't hide from the fact that the only future Mpeg-2 has for now is Blue-Ray

Terrestrial HDTV in the US is based on MPEG-2. I'm not worried. Alot of what we are doing is not specific to MPEG. BTW, there are more MPEG-2 oriented devices hitting the market now than ever.

Also, what does your PSNR Lab software do?

Gen

digigami

24th December 2005, 02:05

HD-DVD decided for compression

Actually HD-DVD does support MPEG-2 VBR and just as wide a range of frame sizes as standard DVD. Believe it or not, it also supports QCIF encoding.

digigami

24th December 2005, 02:36

... you definatly need a PC ...
install a h.264 directshow decoder (ffdshow, nero, etc...), install avisynth, create a script to load the .mp4, load the .avs into virtual dub and save uncompressed frames into avi.
or use some other opensource software to do the job as well on Mac or Linux.

Thanks sharktooth.

JohnV

24th December 2005, 02:48

Windows Media Player has horrible IDCT.

VideoLAN on the Mac seems to be doing a good job. I'll have to look at it on my PC.I can confirm this regarding your Least_Likely_PSA.mpeg clip.
VideoLan and Nero ShowTime play this ok. With Windows Media Player it looks much more blocky and there's like noise pumping.

But I'm still waiting your MPEG-2 HD clips with comparable bitrates to the AVC clips provided in this thread. ;)

digigami

24th December 2005, 03:02

I can confirm this regarding your Least_Likely_PSA.mpeg clip.
VideoLan and Nero ShowTime play this ok. With Windows Media Player it looks much more blocky and there's like noise pumping.

But I'm still waiting your MPEG-2 HD clips with comparable bitrates to the AVC clips provided in this thread. ;)

Hi John,

I think we're going to discover that our HD clips are higher bitrate but in the range of the bitrate for standard DVD, which was the point of our press release. A viable alternative for Blu-Ray and HD-DVD is what we were describing.

I'm looking at an MPEG-1 VBR of the Ice Age trailer right now and it is around 5mbits/s and looks great, but I've just started.

I'm interested in learning more about H.264 and this seems like the place to do it.

Gen

JohnV

24th December 2005, 03:18

Well.. what I wonder in your press releases are sentences like "In our tests here at Digigami, we find that our MPEG-2 encoder is actually outperforming H.264 by a wide margin on 720p/1080p film content. Typically, our HD MPEG-2 encoder can produce VBR files two thirds to one half the bitrate produced by current H.264 encoders."

But turned out you have only used one of the worst AVC encoder QT AVC, and you have no real experience on AVC encoding and good AVC encoders.
This is where your press release goes imo more to the marketing side..

CruNcher

24th December 2005, 03:23

Actually HD-DVD does support MPEG-2 VBR and just as wide a range of frame sizes as standard DVD. Believe it or not, it also supports QCIF encoding.

Sure it does for Backward compatibility reasons, but do you really think Hollywood gonna use Mpeg-2 for HD-DVDs ? i doub't it they gonna use VC-1 and H.264

Terrestrial HDTV in the US is based on MPEG-2. I'm not worried. Alot of what we are doing is not specific to MPEG. BTW, there are more MPEG-2 oriented devices hitting the market now than ever.

Yes for the US thats still right but as we started in Europe later with HDTV we decided for H.264 and all HD Broadcasters gonna use it in the future here and im sure some US and Japanese Broadcasters will also jump onto it because of the bandwith save they can achive with it (Bandwith = Money).

Also, what does your PSNR Lab software do?

PSNR Lab was used by me in the XviD Development process to find problems with different tools it's a simple batch file that does automatic psnr testing of many tools and tool combinations in XviD but it's outdated now as many more tools became available that are not anymore covered.

and i have to agree with JohnV your press release sounds to harsh

bobololo

24th December 2005, 04:59

This is how I compare things:
http://www.digigami.com/mpressionist/screen.php?screen=comparesplit

I wrote this tool specifically so I can have these kinds of discussions without the dialogue descending into mud-slinging and religious fervour.

Also, can compare pixel residue (actual per-pixel frame differences) which is useful for picking up differences too fine for the eyes to see.

How do you compare?

A very rough outlook can be drawn using psnr. If the rate-distorsion curves show 2 dB of difference for two codecs based on the same process scheme (dct or dct-like), then we can reliably deduce which is the best. For finer analysis, visual assessments is imho the way to go.

Anyway, the 6 pages posted within the latest few hours gave me a better understanding of your approach and I would re-formulate it as follow (please correct me if I'm wrong):

For a specific class of HD contents (mostly films) and when operating at the higher bound of HD-DVD/BR bitrate specifications, MPEG-2 can achieve comparable results to H.264.

Basically, even I don't fully agree with this claim, I think it's quite defensible when expressed in that form. However, as JohnV pointed out, your release rather tends to emphasize the superiority of your MPEG-2 over H.264 and this is the point where people strongly disagree. Here is the first sentence of your PR:

Long Beach, CA - December 22, 2005 · Digigami today announced its new MegaPEG HDTV VBR MPEG-2 encoder is capable of matching and, in some cases, exceeding the picture quality while at the same time offering reduced bitrates compared to current H.264 encoders.

Anyway, if we put aside the lower coding efficiency of MPEG-2 vs H.264 that many people are trying to demonstrate to you, and if we get back to your actual concept, I still find a major weakness. As you may know, the HD-DVD/BR specifications set a maximum bitrate for two main purposes. First it insures enough bitrate so that any kind of contents and specificaly the hardest ones (interlaced, noisy, etc) could be reproduced with a good quality. And secondly, it defines the performance requirements on the decoder side.

That specification doesn't imply that all contents is tied to be encoded using the higher bitrate bound. And practically, there are many chances that most contents will be encoded at a lower bitrate. This has been demonstrated for long with DVDs. Even if the max allowable bitrate is set to 9 mb/s, I guess you will agree that most of DVDs don't exploit such bitrate and are rather usually encoded at 5-6 mb/s.

To me the priviledged policy is to squeeze the video as much as possible and I don't see why this would change for HD-DVD/BR. And on that game, I don't see how MPEG-2 could win against H.264 and I'm not doing some H.264 marketing ;)

Finally, I do agree with you when you say that MPEG-1/2 requires less resources than H.264. So do SD against HD. This basically applies to any technology evolutions (faster CPU, wider screen, etc.). Technology improvement has a cost, it needs you to keep going forward with it. You can always decide not to follow it, but in such case don't expect to benefit from its advance.

digigami

24th December 2005, 06:39

Long Beach, CA - December 22, 2005 · Digigami today announced its new MegaPEG HDTV VBR MPEG-2 encoder....[/b].

Yes, well, if you don't write something good in the first line, everybody will ignore your PR ;)

Other than that, I agree with you. Although, in my experience having a high bitrate ceiling allows you to code the most difficult passages without loss, and in the easier sections the bitrate can fall quite low.

BTW, setting aside MPEG-4 Part 10 for the moment, what is the best MPEG-4 simple profile encoder out there?

digigami

24th December 2005, 06:47

But turned out you have only used one of the worst AVC encoder QT AVC, and you have no real experience on AVC encoding and good AVC encoders. This is where your press release goes imo more to the marketing side..

True, it has some marketing in it, but as I read more of the posts on this forum it is clear that some of the more important aspects of H.264 _standards_ work like, for instance, interoperability, conformance and so forth, are still in progress. Like having to go get a new build of VLC just to play a movie. So one could argue that the files given to me by Sagattaire are part of a work in progress, and that x264 is an emerging encoder.

H.264 is not a standard unless everyone's files play on every one else's decoders (assuming the profile is supported by the decoder). That's the hard part of being a part of a standards community. You can complain about Apple all you want, but if you're not diplomatic about interoperating with QuickTime and clear about what their limitations are, then you're not part of a standards effort.

Would you agree?

digigami

24th December 2005, 07:32

But turned out you have only used one of the worst AVC encoder QT AVC, and you have no real experience on AVC encoding and good AVC encoders.

x264 - a free h264/avc encoder
x264 is a free library for encoding H264/AVC video streams. The code is written from scratch by Laurent Aimar, Loren Merritt, Eric Petit (OS X), Min Chen (vfw/asm), Justin Clay (vfw), Måns Rullgård, Christian Heine (asm), Alex Izvorski (asm), and Alex Wright. It is released under the terms of the GPL license.
Development status: BIG FAT WARNING: x264 is still in early development stage

OK, so if QT is the worst AVC and x264 is very early development stage, can you make me a Nero-encoded H.264 AVC of the Ice Age trailer? I am curious to know what the shipping, robust H.264 encoders that are not made by Apple can do.

Also, what methods do you H.264 developers employ to make sure your streams are compliant and interoperable with other H.264 vendors? Is there matrix/table/chart somewhere of the available H.264 decoder implementations and what profiles they support?

Sagittaire

24th December 2005, 10:03

OK, so if QT is the worst AVC and x264 is very early development stage, can you make me a Nero-encoded H.264 AVC of the Ice Age trailer? I am curious to know what the shipping, robust H.264 encoders that are not made by Apple can do.

As I say I make HD test and Ateme High Profil encoder (Nero use only Ateme Main Profil encoder) done comparable visual result (IMO little better). It's very difficult to make test with QT7 because it's not friendly encoder (too slow for example)

In fact difference between H264 and MPEG2 is exponential : If you use less bitrate the difference is more important.

digigami

24th December 2005, 10:35

Ok AVC pros, I have updated my non-volatile ROM with what you have shown me today:
http://www.digigami.com/megapeg/blog/

And I hope you don't mind if I hang out here trying to learn what is really going on in H.264 AVC.

digigami

24th December 2005, 10:41

IMO digigami encoder isn't better than for libavcodec for MPEG2 ... lol

Sagittaire - which version of our codec did you test, or are you reflecting on the samples on our web site:
http://www.digigami.com/megapeg/samples.php

If you would like to do more tests, I would like to assist.

digigami

24th December 2005, 10:43

As I say I make HD test and Ateme High Profil encoder (Nero use only Ateme Main Profil encoder) done comparable visual result (IMO little better). It's very difficult to make test with QT7 because it's not friendly encoder (too slow for example)

In fact difference between H264 and MPEG2 is exponential : If you use less bitrate the difference is more important.

OK, so currently, the only shipping, robust encoder that doesn't come with a warning that is much better than MPEG-2 VBR is Ateme using High Profile?

What is the best that Main Profile can do? Can you re-do your tests (movie trailers) with Ateme Main Profil?

digigami

24th December 2005, 10:45

But turned out you have only used one of the worst AVC encoder QT AVC, and you have no real experience on AVC encoding and good AVC encoders. This is where your press release goes imo more to the marketing side..

Ok, JohnV, with your help and Sagittaire I have narrowed my search down to H.264 Main Profile. Can you re-encode Ice Age trailer with Nero/Ateme Main Profil encoder 1280x720p for comparison? I want to know the difference between H.264 High-Profile and Main Profile, and also want to know the limitations of current QuickTime H.264 encoder.

Xayd

24th December 2005, 10:59

If you go and look at the H.264 chipsets that they are considering for HD-DVD and Blu-Ray, I'd be surprised if they were using H.264 features that aggressively.

i dunno, per a comment from a Microsoft rep about the situation when the techs for HD-DVD and BluRay were being discussed they tried to impress upon Hollywood that new MPEG-4 codecs would require quite a bit of new (and expensive) hardware, but they didn't seem to care about that and declared the codec specs without any regard for the hardware required.

then consider the poor implementation of, say, the HDMI interface as an example, and put two and two together.

never underestimate the potential of people to do stupid things, imo.

Sagittaire

24th December 2005, 12:19

Ateme Main Profil (Nero)
http://multimediacom.free.fr/Download/NDAVCMP_IceAge_720p_1250.mp4

Ateme High Profil
http://multimediacom.free.fr/Download/NDAVCHP_IceAge_720p_1250.mp4

x264 High Profil
http://multimediacom.free.fr/Download/x264MP_IceAge_720p_1250.mp4

x264 Main Profil
http://multimediacom.free.fr/Download/x264MP_IceAge_720p_1250.mp4

XviD MPEG4 ASP
http://multimediacom.free.fr/Download/XviD_IceAge_720p_1250.avi

XivD is MPEG4 ASP codec and is better and by far than all MPEG2 codec. XviD is here like 8*8 DCT codec reference. As you can see DCT isn't the only advantage for H264 ...

JohnV

24th December 2005, 14:27

OK, so currently, the only shipping, robust encoder that doesn't come with a warning that is much better than MPEG-2 VBR is Ateme using High Profile?

No, Sagittaire above already posted streams. And even though I don't represent x264, I'd say it's not in early development stage quality wise. It is a quite mature h264 encoder. Of course since there becomes constantly new development builds of x264 available, there can be sometimes bugs in it (you should follow Doom9 forum), but quality wise it's a mature encoder competing with Ateme/NeroDigital.

The integration of the latest Ateme AVC codec version with High Profile support to NeroDigital has had delays. I was expecting we put it in this year, but it goes to 2006. More info in January.

digigami

24th December 2005, 14:39

No, Sagittaire above already posted streams.

So Nero is only currently shipping Main Profile, you have complaints with Apple, and you won't post streams created by your own encoder?

I think we maybe disagree slightly on what it means to conform to an ISO standard and be interoperable, but I am new here so I will watch and see what is happening. It would be quite useful to know the difference between high profile and main profile.

JohnV

24th December 2005, 14:52

So Nero is only currently shipping Main Profile, you have complaints with Apple, and you won't post streams created by your own encoder?
Umm, Sagittaire already posted that stream few messages above this. Nero Digital is a co-operation between Nero and Ateme. Ateme provides the video encoders (ASP and AVC) to Nero Digital. Nero provides audio codecs and video decoders for Nero Digital.

Sagittaire

24th December 2005, 15:14

Moreover quality differencies between Main and High profil are not considerable (hard to see with medium/high bitrate). If you want make very good encoding with H264 MP try Nero Recode ... it's a very friendly and simple gui. In fact HP advantages are not IMO quality but fonctionality:

Main vs High Profil

- full 8*8 dct for HP
- custom matrix for HP
- lossless mode for HP
- full YUV space mode for HP

Sharktooth

24th December 2005, 16:13

Hi John,

I think we're going to discover that our HD clips are higher bitrate but in the range of the bitrate for standard DVD, which was the point of our press release. A viable alternative for Blu-Ray and HD-DVD is what we were describing.

I'm looking at an MPEG-1 VBR of the Ice Age trailer right now and it is around 5mbits/s and looks great, but I've just started.

I'm interested in learning more about H.264 and this seems like the place to do it.

Gen
If you're interested in h.264, head over the h.264/AVC board: http://forum.doom9.org/forumdisplay.php?f=77

However, your statements on your website are false.
You cant fit a 4 hours movie in HD res with a good quality in a standard DVD, while h.264 can... also mpeg-2 cant in any case "beat" h.264 in efficiency and compression and consequently in quality...

diogen

24th December 2005, 17:38

Since Ben Waggoneer was already quoted once in this thread,
I'll post another of his quotes (taken from AVSforum)
********************************************
I've checked out the Digigami package, and I cover it in a forthcoming DV magazine article.

Basically, I don't think their results are real-world applicable. Their encoder is about 60x slower than the next slowest Mac encoder, and lacks a true bitrate limited 2-pass VBR mode (you can pick peak and min data rate, but have to tune in average data rate via). And even with those caveats, in double-blind evaluation, for SD DVD MPEG-2, it came out as one of the lower-rated MPEG-2 encoders for quality at a given bitrate.

Gen's a smart guy, but I think his passion for his product has outstripped its real-world strengths.
*********************************************
Taken from here
http://www.avsforum.com/avs-vb/showthread.php?t=411600&page=600&pp=30

On the same thread a couple hundred pages earlier Stacy Spears stated that
any encoder with 1/10th real time encoding speed (and slower) is of no interest to studios. I think Apple's encoder is around that speed; that will make Digigami some 1/500th real time or so.

Diogen.

digigami

24th December 2005, 18:16

mpeg-2 cant in any case "beat" h.264 in efficiency and compression and consequently in quality...

Which of the current H.264 AVC encoders support coding interlaced video?
According to x264.nl, it does not. So what do you do when you come to breaks in the telecine and you have 3 or 4 interlaced frames?

Anyone got a sample?

digigami

24th December 2005, 18:18

Since Ben Waggoneer was already quoted once in this thread,
I'll post another of his quotes (taken from AVSforum)
********************************************
I've checked out the Digigami package, and I cover it in a forthcoming DV magazine article.

Basically, I don't think their results are real-world applicable. Their encoder is about 60x slower than the next slowest Mac encoder, and lacks a true bitrate limited 2-pass VBR mode (you can pick peak and min data rate, but have to tune in average data rate via). And even with those caveats, in double-blind evaluation, for SD DVD MPEG-2, it came out as one of the lower-rated MPEG-2 encoders for quality at a given bitrate.

Gen's a smart guy, but I think his passion for his product has outstripped its real-world strengths.
*********************************************
Taken from here
http://www.avsforum.com/avs-vb/showthread.php?t=411600&page=600&pp=30

On the same thread a couple hundred pages earlier Stacy Spears stated that any encoder with 1/10th real time encoding speed (and slower) is of no interest to studios. I think Apple's encoder is around that speed; that will make Digigami some 1/500th real time or so.

Diogen.

You can read that here if you want.
http://www.digigami.com/megapeg/blog/

IgorC

24th December 2005, 18:31

Please , don't start to mix items. first quality, then interlace stuff, other ...

There were clear samples of H.264 and Xvid. And we don't see anysample of MPEG2.

I've seen some MPEG2 HDTV samples from digigami . They look HORRIBLE.
Croma offset. All colors are washed out. It's slow. some samples has 20 fps framerate :sly: . Please. MPEG-2 will never NEVER be on par with H.264.
Oh dear, QT7 AVC is the shame on H.264 family codecs.

JohnV

24th December 2005, 18:37

Which of the current H.264 AVC encoders support coding interlaced video?At least Ateme does.

Sharktooth

24th December 2005, 18:48

digigami

24th December 2005, 18:52

Yeah ateme does.
but interlacing was made for the analog worls... it's truly unnecessary with HD TVs.
The only reason 1080 res is interlaced is because the DVI interface bandwidth is limited (and 1080p requires 2 interfaces) but the new standards (HDMI) arent affected by that problems and 1080p is possible.
So, in any case, if you get interlaced sources, just deinterlace them (the right way) and encode to h.264.

Tell that to all the people buying HDV and DVCPROHD cameras!

The reason for interlacing is increased resolution in the time domain at the expense of vertical spatial resolution. Always was, always will be. Which is why we have 1280x720x59.94p and 1280x720x50p. The latter trade even less spatial resolution for more temporal resolution.

digigami

24th December 2005, 18:54

They look HORRIBLE. Croma offset. All colors are washed out. It's slow. some samples has 20 fps framerate :sly: . Please. MPEG-2 will never NEVER be on par with H.264.

Hope you read the thread. What playback system are you auditioning these files on? Use VideoLAN.

20fps. This is quite funny. MPEG-1 and MPEG-2 start at 23.976 and go up.

digigami

24th December 2005, 18:55

At least Ateme does.

OK, so it is starting to sound as if Ateme is the only robust H.264 AVC encoder in town.

Are there any other H.264 AVC encoders that support interlacing?

bobololo

24th December 2005, 19:05

I finally managed to download your commandN_spotLIGHT clip (what a pain to retrieve 550 MB at a transfert rate of 3.5 kB/s) and made an encode using ateme mp encoder. I uploaded the result here:

ftp://mood.ateme.com/digigami/commandN_spotLIGHT_HD_720p-ateme_avc_mp_1500k.mp4

To me the source is quite irrevelant for such kind of test. The content is very static (an interview with 2 persons speaking in front of the cam with nearly no motion). The source quality is rather poor: I can see smearing on the man's clothes and it's very noisy.

The resulting avc stream features:

main profile
bitrate of 1500 kb/s. So it's less than half of your mpeg-1 which is 3200 kb/s.
2pass rcmode
2 references
extra quality mode
psycholevel enhancement enabled

So you can easily reproduce a similar result with Nero Recode 2 using the above settings.

According to my visual feeling, I prefer the avc version since it's sharper and more stable. Your opinion is welcome of course.

ps: I can host the source QT7 H.264 file and your mpeg-1 encode on our server if it can help since yours are totally overloaded.

JohnV

24th December 2005, 19:06

OK, so it is starting to sound as if Ateme is the only robust H.264 AVC encoder in town.

Are there any other H.264 AVC encoders that support interlacing?
Well, at least Elecard AVC encoder seems to have also, though according to Doom9 test, it's not quite up to Ateme and x264 quality, but not bad like QT.

Sharktooth

24th December 2005, 19:10

OK, so it is starting to sound as if Ateme is the only robust H.264 AVC encoder in town.

Are there any other H.264 AVC encoders that support interlacing?
Here's a list of AVC encoders... commercial and free solutions:
http://forum.doom9.org/showthread.php?t=95939
get them and try them...
there you can find also Hardware solutions...

digigami

24th December 2005, 19:23

It's just a little challenge only for MPEG2 encoder (libavcodec, TMPGEnc, CCE, Procoder, HCEnc, QEnc).
First I think That you must prove that your MPEG2 encoder is better and by far than the other encoder.
My source is an HPII DVD trailler at 5 Mbps.

Sagittaire. Hi, I am down to around 2.2 mbits/s with a semi-constant Q. I would like to see your top two winning .m2v files if I could. The links are there, but there are no files.

JohnV

24th December 2005, 19:30

Sharktooth

24th December 2005, 19:35

some "artifact-noise" on the wall behind the guy but that's why bobololo used psy3 and the source is of poor quality... without psy3 i guess it will be almost perfect.
as noted in the ateme quality feedback thread psy2 is much better than psy3.

digigami

24th December 2005, 19:40

Yes, it's definitely sharper and better quality than Digigami's over twice the bitrate MPEG-1. This AVC encode looks like a HD encode, Digigami's over twice the bitrate MPEG1 encode is blurry (yes with VideoLan) and doesn't really even look like High Def in my opinion.
Interesting to hear Digigami's comments about this..

Yes, it would have been nice to start with uncompressed source. BTW, so you are 2x better than a 1990 codec? Very nice.

BTW, I think our motion estimator was broken in that build. I fixed a bug last night which I think was causing a lot of bad mv's.

Right now, on the Ice Age (Kung Fu), my MPEG-1 is around 3.3 mbits/s. The picture is sharp and, of course, the source is much better.

It would be nice to see a H.264 AVC version of Ice Age (Kung Fu) 1280x720x25p which _does_not_ have blocking artifacts in the darker background regions. If you don't know what I'm talking about, don't bother, but if you can see them, I'd like to see at what bitrate it hits the quality expectation for commercial concerns, rather than ripping and archiving.

Sharktooth

24th December 2005, 19:46

Yes, it would have been nice to start with uncompressed source. BTW, so you are 2x better than a 1990 codec? Very nice.

BTW, I think our motion estimator was broken in that build. I fixed a bug last night which I think was causing a lot of bad mv's.

Right now, on the Ice Age (Kung Fu), my MPEG-1 is around 3.3 mbits/s. The picture is sharp and, of course, the source is much better.

It would be nice to see a H.264 AVC version of Ice Age (Kung Fu) 1280x720x25p which _does_not_ have blocking artifacts in the darker background regions. If you don't know what I'm talking about, don't bother, but if you can see them, I'd like to see at what bitrate it hits the quality expectation for commercial concerns, rather than ripping and archiving.
a bit more than 2x... as you saw that with the sagittaire's encodes... (let's say almost 3x) and however it looks better. so for a real comparison you should produce an encode that looks almost the same as h.264 and then we can compare the bitrates. Use metric tests (PSNR / OPSNR / SSIM) to estabilish level of quality...
However h.264 is AT LEAST 3x better for now...

JohnV

24th December 2005, 19:47

Yes, it would have been nice to start with uncompressed source. BTW, so you are 2x better than a 1990 codec? Very nice. Lol, what ever here is provided or said, it sounds like you downplay it somehow..
Anyway, if you want to go to this x-times better, I'd say it's definitely more than 2x better, cause it's less than half the size and much better quality..

And lets remind what you have said earlier: "According to H.264 marketing you will be able to get it down to 1/2 the size of my MPEG-1/MPEG-2 file. So around 85MB. Complete the test and I will post the results and write about it in my blog."
Interesting to see what you will write. :p

digigami

24th December 2005, 19:51

I finally managed to download your commandN_spotLIGHT clip (what a pain to retrieve 550 MB at a transfert rate of 3.5 kB/s) and made an encode using ateme mp encoder. I uploaded the result here:ftp://mood.ateme.com/digigami/commandN_spotLIGHT_HD_720p-ateme_avc_mp_1500k.mp4

Thanks. Judging by the server you are Ateme employee? How much $ is your HP encoder, BTW?

To me the source is quite irrevelant for such kind of test. The content is very static (an interview with 2 persons speaking in front of the cam with nearly no motion). The source quality is rather poor: I can see smearing on the man's clothes and it's very noisy.

The resulting avc stream features:

main profile
bitrate of 1500 kb/s. So it's less than half of your mpeg-1 which is 3200 kb/s.
2pass rcmode
2 references
extra quality mode
psycholevel enhancement enabled

So you can easily reproduce a similar result with Nero Recode 2 using the above settings.

According to my visual feeling, I prefer the avc version since it's sharper and more stable. Your opinion is welcome of course.

ps: I can host the source QT7 H.264 file and your mpeg-1 encode on our server if it can help since yours are totally overloaded.

Actually the commandN file is from the CommandN blog. Just goog it, or look in the links on my screenshots page.

JohnV

24th December 2005, 19:55

How much $ is your HP encoder, BTW?.If you wait a bit you can get the High Profile encoder from Nero Digital when it's put in. Currently the Nero Digital which includes the AVC Main Profile (which was used in bobololo's clip) is 29,99 Euro (so about $35). Plus you will also get AAC-HE v2 audio in the same package.

http://www.nerodigital.com

Hmm.. now I sound like a marketing guy!! :)

digigami

24th December 2005, 19:58

Lol, what ever here is provided or said, it sounds like you downplay it somehow..
Anyway, if you want to go to this x-times better, I'd say it's definitely more than 2x better, cause it's less than half the size and much better quality..

And lets remind what you have said earlier: "According to H.264 marketing you will be able to get it down to 1/2 the size of my MPEG-1/MPEG-2 file. So around 85MB. Complete the test and I will post the results and write about it in my blog."
Interesting to see what you will write. :p

For someone who won't use your own encoder to compress the Apple trailers, you seem quite chirpy. I'm going to write exactly what I find out. You don't seem to understand that for me to fail here is still a huge success in my books. My main point is about being HD at under 10 mbits/s, and there is no question that we are doing that.

I'm growing to respect a number of the posters here - the ones who think before they act, the ones who are oriented towards metrics and testing and the ones who understand there are 2 points of view. But it seems to me that H.264 is still bleeding edge, except where it comes to Ateme, which I understand (still waiting for an answer) is $9K.

Speaking of bleeding edge, can anyone list the H.264 decoders that "don't suck" and are in a stable release form?

digigami

24th December 2005, 19:59

Hmm.. now I sound like a marketing guy!! :)

That's fine with me. Everybody's marketing something, whether they believe it or not.

JohnV

24th December 2005, 20:01

For someone who won't use your own encoder to compress the Apple trailers, you seem quite chirpy. Eh, Sagittaire and bobololo have been faster than me (it happens to be a Christmas eve and I'm suppose not to spend my time in front of the computer which is not the fastest laptop anyway). They have provided Ateme streams which are used in the Nero encoder.
Point me one Apple trailer you want me to encode, and I'll encode it for you.

Sharktooth

24th December 2005, 20:03

For someone who won't use your own encoder to compress the Apple trailers, you seem quite chirpy. I'm going to write exactly what I find out. You don't seem to understand that for me to fail here is still a huge success in my books. My main point is about being HD at under 10 mbits/s, and there is no question that we are doing that.

I'm growing to respect a number of the posters here - the ones who think before they act, the ones who are oriented towards metrics and testing and the ones who understand there are 2 points of view. But it seems to me that H.264 is still bleeding edge, except where it comes to Ateme, which I understand (still waiting for an answer) is $9K.

Speaking of bleeding edge, can anyone list the H.264 decoders that "don't suck" and are in a stable release form?
well x264 is free and was rated almost on par with ateme encoder in the 2nd round of the doom9's codec comparison (well ateme is better by a very little margin).
yes, it doesnt support interlacing and other "pro" stuff, but it's a significative point that an open source codec can almost be on par with a professional one.
Also we can say h.264 can go HD with less than 2kbps... and that means it will render HD-DVDs and BD almost useless for video distribuition... maybe that's why Sony pushes for MPEG2?

P.S.: i'm not marketing/advertising anything commercial... i always provided encodes from free/open-source stuff.

Have to go now. Dinner time... :)

digigami

24th December 2005, 20:08

How about BBC motion gallery. Like I said in our press release, H.264 has a big edge with saturated color.

JohnV

24th December 2005, 20:11

How about BBC motion gallery. Like I said in our press release, H.264 has a big edge with saturated color.
Fine, just let me know what clip you want me to encode.

digigami

24th December 2005, 20:14

How about BBC Motion Gallery 'Reel'.

JohnV

24th December 2005, 20:19

I find many results. Which clip exactly?

digigami

24th December 2005, 20:25

Since Ben Waggoneer was already quoted once in this thread,
Gen's a smart guy, but I think his passion for his product has outstripped its real-world strengths.
http://www.avsforum.com/avs-vb/showthread.php?t=411600&page=600&pp=30

Regardless as to what Ben says, I'll take Sagittaire's opinion and method of comparing encoders over Ben's anyday. Also if you look at the language that they use, it's obvious that Sagittare (and others up here) are more versed in the underlying technology. At least I learn something when you guys bash on my samples, my encoder and my marketing. I rather enjoy it.

digigami

24th December 2005, 20:26

I find many results. Which clip exactly?

http://www.apple.com/quicktime/guide/hd/bbcmotiongalleryreel.html

digigami

24th December 2005, 20:31

P.S.: i'm not marketing/advertising anything commercial... i always provided encodes from free/open-source stuff.

You are marketing your knowledge, your builds of x264 for Windows, your config files and your quant matrices. Marketing has nothing to do with money - it means promoting what you believe in. Apple markets both H.264 and iPod.

BTW, do the Apple produced H.264 files play on all the H.264 decoders? Are their streams, which are a subset of AVC, compliant and well-formed?

JohnV

24th December 2005, 20:58

At least I learn something when you guys bash on my samples, my encoder and my marketing. I rather enjoy it.
I don't think everybody is bashing your encoder. But it depends quite a lot with what kind of claims and attitude you come here.
If for example you'd ask help for evaluating your encoder and suggestions where it needs improvement, I'm 100% sure the feedback would be totally different.
But when you come to Doom9 with claims that your MPEG1/2 encoder is as or more efficient as h.264, it's the wrong start and people will act accordingly.

But have to say that one positive thing is that you seem to be actually interested what is said here.

I'll encode the Reed Apple clip for you and maybe that commandN with even lower like 1000kbps just to see what it looks like. :)

digigami

24th December 2005, 21:07

But when you come to Doom9 with claims that your MPEG1/2 encoder is as or more efficient as h.264, it's the wrong start and people will act accordingly.

I did _not_ come to doom with claims. If you go up to the start of this thread, about 6mos ago, you will see I came here to defend the unverified complaint about blockiness which was never answered. In fact, it started again until you, JohnV, auditioned our file in VLC where you could see it properly rendered.

I came back because the thread started again.

But you know what they say, "If you can't beat em, join em". And, my friends, I am looking for a H.264 encoder for my freeware:
http://www.digigami.com/moviesformypod/

And I know exactly which one I'm going to use.

JohnV

24th December 2005, 21:14

I did _not_ come to doom with claims. Ok. :) But maybe you shouldn't have come to Doom9 to talk about your encoder if you have just released a press release with such claims.. ;)
Eventhough in that playback blockiness case you were right as I confirmed, WMP playback is much worse with your streams than with VideoLan or Nero ShowTime (for example).

digigami

24th December 2005, 21:28

I'll encode the Reed Apple clip for you and maybe that commandN with even lower like 1000kbps just to see what it looks like. :)

Cool sounds great. I told you I wanted to learn about H.264 and I have my ways of finding the right people to ask.

http://www.digigami.com/dg-ice-age.png
MPEG-1 VBR, semi-constant Q via Q-floor.
I have one more pass. Too much filtering, so again, looks like H.264 AVC done by pros is 1/3 size of MPEG-1 at 1280x720x25fps.

Oh yeah, M$FT DirectShow MPEG-1 decode may not be able to play IBBBP. It's either 3 or 4 B-frames that make it choke. But VLC works great.

digigami

24th December 2005, 21:36

Ok. :) But maybe you shouldn't have come to Doom9 to talk about your encoder if you have just released a press release with such claims.. ;)
Eventhough in that playback blockiness case you were right as I confirmed, WMP playback is much worse with your streams than with VideoLan or Nero ShowTime (for example).

I have a number of macs and pcs to work with and if using VLC, our streams look great. Both QT Player and MSFT MPEG-1 decoders are second class citizens, and so they just work OK. But at least the streams play. Also, it is the agressive quant that brings out the blocks with poor DCT accuracy. Like I said, it is the DC coefficient inverse quantize and off by one gray level or more that causes the problems. However, it is not restricted to MPEG-1. If you single step through most Hollywood pro-encoded DVDs with MPressionist on fixed point DCT, you can usually see blocking on B-frames. Very light, but it is definitely there. Switch to 32-bit float reference and it looks perfect.

digigami

24th December 2005, 22:33

http://digigami.in-long-beach-ca.com/megapeg/Doom9/Pour-Sagittaire/dg-trailer-1700kbps.png

This encode represents _my preferences_ with respect to low bitrate coding of video. I'll be happy to hear how the PSNR stacks up, but to my eye (and you will need good IDCT for this), there are blocks, but they are well-hidden and do not detract from the program at a normal viewing distance. For exaggeration, I have upsampled the frame to 1280x720. I am using the split-view of MPressionist to lay up the original .m2v and the bitrate-reduced version side by side (vertical split). Which is which?

Audition with VLC.
http://digigami.in-long-beach-ca.com/megapeg/Doom9/Pour-Sagittaire/megapeg-1.7mbs.m2v

My upload is only 1/3rd done, so keep that in mind when downloading. Server is slow and I am too lazy to walk to the next bldg to plug firewire in for a quick transfer.
Pound away.

digigami

24th December 2005, 22:38

Also, since I was under the target bitrate, I decided to do a couple more. One note: our quant matrices tend to hammer the chroma at very low bitrates. But overall, I like the way they degrade better than the ISO or usual MPEG-2 suspects.

digigami

24th December 2005, 22:59

Quick notes about the screen grab. Once again, I am using the split screen feature to lay up A|B side by side. However, I encoded at 24fps and the source is 23.976 and it is off by one, which I am using to show the dividing line. As you can see, there is some fine detail in the Apple version not present in the MPEG-1, but it is very slight.

According to Mr. Waggggonnner, his fastest encoder is 60x faster than this. My encoding ratio for 1280x720x24p on "Mastering Quality", which is exhaustive half-pel search equivalent (exactly) to a hardware MPEG encoder is 20:1. So that means he has an encoder that runs 3x real time doing exhaustive motion compensation. I don't think so.

http://digigami.in-long-beach-ca.com/megapeg/Doom9/dg-ice-3630.png

Ok, this is as low as my quality meter will allow. 3.6 mbs/s, but I think the quality is on target. Complain about the bitrate if you must, but I'd like to see the H.264 version without blocks. There is some very minor blocking in this, can you find it with VideoLAN?

http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v

At this time, 20min remaining. Christmas eve. Made some new friends who are smart, doing cool stuff. Hope you feel the same. Time to take a ride, get some sunlight.

kiyooka-san!
hai!

CruNcher

24th December 2005, 23:05

sure but hey their is still the tractor that want's to be encoded ;)
i also redid Ben Waggoners 8 Mbps The Island Trailer with the Ateme Encoder and X264 alot more complex then his quicktime encode, unfortunately i don't have so much space available for upload i see that i can host it maybe on Rapidshare :)

digigami

24th December 2005, 23:06

BTW, if someone wants to whip up a .torrent file for these, I'm sure my friends whose web sites I host will appreciate it.

bobololo

24th December 2005, 23:06

Thanks. Judging by the server you are Ateme employee? How much $ is your HP encoder, BTW?

Yes I work at Ateme, though I'm not from the sales so I won't be able to give you accurate info about pricing and licensing model of our encoder. Just keep in mind that our primary role is to provide technology to industry integrators and therefore our customers are mostly professionals. These latters offer our IP in their products to end users. So our price isn't very significant since it's not directly related to the end user product's price which can range from 30 euros to some stratospheric amounts ;)

digigami

24th December 2005, 23:08

sure but hey their is still the tractor that want's to be encoded ;)

Doood! my tractor.yuv gets to 100% and just hangs. I restarted it 2x.

BTW, can QT read this file?

JohnV

24th December 2005, 23:20

digigami:
Here's some more AVC clips for you
http://morbo.org/nero/commandN_spotLIGHT_HD_720p-ateme_avc_hp_1000k.mp4
http://morbo.org/nero/bbc-r_m720p-NeroDigital_3500k.mp4

There's the 1000kbps commandN with High Profile latest Ateme codec. Still looks good.
Then there's 3500kbps BBC motion clip with NeroDigital AVC (Ateme Main Profile). I'll do a 2500kbps encode also. Notice that there's some heavy blocking already in the original Apple trailer especially in the blue sky areas.
Your comments are welcome. :)

IgorC

24th December 2005, 23:21

Why you are showing us 3.6 mbps MPEG-1 encoding when bitrate was 1.250 mbps?

http://forum.doom9.org/showthread.php?p=756651#post756651

digigami

24th December 2005, 23:59

Why you are showing us 3.6 mbps MPEG-1 encoding when bitrate was 1.250 mbps?

http://forum.doom9.org/showthread.php?p=756651#post756651

I've already described in my blog that JohnV, Sagittaire, and others have demonstrated that emerging H.264 coders and Ateme (which is robust), are capable of the bitrate claims of reduced bitrate burden of MPEG-2. However, this thread was largely about our press release. Our press release was about the suitability of MPEG-2 for HD-DVD and Blu-Ray, and that we were able to achieve HD at SD DVD bitrates. Well fella, 3.6 mbits/s is a SD DVD bitrate, and the quality is definitely there. And that gives us 24 mbits/s to do whatever else we want 1080i etc. Top notch quality.

Personally, what I am most interested here at Doom9 is intelligent, thoughtful conversation. If you're not interested in that, go join another thread. This one started with someone questioning the 'blockiness' of our MPEG-1 streams 6 months ago. Today that matter was resolved as well.

Merry xmas to ya all.

JohnV

25th December 2005, 00:05

Here's the BBC Reel with 2.5Mbs HP
http://morbo.org/nero/bbc-r_m720p-ateme_hp_2500k.mp4

JohnV

25th December 2005, 00:08

3.6 mbits/s is a SD DVD bitrate, and the quality is definitely there. Can you now provide 3.6Mbps BBC Reel MPEG-1/2 clip for comparison?

digigami

25th December 2005, 00:09

Why you are showing us 3.6 mbps MPEG-1 encoding when bitrate was 1.250 mbps?

http://forum.doom9.org/showthread.php?p=756651#post756651

Also, if you want to 'pipe in', I have requested a version of Ice Age (Kung Fu) which does not have blocking artifacts in the low-contrast, low-light regions of the clip. You cannot claim 1250 as a HD bitrate for feature films if you have the artifacting. Fine for archiving HD on whatever, but not for original material to be replicated. I want to know what the real HD bitrate is, probably more in the 2000-3000 range for perfect quality. There is also some aliasing in the Ice Age trailer, and some banding.

digigami

25th December 2005, 00:11

Can you now provide 3.6Mbps BBC Reel MPEG-1/2 clip for comparison?

I will look at this one, but I can tell you it will not be pretty. However, our motion comp has improved since the last time I encoded it, so we will see.
right now, food, etc.

Fun meeting you guys. As we say in California, Totally! Of course, I'm from Canada myself.

IgorC

25th December 2005, 00:12

digigami

Ok. Sorry. I've downloaded this sample http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v.
Quality isn't bad for such low bitrate as 3.5 mbps. However it isn't nor even closer to call it HDTV. Already knows issues like ringing, artefacts, visual macroblcoks on palet of soft colors. It's typically for MPEG-1/2.
I'm not sure but I think mencoder MPEG-2 in 3 passes with slowest setings can do the same job or even better.

I can't say the same thing for H.264 at 3.5 mbps . All details are preserved, no visual blocks. Excelent visual quality.

Don't see any point to discuss. Your trailer clear shows what MPEG-1 can do (can not do).

CruNcher

25th December 2005, 04:11

Here is the Tractor @ 3 Mbps 1920x1080x25p
http://rapidshare.de/files/9780965/Tractor_HD-High_Profile-x264.mp4.html

and here is the more complex High Profile X264 version of Ben Waggoners The Island Trailer 8 Mbps test (better detail preservation) 1920x800x23.98 manualy croped and keeped the 2.40 AR :) (dunno how Ben Waggoner comes to the idea that this clip is 2.35 :P )
http://rapidshare.de/files/9781798/The-Island_HD-Trailer-High_Profile-x264.mp4.html (more then 2.4 GHz needed)
http://rapidshare.de/files/9787718/The-Island_HD-Trailer-High_Profile-x264-nodeblocking.mp4.html (less complex) (at least 2.4 GHz needed)

here is the XviD EDP version (needs less CPU resouces for playback) (still comparable detail preservation to X264 HP @ this bitrate)
http://rapidshare.de/files/9784778/The-Island_HD-Trailer-XviD-EDP.mp4.html (2 GHz needed could also run on slower)

Ben Waggoners Quicktime Encode

http://cruncher.mufflastig.com/x264/The-Island-BenQuicktime.png

CruNchers High Profile X264 Encode

http://cruncher.mufflastig.com/x264/The-Island-X264.png

CruNchers XviD EDP 1.4 Encode (Better Profile)

http://cruncher.mufflastig.com/XviD/test/The-Island-XviD-EDP.png

Sagittaire

25th December 2005, 11:07

Merry Christmas everyone.

@ digigami

your link don't work (no access)
http://digigami.in-long-beach-ca.com/megapeg/Doom9/Pour-Sagittaire/megapeg-1.7mbs.m2v

For good comparison you must target the perfect bitrate (PAL MPEG2 MP@ML DVD Compliant stream at 1800 Kbps if possible). Because it's low quality for MPEG2 and in low quality zone Metric variability is very high ...

For MPEG2 vs H264

As I say differencies are exponential : lower bitrate and higher difference. I think that metric comparison for each quant can be usefull ...

digigami

25th December 2005, 11:38

It's typically for MPEG-1/2.
I'm not sure but I think mencoder MPEG-2 in 3 passes with slowest setings can do the same job or even better.

If you want to be a scientist/engineer, please describe your playback environment, IDCT precision, and qualify your criticisms with screenshots. I know for a fact that what you have said is not true. Go back up to the top of the thread and look at this (http://forum.doom9.org/showpost.php?p=684002&postcount=8). Nobody responded to my post. Not one.

I have eyes. I have a high end PC. I have MPEG analysis tools that I have built myself for the purpose of performing this analysis.

What method do you use to compare? Even a screenshot from your program is better than nothing.

If you look up this thread, you will see that I am "not allowed" to use QT player to view the H.264 streams because it is crap. Therefore, it is my right to request a certain level of competence in MPEG-1 playback. Thank you.

BTW, the Nero streams seem to work partially with QT player. Does Nero have a template for making QT compatible streams?

digigami

25th December 2005, 11:42

I'm not sure but I think mencoder MPEG-2 in 3 passes with slowest setings can do the same job or even better.

You are not sure but you think? Either do your own encode to compare, and post the file, or please go back to lurking. If you post a file, I will bring it up under MPressionist and I will show you exactly how the files differ. So if you are used to 'comparing' by simply watching, could be problematic.

digigami

25th December 2005, 11:54

For the record, these are my outstanding 'questions/requests' without answers as of yet:

1. Ice Age (Kung Fu) trailer without blocks in darker low contrast areas. What bitrate is it free of artifacting of any kind.
2. How your bitstreams are checked for conformance/compliance with spec. It is not possible to do this by watching as I'm sure H.264 is very resilient to errors as is MPEG-2.
3. Exactly what subset of the H.264 AVC is QuickTiime player implementing?
4. Exactly what subset of the H.264 AVC is QuickTiime encoder implementing?
5. Who is interoperable with who? Does someone have a compatibility matrix with decoders/encoders and what parts of spec have been implemented? For instance, does VLC decode interlaced H.264?
6. What is the minimum PC required to decode full (720p,1080p) H.264 AVC feature set with VLC? Clock speed, video card and CPU (mmx,sse,sse2 etc).
7. What is the minimum Mac required to decode same.
8. New: what H.264 playback program allows you to pause and look at individual frames?
9. New: how do I obtain Ateme HP encoder?
10. New: what subset of H.264 is iPod supporting. Frame rez is max 320x240, but what subset of features.

savage747

25th December 2005, 13:17

Heh, what a nice "discussion". ;)

I think nobody here seriously doubts h264 is really really really superiour to MPEG2 when it comes to "low" bitrate video. However, when it comes to storage of HD-content on next-gen optical media (HD-DVD or BluRay) you have a lot of bits to spend. And if MPEG2 can deliver basically "perfect" HD video at bitrates > 15 mbit there is no real reason why you shall use h264.

Seems at least Sony Pictures thinks so: http://www.engadget.com/2005/11/29/sony-pictures-to-use-mpeg-2-on-blu-ray/ (by now this has been confirmed to be more than a rumor)

I don't really what made Sony to decide the way they did - perhaps they are just very used to their MPEG2 production chain or want to make sure the higher storage capacity is really used (or lower license fees?).

However, it seems MPEG2 will be with us a long time to go...

Doom9

25th December 2005, 13:55

Sony needs a reason to justify the size premium over HD-DVD.

2. How your bitstreams are checked for conformance/compliance with spec. It is not possible to do this by watching as I'm sure H.264 is very resilient to errors as is MPEG-2.
Naturally, you'd have to use the reference JM decoder to conformance testing, just as for any other standard.

4. Exactly what subset of the H.264 AVC is QuickTiime encoder implementing?Baseline + 1 B-frame. Technically that makes it Main goes a lot further. And in terms of performance and quality, QuickTime lags severely behind good AVC encoders.
3. Exactly what subset of the H.264 AVC is QuickTiime player implementing?The same as the encoder. Bottom line, QT gives AVC a bad name.
5. Who is interoperable with who? Does someone have a compatibility matrix with decoders/encoders and what parts of spec have been implemented? For instance, does VLC decode interlaced H.264?Have a look here for decoder features: http://forum.doom9.org/showthread.php?t=99402
8. New: what H.264 playback program allows you to pause and look at individual frames?Only every DirectShow based player ;)
9. New: how do I obtain Ateme HP encoder?I believe they put the contact information right on the product description page: http://www.ateme.com/products/pcencoder_fr.php

like to see at what bitrate it hits the quality expectation for commercial concerns, rather than ripping and archiving.commercial concerns eh? Why is it that when watching DVDs on good equipment (flatscreen with a native HDTV resolution, fed via DVI) I always find severe imperfections? Hollywood doesn't care about quality a lot.

Last but not least, it's very convenient if the encoder just happens to be on a platform nobody around here has ;) Makes it very hard to disprove a point...

digigami

25th December 2005, 14:53

digigami

25th December 2005, 14:56

Sony needs a reason to justify the size premium over HD-DVD.

I'm not exactly sure what you mean by this, but when has anyone ever wanted a smaller optical disc format rather than a larger one?

All Sony has said is that they are putting their features on Blu-Ray in MPEG-2 format. I believe they continue to support the HD codecs of H.264 and M$FT.

Gen

digigami

25th December 2005, 15:00

Last but not least, it's very convenient if the encoder just happens to be on a platform nobody around here has ;) Makes it very hard to disprove a point...

? What is the problem, I have a Mac, I have a PC.

According to this thread, Ateme HP is 'publicly unavailable'.
http://forum.doom9.org/showthread.php?t=99402

In my books, this says "Unobtanium" and should not be allowed to compete.
That leaves us the version of Ateme included in Nero.

Does the Nero version support interlaced encoding?

digigami

25th December 2005, 15:06

Bottom line, QT gives AVC a bad name.

I disagree. Between Apple's HD Trailers and the iPod, Apple is raising public awareness for H.264 and AVC and this will inevitably help H.264 _and_ lead to more awareness for the superior encoders.

Instead of simply criticizing, consider partnering with people like commandN who do not understand the difference and help them to Podcast H.264 with a better encoder.

Interoperability is the key to standards.

In 1997, CreativeLabs released a hardware player that refused to play my MPEG streams. I could have blamed them, but instead, I made changes to what I was doing to support their limitations. Consumers expect files to play on all devices. The acceptance of the standard is accelerated not by showing off the ultimate encoder performance but by creating the most seamless and uniform experience for the consumer, so that they can realize the benefits of the technology as soon as possible. In that respect, Apple's H.264, despite its technical limitations, reaches the widest audience, has robust playback and user experience. Eventually they will play your advanced H.264 streams just fine. In the meantime, if you are in support of the acceptance of the standard, it is a good idea to make it easy for people using these more advanced H.264 encoders to create files compatible with QT.

Other than the Apple site and DVB transport stream grabs, where do you find your HD content to play with? Apple and MPEG-2.

"We stand on the shoulders of giants".

digigami

25th December 2005, 15:13

However, it seems MPEG2 will be with us a long time to go...

No question about that. Basic television unchanged for 40+ years.

Also, when FCC mandates switch to HDTV in the US, it will be all MPEG-2 + optional H.264. Count on it. The companies that have already geared up for MPEG-2 HDTV transmission are not going to change again. They might add channels/features but not change.

As I understand it DVB is a pay-for-service, yes? Like DirecTV. ATSC HDTV is different. It is public airwaves. Just buy the receiver. Huge difference.

Also, you can expect MPEG-2 encoders to get better. Count on it. Refer to my comment about gigabit ethernet over twisted pair and you will understand.

digigami

25th December 2005, 15:16

Why is it that when watching DVDs on good equipment (flatscreen with a native HDTV resolution, fed via DVI) I always find severe imperfections? Hollywood doesn't care about quality a lot.

Can you give me an example? I believe there are also counter-examples. Superbit DVDs are encoded by people that care about quality. You can see it in their use of bitrate, custom quant matrices and other details.

digigami

25th December 2005, 15:37

Ateme Main Profil (Nero)
http://multimediacom.free.fr/Download/NDAVCMP_IceAge_720p_1250.mp4

OK, I am seeing playback problems with this stream using VLC 084a on both Mac/G5 and WinXP/P4HT.

The problem occurs when the big closeup of the 'eyeballs' comes on screen, and immediately after, the playback freezes for 2-3 seconds before continuing.

Would anyone like to address this problem or submit a different H.264 file encoded with a different encoder?

UPDATE: If this stream is bad, the encoder is disqualified. If the stream is good, then this entire profile is disqualified because my press release refers to 'current H.264' and noone can tell me that VideoLAN is not a current H.264 player. And grabbing a nightly build is not current, it is pre-production. Either way, I need a Ice Age (Kung Fu) trailer that plays without glitching on all the primary H.264 decoders on PC.

savage747

25th December 2005, 16:09

As I understand it DVB is a pay-for-service, yes?

Depends.

DVB-T (terrestrial) is 100% free at the time being (no cable fees, no encryption that'd force you to hire a key card or whatsoever). I'm using DVB-T and it's usually nice. Some tv networks broadcast with 2.5 mbit/s on average, though (because their stupid Dolby Digital (mostly 2.0, sometimes 5.1) consumes ~400 kbit/s - and they have to broadcast 192 kbit/s MP2 audio, too).

<off-topic>
Oh, and some stations re-encode their DVB-C/S streams to generate a lower-bitrate DVB-T stream... argh.

I'd really appreciate an efficient MPEG-2 encoder for standard definition at bitrates between 2 und 3 mbit/s... no idea how efficient the usual rack-mounted "hardware" MPEG-encoders are.
</off-topic>

However, on DVB-C (cable) and DVB-S (satellite) some tv networks begin to use encryption even when *not* charging a monthly fee - that's because Hollywood is paranoid and gives some nice discount if the consumer cannot record their content.

Pay-TV per DVB-S is something really common, too.

bobololo

25th December 2005, 16:23

Hi bobololo,

I looked at your file and it looks great. However, for this particular example, since the commandN (http://commandn.typepad.com/) audience are regular PC/Mac users and just want to watch the movie, can you re-encode it using a H.264 profile that QuickTime can play on Mac/Windows. Otherwise, in this case it is not a fair comparison. It is unrealistic for regular PC/Windows users to have to go get a fresh build of their player before viewing every file. Also, many internet users do not have Admin priviledges and cannot necessarily even install new software on their computers.

Gen

Gen,

I don't really see where you're trying to get, but please consider that the only purpose of this encoding was to answer to your request to demonstrate that H.264 could perform much better than MPEG-1/2 and nothing more.

If you want to learn more about H.264 and especially if you want to check what is achievable with QT restrictions, I'd suggest you to spend some time on the corresponding threads on this board where those topics have been deeply discussed and to make you own tries. Doing it yourself is often the best way for learning ;)

Sagittaire

25th December 2005, 16:33

OK, I am seeing playback problems with this stream using VLC 084a on both Mac/G5 and WinXP/P4HT.

The problem occurs when the big closeup of the 'eyeballs' comes on screen, and immediately after, the playback freezes for 2-3 seconds before continuing.

Would anyone like to address this problem or submit a different H.264 file encoded with a different encoder?

UPDATE: If this stream is bad, the encoder is disqualified. If the stream is good, then this entire profile is disqualified because my press release refers to 'current H.264' and noone can tell me that VideoLAN is not a current H.264 player. And grabbing a nightly build is not current, it is pre-production. Either way, I need a Ice Age (Kung Fu) trailer that plays without glitching on all the primary H.264 decoders on PC.

No problem for me ... VLC 0.8.2

but you have very strange and not objectif judgement : "If my prefered player is not able to play the stream then stream is bad" ... :confused:

1) H264 is compliant format with HD-DVD and BR
2) H264 is always better than MPEG2 in all situation (with small advantage for low quantizer but high advantage for high quant)

digigami

25th December 2005, 16:38

Gen,

I don't really see where you're trying to get, but please consider that the only purpose of this encoding was to answer to your request to demonstrate that H.264 could perform much better than MPEG-1/2 and nothing more.

I disagree. The purpose of compressing video is to allow people to view it. The commandN clip was distributed over the internet to allow viewers of the commandN TV show to see the show in HD, instead of 320x240. I transcoded the file to MPEG-1 and reduced the size to make the download manageable. To provide an "alternative" H.264 encoding, using a current encoder, it must be realistic comparison. It is not realistic to say that this H.264 file that has been produced will play on a DVB set-top box with H.264 built in, or only on a special H.264 decoder on a PC. If H.264 is a standard, your comparison should be a file which plays back on all the popular H.264 decoders. For this example. HD-DVD and Blu-Ray are another matter altogether.

But, if you don't want to do it, that is fine. I just bought Nero Digital, so I will post my own results.

You have complained that the QuickTime AVC encoder is crap, so I would think you would like an opportunity to demonstrate that another encoder can produce a file to the exact same H.264 profile that QT supports, but at a vastly reduced file size.

digigami

25th December 2005, 16:40

No problem for me ... VLC 0.8.2

but you have very strange and not objectif judgement : "If my prefered player is not able to play the stream then stream is bad" ... :confused:

Rather: if H.264 is a standard, the _current_ support for that standard is what I am interested in. My press release describes 'current', not emerging or forthcoming, or futuristic. It talks about what is available, proven, ready right now.

Sagittaire

25th December 2005, 16:43

Can you give me an example? I believe there are also counter-examples. Superbit DVDs are encoded by people that care about quality. You can see it in their use of bitrate, custom quant matrices and other details.

yes try this with superbit bitrate
http://multimediacom.free.fr/Download/lossless.avi

DVD for complexe motion/texture is unable to make good quality

http://multimediacom.free.fr/Download/MPEG2-hightexture.jpg

Try with your MPEG1/2 encoder if you want

And here my best libavcodec result for my HPII trailer challenge (it's a new record with 47.35 dB by far better than CCE, TMPGEnc, HCEnc, Procoder ... )
http://multimediacom.free.fr/Download/Libavcodec.m2v

digigami

25th December 2005, 16:45

Depends.

DVB-T (terrestrial) is 100% free at the time being (no cable fees, no encryption that'd force you to hire a key card or whatsoever). I'm using DVB-T and it's usually nice. Some tv networks broadcast with 2.5 mbit/s on average, though (because their stupid Dolby Digital (mostly 2.0, sometimes 5.1) consumes ~400 kbit/s - and they have to broadcast 192 kbit/s MP2 audio, too).

<off-topic>
Oh, and some stations re-encode their DVB-C/S streams to generate a lower-bitrate DVB-T stream... argh.

I'd really appreciate an efficient MPEG-2 encoder for standard definition at bitrates between 2 und 3 mbit/s... no idea how efficient the usual rack-mounted "hardware" MPEG-encoders are.
</off-topic>

Yes, an interesting problem. I think if you look at the upstream compression environments for your DVB H.264 future, you are going to see different encoding results than you are getting with your multiple pass encoder samples produced here. In many broadcast situations, multi-pass simply is not feasible because there is a variety of diverse source material all feeding into a single pipe.

So even if the DVB set-top box supports all the advanced H.264 features, I doubt you will see 15 reference frames in a broadcast feed unless it was compressed offline for ad-insertion or something similar.

However, on DVB-C (cable) and DVB-S (satellite) some tv networks begin to use encryption even when *not* charging a monthly fee - that's because Hollywood is paranoid and gives some nice discount if the consumer cannot record their content. Pay-TV per DVB-S is something really common, too.

Since I am in the US, perhaps you can give a breakdown of what HD formats are available via DVB in your area.

digigami

25th December 2005, 16:47

yes try this with superbit bitrate
http://multimediacom.free.fr/Download/lossless.avi

Sagittaire, he was looking for an example from Hollywood that cared about video quality.

digigami

25th December 2005, 16:49

but you have very strange and not objectif judgement : "If my prefered player is not able to play the stream then stream is bad" ... :confused:

Fine, you tell me what decoder to use and I will use yours.

bobololo

25th December 2005, 16:50

? What is the problem, I have a Mac, I have a PC.

According to this thread, Ateme HP is 'publicly unavailable'.
http://forum.doom9.org/showthread.php?t=99402

In my books, this says "Unobtanium" and should not be allowed to compete.
That leaves us the version of Ateme included in Nero.

Does the Nero version support interlaced encoding?

Ok to make things clear about the availability of our codecs. The main profile encoder is available from Nero Digital Pro bundle. This latter aims "mass" market. The retail price is 30 euros. And in this version, the interlaced encoding isn't supported.

The high profile encoder is currently only available from our pc encoder suite (which includes much more than the avc encoder). It aims professional market. The indicative price for a single seat license is 9k euros. It supports interlaced encoding. This high profile encoder is expected to be integrated into Nero Digital in the near future.

Beside those official products, a few testers from doom9 forums had the chance to get both main & high profile codecs during semi-public beta test.

Sagittaire

25th December 2005, 16:56

Sagittaire, he was looking for an example from Hollywood that cared about video quality.

But it's really simple to find that because DVD MPEG2 MP@ML are unable to make good quality because vbv limitation are too high in most situation.

Try to encode my example. There are no MPEG2 encoder in the world able to encode this scene with good quality for MPEG2 DVD compliant stream.

Doom9

25th December 2005, 17:04

Can you give me an example?My codec comparison sources. Look at my review scenes of Matrix3.. the source simply isn't good enough.

You have complained that the QuickTime AVC encoder is crap, so I would think you would like an opportunity to demonstrate that another encoder can produce a file to the exact same H.264 profile that QT supports, but at a vastly reduced file size.Codec comparison again.. at the same bitrate, other AVC encoders (ateme, x264, elecard, xvid avc to put some names.. you can download and try x264 on your own so just that we have the availability issue sorted, and you can buy the mainconcept encoder or the ateme encoder which both contain essentially the engines I've tested) provide significantly better looking picture. Personally, I find the comparison of same bitrate more productive than trying to find corresponding bitrates that yield the same quality.. you may not be able to see the effect of small bitrate changes on the output..

And if you want x264 output that is compatible with QuickTime, head over to the AVC forum.. there's plenty of threads on that particular subject. We have a rule here: it's read up first (and that includes making forum searches), ask questions later.

The problem occurs when the big closeup of the 'eyeballs' comes on screen, and immediately after, the playback freezes for 2-3 seconds before continuing.No problems here, with either DirectShow based players or VideoLan using either my custom 0.84 build with Dirac support or the release 0.84a version.

IgorC

25th December 2005, 17:18

If you want to be a scientist/engineer, please describe your playback environment, IDCT precision, and qualify your criticisms with screenshots. I know for a fact that what you have said is not true

Yes, it's true.
http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v

Player Videolan. It's impossible to see clearly ringing on screenshots. But it's obvious during playback. It's defeneatly not HD-DVD quality.
Maybe somebody can play with some zones of video samples to improve rate control. I think it's not fair.

Ringing.
http://www.bilder-hosting.de/uploads/10e7094ac2.png

Artefacts
http://www.bilder-hosting.de/uploads/fd91b42ded.png

Artefacts
http://www.bilder-hosting.de/uploads/c71ecf9901.png

There was also a MPEG2 test. http://forum.doom9.org/showthread.php?t=101735. Where mencoder has shown very good results at low bitrate. Maybe it is will be usefull to see. There's nothing new that MPEG-2 can provide acceptable quality at low bitrates as 3.6-5 mbps for 720p but it's still not on par with H.264.

digigami

25th December 2005, 17:19

No problem for me ... VLC 0.8.2)

Are you sure? Have you auditioned the file in QuickTime to see what it is supposed to look like?

UPDATE: Ok I have downloaded the only Nero AVC sample from the Nero movie trailers gallery. I also downloaded Nero showtime 30 day trial.
I bought Nero Digital, but am still waiting for an email with my download/license.

Machine specs: Dell XPS 3.4 Ghz HT P4. High end gaming laptop.

In trying to play the H.264 AVC with Nero Showtime I get continous frame dropping. Am I missing something here? You can call QT crap if you want, but at least their HD files give you a HD experience.

digigami

25th December 2005, 17:23

Player Videolan. It's impossible to see clearly ringing on screenshots. But it's obvious during playback. It's defeneatly not HD-DVD quality.

This is a ridiculous statement. When dealing with compressed video the reverse is true: when choosing single frames it is usually quite easy to see artifacts, but at 24fps (or 25fps) it is difficult.

The human eye uses something called foveation to scan the visual field. The eye is always moving and has a response time for adjusting to changes in content. 100ms, which is usually approx 1 or 2 frames, depending on the base framerate. So an artifact in a single frame will generally never be seen. (The H.264 clip has artifacts as well. Demonstrate objective analysis and go find them).

Usually, the eye will never catch a single video artifact unless it is the most interesting thing in the visual field. http://lists.apple.com/archives/Dvdlist/2005/Jun/msg00015.html

You say there are artifacts. Show the H.264 frame and the same MPEG-1 frame side by side, or better yet, find the artifacts in the H.264 file.

If you need an example of 'ringing', I will show you one. What you are showing is very minor block artifacts. I already said there was minor block artifacts in my stream (see the original post). Ringing occurs near sharp edges (like titles) and is a result of the inverse quantization process.

IgorC

25th December 2005, 17:30

You obligated me to provide some screenshots if I want to do serios statements and now saying that's it's ridiculuos to judje quality looking at sreenshots. Have I understood you correctly?

So I can't say my opinion about what I've seen and also I can't post sreecshot. SO What can I do.

My visual percepcion and my screenshots are here. If you don't beleive them let's do some syntethic test like SSIM and OPSNR.

IgorC

25th December 2005, 17:34

What you are showing is very minor block artifacts.

Minor? I can see them perfectly during playback.

IgorC

25th December 2005, 17:36

Ringing occurs near sharp edges (like titles) and is a result of the inverse quantization process.

I know what you want to say. But ringing is still here.

digigami

25th December 2005, 17:41

Minor? I can see them perfectly during playback.

Go look at the H.264 Clip of the same trailer that Sagittaire has produced as an example of perfect quality at 1250.

In the last 2/3rd of the movie the one character is standing in a cave. Look to the left of the character in the darker area. In the Apple version, there is banding here, but in the Sagittaire version blocking is highly evident.

It is a fact of video compression that block artifacts are difficult to hide in low-contrast areas. Such is the case here, which is why I have asked for a second encoding at the bitrate required to eliminate the blocks. My clip also exhibits blocking in certain areas.

Yes?

digigami

25th December 2005, 17:43

You obligated me to provide some screenshots if I want to do serios statements and now saying that's it's ridiculuos to judje quality looking at sreenshots. Have I understood you correctly?

No, you did not understand. You said that you could see artifacts that could not be seen in the screenshots. I said the reverse. I said that artifacts could be present that would be hard to detect with the eyes, but easy to find with screenshots.

I would say that the block artifacts in your screenshot would be difficult to see under normal viewing conditions, but the artifacts in the cave, which are present both in the MPEG-1 and H.264 by Sagittaire would be easy to see.

digigami

25th December 2005, 17:47

And if you want x264 output that is compatible with QuickTime, head over to the AVC forum.. there's plenty of threads on that particular subject. We have a rule here: it's read up first (and that includes making forum searches), ask questions later.

Noted.

No problems here, with either DirectShow based players or VideoLan using either my custom 0.84 build with Dirac support or the release 0.84a version.
No problems here with Nero Showtime. I am now going to audition my MPEG-1 VBR on Nero Showtime.

IgorC

25th December 2005, 17:47

I would say that the block artifacts in your screenshot would be difficult to see under normal viewing conditions

You're saying to me that I don't see them while I'm saying you that I see them easily.

digigami

25th December 2005, 17:50

I know what you want to say. But ringing is still here.

In your 3 screenshots there is no ringing. But there is light block artifacting. I think if you inspect the H.264 version more closely you will see similar artifacts - not necessarily in the same places. I already told you where to look for one example. However, you might have trouble finding them with VLC, because I believe VLC only allows random access to key-frames, which almost never have artifacts.

I wish to see an example of H.264 Ice Age (Kung Fu) that does not exhibit block artifacts in the cave sequence. In the Apple trailer there is banding, so banding is AOK.

I myself am having difficulty taking snapshots with VLC. Does anyone know if this features is broken in 084a?

Also, recommend screen grabber that works with ATI cards. I believe VLC, InterVideo, and ShowTime are using the video overlay, either impossible or hardware specific to grab frames. Have not been using PC much so...

digigami

25th December 2005, 17:52

Minor? I can see them perfectly during playback.

Yes, as can I see the H.264 artifacts in the cave clearly. It is a function of bitrate, which is why I would like to see what bitrate H.264 requires to encode this trailer without artifacts.

Doom9

25th December 2005, 17:57

This is a ridiculous statement. When dealing with compressed video the reverse is true: when choosing single frames it is usually quite easy to see artifacts, but at 24fps (or 25fps) it is difficult.Actually, it's not so ridiculous at all. On the one hand, it's true, a picture is no true representation of what you percieve while watching the video, but that can go both ways. On the one hand, you are likely to miss individual artifacts in moving video, on the other hand, screenshots cannot show you changes over time. Take what I like to call the "keyframe readjustment" issue. While the term isn't technically correct, the phenomenon is that from one frame to another there's a significant change between two frames. So, while a screenshot from frame X and frame X+1 may look just fine, if played after another, you'll notice something is off.
Another example would be smearing. You can't spot that in individual screenshots, it's an effect you only see when watching the video.
Yet another example is wobbling, or texture movements.. scenes where you feel that an area is moving that should not be in movement. This usually happens when there's a lot of noise in the source. Individual screenshots may not bad and you'll certainly never get the idea that anything is off unless you watch the video at full speed.

These two kinds of problems make a video review with screenshots and without sample video really hard.. one the one hand you try to pick screenshots that are representative of what you see when watching the video, on the other hand you have to write about phenomenon you only see in the video, not the screenshot.

Go look at the H.264 Clip of the same trailer that Sagittaire has produced as an example of perfect quality at 1250.I wouldn't call that perfect by a long shot ;)

digigami

25th December 2005, 17:59

The high profile encoder is currently only available from our pc encoder suite (which includes much more than the avc encoder). It aims professional market. The indicative price for a single seat license is 9k euros. It supports interlaced encoding. This high profile encoder is expected to be integrated into Nero Digital in the near future.

OK, your pro version looks quite nice. GUI etc.

Our "Pro HD" encoder which can do HD MPEG-1 and MPEG-2, is $299.

So for 20x the price, you can have files 1/3 the size with a sharper picture (my opinion after looking at the files). I think this is a fair trade. Our 299 package also handles interlaced video and provides a wide range of filtering options, a very nice GUI etc.

Given the vast storage space of next gen optical discs, I think we have demonstrated that MPEG-2 is viable, and that H.264 is a better performer. But is also clear that the price for that performance is Ateme.

I have not auditioned JohnV's Main Profile streams yet, but I am sure they are very good.

So for professionals, at this time, Ateme Pro is the only real choice. Any opinions on Squeeze and some of the other H.264? My guess is they are commercializing open source compressors. Sorenson took H.263 and called it "Sorenson". Also what DiVX is doing I understand. May these companies die an ugly death at the hand of universal standards!

Sharktooth

25th December 2005, 18:00

I wish to see an example of H.264 Ice Age (Kung Fu) that does not exhibit block artifacts in the cave sequence. In the Apple trailer there is banding, so banding is AOK.
If the source is flawed the encoder cant to miracles. However reducing or turning off the psychovisual enhancements should lower the artifacts on dark areas.
Using a more "sensible" AVC quantization matrix for low luma should reduce the artifacts as well (AVC matrices are much more complex than mpeg2 quant matrices).
H.264 default quantization is far from perfect... it's a general purpouse quantization (as h.263 in MPEG-4 ASP).
But i feel we'll fall in an infinite discussion if we start to talk about custom quant matrices...

digigami

25th December 2005, 18:01

I wouldn't call that perfect by a long shot ;)

Yes, now that we have gone over the "my bitrate is lower than yours", I am interested to find out what the bitrate cost is for 720p with no artifacts. And it looks like the only one standing is Ateme, so whoever has access to that, perhaps they could encode the Ice Age trailer without artifacts and post it.

Thank you. In the meantime, I will give Recode a try while I am waiting for Nero to send me a license key.

JohnV

25th December 2005, 18:07

Imo you can discuss until the world ends if some encode has artifacts or not and how visible those are, unless you do lossless... It quite subjective. Some people might not see all artifacts, some are very sensitive and have huge LCD HDTV screens.
I agree with Doom9 that it's better to use one bitrate for both codecs and then judge the quality.

digigami

25th December 2005, 18:09

Actually, it's not so ridiculous at all. On the one hand, it's true, a picture is no true representation of what you percieve while watching the video, but that can go both ways. On the one hand, you are likely to miss individual artifacts in moving video, on the other hand, screenshots cannot show you changes over time. Take what I like to call the "keyframe readjustment" issue. While the term isn't technically correct, the phenomenon is that from one frame to another there's a significant change between two frames. So, while a screenshot from frame X and frame X+1 may look just fine, if played after another, you'll notice something is off.

I totally agree with you. He was talking about ringing and pointing to block artifacts. For what you are talking about, auditioning our MPEG-1 on Windows Media will make the phenomenon you describe very noticeable. This is related to IDCT performance, which is why it's harder to see in VLC.

If it were an encoder wart, it would be bitrate control. In fact, if you use the QT MPEG-4 encoder at low bitrates you will see strobing related to rate control problems.

While I agree with you that there can exist such temporal artifacts, I dont' see anything significant in our clip. Aside from what I am doing right now, my primary time is spent looking for artifacts produced by our encoder, and I have a pretty good eye for them, spatial, temporal. I just watched our Ice Age .m1v stream in VLC at a distance of 2.5 feet on my powerbook and I saw some slight blocking in the cave sequence and some slight strobing out of the Region of Interest in the final credits. And I can see artifacting in the H.264 sequence in the cave. But the assertion that it doesn't look HD is just ridiculous. At the top of the thread there are posters who think our HD streams look just fine.

I think the H.264 shown to me looks great, as well. It has a bitrate advantage, it has a detail advantage at those bitrates. But in the 15-27 mbits/s range afforded by next-gen optical discs, I think these advantages diminish. I also fielded an email from someone who believes at the higher bitrate MPEG-2 has the visual advantage. I have no comment on this, but the writer demonstrated a fine knowledge of compression.

And I very much would like to see how Ateme performs on interlaced material. Like football(soccer) and american football/basketball. 1920x1080x60i would be a nice test.

Anyone aware of such a sample or can make one?

IgorC

25th December 2005, 18:18

I totally agree with you. He was talking about ringing and pointing to block artifacts. For what you are talking about, auditioning our MPEG-1 on Windows Media will make the phenomenon you describe very noticeable. This is related to IDCT performance, which is why it's harder to see in VLC.

I said it was VLC player.

CruNcher

25th December 2005, 18:22

@digigami
Sorry but you can't say every time if someone sees standard Artifacts coused by the compression mathematics itself that IDCT in their player is allways @ fault that is scientific nonsense their will allways be such problems visible in close inspection of the compressed material thats why it's lossy compression. The difference between Mpeg-2 and H.264 is just how much not that one of them completly eliminates them that is as we all know impossible with lossy compression and in that regards even ASP is better then Mpeg-2.

IgorC

25th December 2005, 18:26

Guys, what we are discussing about?
just look at digigami's video. http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v. He said there is no visual artefacts.

just look . There are obvious artefacts that I can see during playback via VLC. And don't say me that those artefacts were here in QT 1080p source.

digigami

25th December 2005, 18:26

I agree with Doom9 that it's better to use one bitrate for both codecs and then judge the quality.

At this point, since I've already described what I have seen as being superior at low bitrates, I am most interested in seeing H.264 at lossless quality, and how much extra bitrate is required to get there.

Also, in the Apple HD content derived from CG (like Ice Age), there is noticeable aliasing on edges. I am sure this is not from the sources, since the pure CG animation folks like to make things look organic. Anyone know if this is a by-product of one of the coding transforms?

So we are clear about the cost of the superior quality and bitrate: from the compatibility matrix it is evident that the support for all H.264 AVC features is spotty across the board, which suggests a fair comparison is an H.264 file which can play across all the PC decoders (excluding QT) without playback glitches. Because if there are glitches, I say it's not yet standardized.

And I re-iterate this especially for the commandN sample, which was aimed at broad distribution, cross-platform to ordinary readership. I think the QT H.264 profile should be the baseline comparison for that clip.

Gen

digigami

25th December 2005, 18:30

Guys, what we are discussing about?
just look at digigami's video. http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v. He said there is no visual artefacts.

Go back and read my first post about our MPEG-1 clip. I said there were block artifacts from the onset. However, I do not believe they detract from the presentation.

just look . There are obvious artefacts that I can see during playback via VLC.

At what distance from the monitor are your eyes? You do understand that your eye has a limited ability to perceive spatial resolution which diminishes with distance?

digigami

25th December 2005, 18:32

@digigami
Sorry but you can't say every time if someone sees standard Artifacts coused by the compression mathematics itself that IDCT in their player is allways @ fault that is scientific nonsense their will allways be such problems visible in close inspection of the compressed material thats why it's lossy compression.

I did not say this. I said that temporal artifacts due to IDCT accuracy would be much more evident on Windows Media Player due to (what I believe) is decreased IDCT accuracy.

I also said that such artifacts can be the result of bitrate control problems.

I think I have a grasp of the science here. I'm not sure that all of the H.264 advocates have a grasp of their emotions.

IvS

25th December 2005, 18:32

The aliasing may in fact be the CG people's fault. Rendering such complex 3D at such a huge bitrate takes a huge amount of time. So maybe they chose to have low antialiasing like 4x or 8x, and not 16x which is of course by far the most optimal (as far as quality goes) and best for video codecs as well.
Or maybe the HD movie's source was a higher resolution completely non antialiased source, who knows?

Doom9

25th December 2005, 18:35

Mainconcept's encoder also does a good job, at $599. And x264 is free of charge. Just because there's no GUI and whatnot, shouldn't it be about quality first and the rest second? I'm still working on the final of my codec comparison but the final call on whether ateme or x264 does better is not in yet.. so I'd say it's fair to say there's one publicly available encoder that works on different platforms and delivers excellent quality.

Also what DiVX is doing I understand.Huh? DXN implements MPEG-4 level2, simple and advanced simple profile. QuickTime includes another implementation of that standard, just offering Apple's usual exceptional quality ;) (sarcasm turned all the way on..)

And I would be very careful before alleging that any company is selling open source products and violate open source licenses, because not only do you completely disqualify yourself, but you bring potential legal problems my way.. I don't want companies starting slander lawsuit and send subpoenas my way to get access records.

digigami

25th December 2005, 18:35

I said it was VLC player.

At what distance from VLC player's display are you sitting. Is the playback enlarged to the full screen? And what is the dimension of the screen in inches?

If you're going to say there are obvious artifacts in my clip, please qualify it.

I have 4 machines here to look at this stuff on, have a good eye for artifacting, and have no trouble reporting it for both MY clip and the H.264 clip. You have only demonstrated a willingness to look at mine. Look at the H.264 clip. Or say you lack objectivity. Either is fine with me.

digigami

25th December 2005, 18:36

And I would be very careful before alleging that any company is selling open source products and violate open source licenses, because not only do you completely disqualify yourself, but you bring potential legal problems my way.. I don't want companies starting slander lawsuit and send subpoenas my way to get access records.

This is a public record, just google it. UPDATE: I did for you:
http://weblogs.asp.net/jezell/archive/2004/04/30/123529.aspx

Also I did not say they were in violation of an open source license. Rather, the history of the company was codec development/invention. I just found it interesting that they chose to adopt a standard rather than continuing to improve their own.

IgorC

25th December 2005, 18:36

Sagittaire

25th December 2005, 18:37

There was also a MPEG2 test. http://forum.doom9.org/showthread.php?t=101735. Where mencoder has shown very good results at low bitrate. Maybe it is will be usefull to see. There's nothing new that MPEG-2 can provide acceptable quality at low bitrates as 3.6-5 mbps for 720p but it's still not on par with H.264.

Be carefull ... it's always challenge for 720*576 DVD MPEG2 MP@ML stream. I use HD source for to have best possible source ... :readrule:

Go look at the H.264 Clip of the same trailer that Sagittaire has produced as an example of perfect quality at 1250.

certainely not ... IMO perfect is only lossless. It's an example for you. H264 can make good HD encoding in 1-2 Mbps interval.

I wait always the encoding with MPEG1/2 with the same source and the same bitrate (it's useless to try ... MPEG1/2 are unable to make that)

I think that it is useless to continue the discution. We proved what we want : H264 is by far better than MPEG2

I think that you should satisfy to prove that your codec is best MPEG2 codecs and only that. That would be a great achievement for you and your encoder.

JohnV

25th December 2005, 18:37

digigami

25th December 2005, 18:41

there will allways be such problems visible in close inspection of the compressed material thats why it's lossy compression.

I agree with you there. At this point, I am interested in seeing an Ateme encoded version of Ice Age (Kung Fu) at lossless quality.

Now that I know Nero is using Ateme and have purchased it, I can learn the rest for myself.

I have found this discussion to be quite informative. And as I said before, I am looking for the right h.264 encoder for:
http://www.digigami.com/moviesformypod/

Before this thread, I didn't really have an idea of who to ask about what was good. Now I know who to ask and have a clue.

IgorC

25th December 2005, 18:44

Distance between my eyes and monitor 15' 1024x768 fullscreen mode is full lenght arm (from sholder untill fingers). I evaluted whole picture and not each marcoblock separatly. I can see obvious visual macroblocks. If you don't see them I don't know what to tell you.

I understand what you are talking about some people evaluate quality of video by each macroblock and imo it's not correct. But it's not the case.

digigami

25th December 2005, 18:46

bond

25th December 2005, 18:46

so yeah there has been a lot of talk till now, but i still dont see one single independant sample encoded with both digigami and a good h.264 encoder! or did i miss something?

i saw some clips proposed for such a comparison but i never saw digigami actually making an encode with them! why?

digigami

25th December 2005, 18:50

Imo more artifacting than in the 1.25Mb AVC streams, and I don't know if it's just me, but imo it also looks slightly more blurry.

It is definitely less sharp than the H.264.

Downstairs my friends have a large plasma TV (5' across) with HDTV service and sitting on their viewing couch I do not believe these artifacts would be visible. But from 15" this is another matter altogether. I do not watch movies this close unless I am on a plane.

digigami

25th December 2005, 18:51

IgorC

25th December 2005, 18:52

Thank you. Now please evaluate the H.264 stream and tell me if you see the artifacts in the cave sequence (1-2s duration) near the last 3/4 of the running time.

yes. QT7 h.264 isn't ideal source but there aren't so obvious artefacts like on frame 560 where appears blue sky. It's most annoying artefactas I've seen on mpeg-2 video and there aren't on Qt7 1080p source.

CruNcher

25th December 2005, 18:52

Sorry but Quicktime as it is now doesn't use the full Potential of H.264 yet sooner or later it has to be HD-DVD and Blue-Ray compatible as both require High Profile and that's what Hollywood is gonna use. Quicktime as it is used now is just a Lower Base for Online Distributeing Trailer and TV Shows (and one of the badest solutions) and im not gonna wait for Apple till they Advance if they don't sooner or later Show High Profile support and a more Advanced Encoder they will be out of the Game for Professionals. I don't understand how they can Advertise with HD-DVD support for DVD Studio Pro 4 if they don't meet the Full Criteria for it and Frext is definatly one.

And now encode the Tractor source Gen it is loosless source coming from a Professional HD Cam and was made especialy for Testing Efficiency of Compression standards. The clips from TUM where also used in the H.264 verfication process.

http://www.itl.nist.gov/div895/papers/IBC-Paper-AVC%20VerifTestResults.pdf

digigami

25th December 2005, 18:57

We proved what we want : H264 is by far better than MPEG2

Good for you. But even better for the designers of the codec.

My request for a lossless version of Ice Age H.264 AVC is to establish the true bitrate that will be used to encode HD-DVD and Blu-Ray. I do not believe anyone will professionally encode for those formats at 1250.

Is there anyone here who can answer this using Ateme? How about the Ateme guy...

digigami

25th December 2005, 19:03

And now encode the Tractor source Gen it is loosless source coming from a Professional HD Cam and was made especialy for Testing Efficiency of Compression standards. The clips from TUM where also used in the H.264 verfication process.]

What PC program will allow me to transcode this to a YUV format readable by QT?

Sagittaire

25th December 2005, 19:05

My request for a lossless version of Ice Age H.264 AVC is to establish the true bitrate that will be used to encode HD-DVD and Blu-Ray. I do not believe anyone will professionally encode for those formats at 1250.

Transparant quality is subjective ... try yourself with Nero Recode or MeGui (Gui for x264 ... free and 0€).

MeGui can encode with special QT7 profil ... QT7 player is simply not compliant with the complete norme. It's not a H264 problem but only Apple problem ...

digigami

25th December 2005, 19:06

I don't understand how they can Advertise with HD-DVD support for DVD Studio Pro 4 if they don't meet the Full Criteria for it and Frext is definatly one.

Are you saying that HD-DVD will not play the streams currently produced by QT H.264? Or that HD-DVD supports a superset? It would seem normal that HP decoders also play less featured profiles.

Gen

digigami

25th December 2005, 19:08

Transparant quality is subjective ... try yourself with Nero Recode or MeGui (Gui for x264 ... free and 0€).

OK, I have Nero now and was only asking because Ateme HP is not available.

Sagittaire

25th December 2005, 19:09

H264 MP is very good too ... try and see yourself.

IgorC

25th December 2005, 19:09

bond

25th December 2005, 19:12

The Ice Age sample chosen by Sagittaire was encode by him with Ateme ($9K) and by me with MegaPEG.thx, where can i download both clips and the source?

digigami

25th December 2005, 19:14

If the source is flawed the encoder cant to miracles.

No argument there. In this case, the Apple source exhibits light banding, so I think a faithful representation of the banding would be what we want.

IgorC

25th December 2005, 19:14

http://www.apple.com/trailers/

digigami

25th December 2005, 19:15

thx, where can i download both clips and the source?

The source is on the Apple Trailers site. There are two Ice Age trailers so download them both.

Regarding the compressed clips, you'll need to read the thread. Sagittaire's clips are up near the top, mine is around 2/3.

Speaking of clips, I have a lot of stuff to download, and the one web-based download services with the 60s delay is a bit of a hassle.

digigami

25th December 2005, 19:17

H264 MP is very good too ... try and see yourself.

Will do.

BTW, all your clips for your MPEG-2 tests are 401 - file not found.

digigami

25th December 2005, 19:19

Ateme HP encoder will be enable in Nero7 (approx. less than 100$ ) package in 2006 , while your MPEG-2 codec cost 300$ . I would prefer 0$ mencoder another excelent MPEG-2 with the same or better quality.

That would be up to you. Our encoder has other features for $300.

Referring to products that ship next year is called "vaporware".

IgorC

25th December 2005, 19:22

Referring to products that ship next year is called "vaporware".

There is already quite stable Ateme encoder HP version. Some betatesters (most of guys here ;) ) have already tested. It will be here soon :)
there is also x264 opensource with a great quality.

CruNcher

25th December 2005, 19:25

@digigami
you could use a raw to avi converter or avisynths raw source plugin

here is the complete verification report
http://www.chiariglione.org/mpeg/working_documents/mpeg-04/avc/avc_vt.zip

as you can see @ that time 2003 HD superiority over Mpeg-2 was by arround the factor of 1.7x today in 2005 we reached the factor of 2x and 3x and it's still improved by different companies as Ateme, Mainconcept/Elecard, Videosoft and many others and not to forget OSS projects like X264 with still projects to come.

IgorC

25th December 2005, 19:27

Our encoder has other features for $300.

Like what? Mencoder MPEG-2 is on par with your encoder and it's opensource.

Other features are enable in other open source projects. Like audio mp2, tool to work with MPEG-2 streams and etc.

savage747

25th December 2005, 19:28

Since I am in the US, perhaps you can give a breakdown of what HD formats are available via DVB in your area.

HD television just launched regular operation here in Germany. There have been some experimental broadcasts using DVB-S using MPEG2 - but they now are discontinued. All HD broadcasts (I am aware of) now use DVB-S2 (more efficient version of DVB-S - squeeze more bandwith out of one satellite transponder) and h264 - oh, and "nice" encryption.

Basically those programs have zero audience because fitting settop boxes are not available yet (shall be in Q1 2006).

Well, at least they start with the most modern technology... no MPEG2 HD in mid-europe.

<off-topic>
Oh, I took the liberty of encoding some clips with libavcodec single pass (-lavcopts vcodec=mpeg2video:aspect=16/9:vbitrate=2000:vhq) and this already gives much better quality then what can be seen on most DVB-T broadcasts... I wonder what hellspawn of an encoder they use.
</off-topic>

digigami

25th December 2005, 19:36

yes. QT7 h.264 isn't ideal source but there aren't so obvious artefacts like on frame 560 where appears blue sky. It's most annoying artefactas I've seen on mpeg-2 video and there aren't on Qt7 1080p source.

So now my question is what video card and if you know, whether VLC is using the video card for IDCT or using the built-in routines. I say this because when I look at VLC in 082 and 084a - not a problem. 24" viewing distance, approx 15" window on 17" display LCD screen. Mac gamma.

On PC, 24" from screen. 17" LCD.

Here are my audition results on Nero Showtime. 17" laptop, full screen playback.

24" H.264 is sharper. Both H.264 and MPEG-2 noticeable banding/blocking and blocking in the blue background behind fight scene and at end when elephant's ass is walking away.

36" H.264 clarity not so evident. Artifacts in cave (dark) not visible on either clip. Blocking/banding in blue sky/blue cave visible in both clips.

digigami

25th December 2005, 19:38

and not to forget OSS projects like X264 with still projects to come.

And not to forget...Digigami! ;)

BTW, all you H.264 proponents - you can now consider me to be H.264 proponent. Although I am sure you will not mind if we continue to improve our MPEG-2 encoder.

Sharktooth

25th December 2005, 19:41

Good for you. But even better for the designers of the codec.

My request for a lossless version of Ice Age H.264 AVC is to establish the true bitrate that will be used to encode HD-DVD and Blu-Ray. I do not believe anyone will professionally encode for those formats at 1250.

Is there anyone here who can answer this using Ateme? How about the Ateme guy...
no-one will encode professionally HD stuff with your or any other mpeg-2 at 3.3mbps or even at 8mbps as well...
however h.264 can be LOSSLESS too (that's different from transparent).
a professional h.264 encode with all the quality needed for distribution will be something around 3000kbps but you will not notice ANY kind of artifacts even with a frame by frame comparison unlike some actual DVDs.

digigami

25th December 2005, 19:41

<off-topic>
Oh, I took the liberty of encoding some clips with libavcodec single pass (-lavcopts vcodec=mpeg2video:aspect=16/9:vbitrate=2000:vhq) and this already gives much better quality then what can be seen on most DVB-T broadcasts... I wonder what hellspawn of an encoder they use.
</off-topic>

Yes, OK, now do it in real-time with capture. Most real-time encoders use 1/2 width frame for motion compensation and do exhaustive search. Also, make sure you are using MPEG standard quant matrices, because I think it is rare to find broadcast with anything else. DVD is another matter.

digigami

25th December 2005, 19:45

mpeg-2 at 3.3mbps or even at 8mbps as well...

You are right about that. I recommend to customers to use the highest bitrate they can possibly use. Blu-Ray as a 54 mbits/s ceiling and a big drive capacity. Looking forward to that...

HD-DVD looks like around 27 mbits/s ceiling and 5x the size of DVD5.

IgorC

25th December 2005, 19:45

So now my question is what video card and if you know, whether VLC is using the video card for IDCT or using the built-in routines.

What? Now you are saying that´s matter of harware. mpeg-2 has artefactas on my system due to my old Geforce 64 ddr videocard while H.264 is looking better. Is it your point?

Artefacts has nothing to do with hardware. QT7 source video has not such artefacts.

digigami

25th December 2005, 19:48

Thanks for all your posts. I'm going to play around with Nero MP and x264.

Like I said earlier, I don't mind being bested by H.264 - and I am glad to have learned that Ateme is a professional solution for the PC.

I will be surprised if they put a $9000 encoder into $100 program at this point in the adoption curve. I guess they worry about x264!

Sharktooth

25th December 2005, 19:49

You are right about that. I recommend to customers to use the highest bitrate they can possibly use. Blu-Ray as a 54 mbits/s ceiling and a big drive capacity. Looking forward to that...

HD-DVD looks like around 27 mbits/s ceiling and 5x the size of DVD5.
i corrected my sentence.. i didnt mean "your"... i meant mpeg-2 in general.
Thanks for all your posts. I'm going to play around with Nero MP and x264.

Like I said earlier, I don't mind being bested by H.264 - and I am glad to have learned that Ateme is a professional solution for the PC.

I will be surprised if they put a $9000 encoder into $100 program at this point in the adoption curve. I guess they worry about x264!
They already did it with the main profile encoder.
however they dont provide customer service/care for nero customers... it's up to nero... and nero customer service is aimed at non-professional audience.

digigami

25th December 2005, 19:51

What? Now you are saying that´s matter of harware. mpeg-2 has artefactas on my system due to my old Geforce 64 ddr videocard while H.264 is looking better. Is it your point?

Artefacts has nothing to do with hardware. QT7 source video has not such artefacts.

Do you understand that different video cards can handle different amounts of the MPEG decoding pipeline? On the Mac, with the DVD framework, almost the entire decode is done by the hardware. VLC does not access this AFAIK.
QuickTime MPEG-1 does not use this. Thus, on my PowerBook, each of these 3 MPEG decoders gives different results. My PC laptop has high end video card, probably does the entire MPEG decoding in hardware.

http://www.ati.com/products/dtv/revolutionary.html
google search MPEG decoding in hardware ATI video cards (http://www.google.com/search?q=hardware+mpeg+decoding+site:ati.com&ie=UTF-8&oe=UTF-8)

Do you understand?

Sagittaire

25th December 2005, 19:52

I will be surprised if they put a $9000 encoder into $100 program at this point in the adoption curve

Ateme encoder is professionnal encoder (interlaced support with transport stream output for example) certainely for industry with small potential client numbers.

Nero suite are not for the same objective ...

digigami

25th December 2005, 19:52

digigami

25th December 2005, 19:53

Ateme encoder is professionnal encoder (interlaced support) certainely for industry with small potential client numbers.

Nero suite are not for the same objective ...

So the HP in Nero will still not represent the full AVC capabilities? Progressive encoding only? What about other features. Do you know?

IgorC

25th December 2005, 19:53

I will be surprised if they put a $9000 encoder into $100 program at this point in the adoption curve. I guess they worry about x264!

I will be surpisied if you put a 300$ encoder into 5$ program. You would worry about mencoder.

IgorC

25th December 2005, 19:58

digigami. Nobody will use MPEG-2 to archieve HDs at such low bitrates 3.6-5 mbps. You can make a good marketing and propaganda here but quality isn't here.
You are only the one man here talking about MPEG-2 HD at 3-5 mbps, don´t use. We don't see another guy.

digigami

25th December 2005, 19:59

I will be surpisied if you put a 300$ encoder into 5$ program. You would worry about mencoder.

Do you see similar artifacts in H.264? Blue background during fight scene and at the end? If not, you are biased. I can see them in both clips.

Do you?

Sharktooth

25th December 2005, 19:59

So the HP in Nero will still not represent the full AVC capabilities? Progressive encoding only? What about other features. Do you know?
Yes it will but i suppose other "pro" features (TS output, interlaced encoding and other stuff meant for broadcast or professional use) wont be there.

digigami

25th December 2005, 19:59

digigami. Nobody will use MPEG-2 to archieve their HDs at such low bitrates 3.6-5 mbps. You can make a good marketing and propaganda here but quality isn't here.
You are only the one man here talking about MPEG-2 HD at 3-5 mbps, don´t use. We don't see another guy.

Do you see similar artifacts in H.264? Blue background during fight scene and at the end? If not, you are biased. I can see them in both clips.

digigami

25th December 2005, 20:00

Yes it will but i suppose other "pro" features (TS output, interlaced encoding and other stuff meant for broadcast or professional use) wont be there.

Interesting. Thank-you. I plan to try your quant matrices with our MPEG-2 just for kicks. 8x8 DCT is 8x8 DCT.

Sharktooth

25th December 2005, 20:01

Do you see similar artifacts in H.264? Blue background during fight scene and at the end? If not, you are biased. I can see them in both clips.
Yes artifacts are there, but the h.264 clip still looks better.
My matrices are for xvid (mpeg4 ASP)... i dont guarantee good results for mpeg2.
Time to go now :)
l8r.

IgorC

25th December 2005, 20:02

QT7 source artefactas are nothing comparing to your MPEG-2 sample.
looks like you're nerving right now ;)

digigami

25th December 2005, 20:03

Yes artifacts are there, but the h.264 clip still looks better.

Thanks, but I am asking IgorC. He's not evaluating both clips, only mine.

And I agree, the H.264 looks sharper.

digigami

25th December 2005, 20:05

QT7 source artefactas are nothing comparing to your MPEG-2 sample.
looks like you're nerving right now ;)

I'm not talking about QT source. I'm talking about Sagittaire's H.264 AVC at 1250. Get a grip.

JohnV

25th December 2005, 20:05

So the HP in Nero will still not represent the full AVC capabilities? Progressive encoding only? What about other features. Do you know?
Even I can't say yet. But it should be the same codec with practically all features, interlacing included. Btw. IIRC interlacing is Main Profile feature anyway, not HP.

digigami

25th December 2005, 20:07

Yes artifacts are there, but the h.264 clip still looks better.
My matrices are for xvid (mpeg4 ASP)... i dont guarantee good results for mpeg2.
Time to go now :)
l8r.

OK - I will be investigating XviD too for MoviesForMyPod.

CruNcher

25th December 2005, 20:07

ok just on the matter of Hollywood and H.264 i think you remember the time when Blue-Ray said hey AVC Main Profile doesn't looks so good as our 20 Mbit Mpeg-2 Stuff

here is the official background to that matter and when Frext was introduced also it should be covered that Atemes Encoder i think as the only one in the world supports Film Grain Modeling in the Encoder/Decoder as a bonus to High Profile.

http://ftp3.itu.ch/av-arch/jvt-site/2004_03_Munich/JVT-K025r1.doc

digigami

25th December 2005, 20:08

Even I can't say yet. But it should be the same codec with practically all features, interlacing included. Btw. IIRC interlacing is Main Profile feature anyway, not HP.

Thanks JohnV, I am now bonafide Nero customer for NeroDigital, just waiting for key license key. Can you tell them to hurry? ;)

IgorC

25th December 2005, 20:09

Talking about Ateme H.264 , yes I agree with. but are you comparing Ateme H.264 video that´s 3 times smaller than your MPEG-2 video and you´re saying that H.264 is sharper than MPEG-2. Imagine what conlcusion you will do at the same bitrate. H.264 will be much better than MPEg-2.

digigami

25th December 2005, 20:10

digigami

25th December 2005, 20:12

H.264 will be much better than MPEg-2.

I guess you haven't been reading the thread, but we all agreed on that about 3 pages ago.

IgorC

25th December 2005, 20:15

I'm really happy now :)

savage747

25th December 2005, 20:34

Yes, OK, now do it in real-time with capture. Most real-time encoders use 1/2 width frame for motion compensation and do exhaustive search. Also, make sure you are using MPEG standard quant matrices, because I think it is rare to find broadcast with anything else. DVD is another matter.

Well, my encoding speed was something like 64fps (Athlon64 3000+) - so this gives plenty of headroom for capturing. I didn't use a custom quantization matrix (most likely http://www1.mplayerhq.hu/cgi-bin/cvsweb.cgi/ffmpeg/libavcodec/mpeg12data.h?rev=1.28&content-type=text/x-cvsweb-markup&cvsroot=FFMpeg was used - is that standard?). In my experience those files can be played with everything that claims to be able to decode MPEG2... I couldn't find the exact spec (GOP length etc.) for DVB by a simple google query, though (they should be over at ETSI, though - but their site is a mess). I did another quick test (forcing keyframe interval to 10 - which most likely is more restrictive than needed) and the result still looks way better than what can be seen on DVB-T... so I think that much could be done to improve picture quality if the broadcasters would care.

Meh, think I'm fully off-topic now... (SD TV over DVB)

Sagittaire

25th December 2005, 21:27

And finally (for me) here little comparison

your best possible encoding (I imagine) at 3800 Kbps
http://digigami.in-long-beach-ca.com/megapeg/Doom9/ice_age_2-tlrD_720p.010.m1v

and my best possible encoding at 3800 Kbps for libavcodec MPEG2 encoder
http://multimediacom.free.fr/Download/MPEG2-HQ5.m2v

Sorry ... but as you can see there are no possible discution ... ;)
Before this thread I thought that Libavcodec was the best MPEG1/2 encoder in the world.
After this thread I think always that Libavcodec is the best MPEG1/2 encoder in the world.
After this thread I think always that H264 is very better than all MPEG1/2 encoder in the world.

PS: Setting for Libavcodec:
- GOP PAL compliant 15 frames max
- 2 Bframes - No adaptative with low ratio
- 128 for Max MV range
- VBV compliant with DVD format
- 20 Mbps for Max Bitrate
- Highest possible ME search and RDO search
- Masking HVS optimisation
- Multipasse Encoding
- Sagittaire Custom Matrix

digigami

25th December 2005, 21:55

And finally (for me) here little comparison
and my best possible encoding at 3800 Kbps for libavcodec MPEG2 encoder http://multimediacom.free.fr/Download/MPEG2-HQ5.m2v

Sounds great. Taking a look at it now.

First try with Nero ReCode:
12:09:24 File burn_ndigital.cpp, Line 722
Nth-pass file = C:\DOCUME~1\gen\LOCALS~1\Temp\\ice_age_2-tlrD_1080pa02716.bin

ERROR in file:E:\Sources\Current\Recode 2.2 Compare\Nero Recode II\burning\burn_ndigital.cpp, line:1884, error:Unspecified error (0x80004005)
12:09:24 File burn_ndigital.cpp, Line 200
NeroDigital: exit

digigami

25th December 2005, 22:07

And my best possible encoding at 3800 Kbps for libavcodec MPEG2 encoder
http://multimediacom.free.fr/Download/MPEG2-HQ5.m2v

Very nice Sagittaire. You have demonstrated, beyond a doubt, that MPEG-2 VBR has unlimited potential for HD-DVD and Blu-Ray.

X-mas today, I got 3 gifts for myself. 1) H.264 spec from ITU webstore. 2) NeroDigital (still waiting for license key) and 3) confirmation that VBR MPEG-2 is great for HD.

Thank you all!

PS.
svn co
./configure
make

x264 build succeeded.

digigami

25th December 2005, 22:12

Best possible encoding at 3800 Kbps for libavcodec MPEG2 encoder
http://multimediacom.free.fr/Download/MPEG2-HQ5.m2v

Now, I wait for someone to beat Sagittaire MPEG-2 VBR at 3800 with H.264 AVC - and no blocking artifacts.

Only small blocking in Sagittaire. Fade in to 20th Century Fox logo - difficult to see. His encode is almost perfect constant Q at 6, very high quality. Superbit DVD is 1-3 with very low Q-matrix. So at 3.8 mbits/s MPEG-2 VBR is very, very good for 1280x720x25p. At 10 mbits/s I think perfect quality is possible. Less than 1/3 of HD-DVD bitrate and 1/12 of Blu-Ray. MPEG-2 c'est bon, n'est pas?

Sagittaire - c'est champion. Avec custom matrix. C'est bon. J'attend pour H.264 at 3.8 mbits/s.
http://digigami.in-long-beach-ca.com/megapeg/Doom9/sagittaire-champion.png

Sagittaire

25th December 2005, 22:32

Now, I wait for someone to beat Sagittaire MPEG-2 VBR at 3800 with H.264 AVC - and no blocking artifacts.

But there are blocking artifacts on this MPEG2 encode (because average quant 6). You are really incredible ...

If I say H264 is better it's because I was already test the best MPEG2 encoder (IMO libavcodec) and the best H264 encoder (IMO Ateme HP).

The ratio quality is variable (source and level quality) but there are certainely something like 2.0 for ratio for this source at this quality level.

IgorC

25th December 2005, 23:01

Actually I agree with Sagitarrie. Libavcodec (enable in mecnoder) MPEG-2 provides best possbile for MPEG-2 quality. But it's still worse than H.264 High Profile.

digigami

25th December 2005, 23:14

I'm really happy now :)

Pour IgorC - la question d'accuracy de 'matices de quant'. Excuse-moi, mon francais est terrible.
-----
For IgorC - why I asked you questions about your system trying to determine the accuracy of the IDCT used by VLC.

This clip is from Sagittaire - Harry Potter trailer ripped from DVD, showing 4 different IDCT implementations (made png with Photoshop):
MPEG-2 decoder IDCT (http://local.digigami.com/mpressionist/screen.php?screen=idct-accuracy) accuracy comparison.
Notice AAN is very famous IDCT implementation, but compared to ChenSmith97, not so good for MPEG decoding. What IDCT does x264 use?

Windows Media Player is especially poor IDCT performance. VideoLAN is good. Hardware is?.... hardware dependent.

digigami

25th December 2005, 23:16

Actually I agree with Sagitarrie. Libavcodec (enable in mecnoder) MPEG-2 provides best possbile for MPEG-2 quality. But it's still worse than H.264 High Profile.

Yes, but a viable choice for HD-DVD and Blu-Ray which allow for 27 mbits/s MPEG-2 and 54 mbits/s MPEG-2 respectively.

J'attend pour H.264 AVC Ateme a 3.8 mbits/s.
----
I wait for a H.264 AVC Ateme at 3.8 mbits/s. Better yet, do one at 3.8 and one at 6 mbits/s.

digigami

25th December 2005, 23:20

Actually I agree with Sagitarrie. Libavcodec (enable in mecnoder) MPEG-2 provides best possbile for MPEG-2 quality. But it's still worse than H.264 High Profile.

Mais, MPEG-1 avez Q-2 stepsize seulement. Je pense que digigami avez une autre chance.

------
But, MPEG-1 has Q2 stepsize (in quantizer) only. I think that digigami has another chance. We have feature classification like libavcodec, but not I-frame injection (cut scene detection).

http://www.digigami.com/megapeg/blog/2005_03_16_archive.html

IgorC

25th December 2005, 23:23

Pour IgorC - la question d'accuracy de 'matices de quant'. Excuse-moi, mon francais est terrible.

You're violarion rules of this firum speak i english

I also can say you some pair of words on spanish,portgues, russian, tatar.

digigami

25th December 2005, 23:24

But there are blocking artifacts on this MPEG2 encode (because average quant 6).
Oui. Je pense que vous encodez une autre MPEG-2 VBR libavcodec a 6 mbits/s?
---
Yes. I think that you encode another MPEG-2 VBR with libavcodec at 6 mbits/s? I think it will be perfect quality for HD-DVD and Blu-Ray. At 6 mbits/s! Do you agree?

You are really incredible ....
Merci beaucoup. You are incredible yourself. Much better than Waaaggggonnnner. You should write for DV magazine.

Je pense que libavcodec utilizez I-frame injection and exhaustive ME? Oui?

digigami

25th December 2005, 23:26

You're violarion rules of this firum speak i english

I also can say you some pair of words on spanish,portgues, russian, tatar.

Pobre sos vos pelotudo de mierda. Ya se que cuestion de accurancy esta en matrices de quantizacion. Disculpa mi español. Pero la concha de tu madre era una puta y vos sos un puto del orto.

C'est bon, parce-que I do not speak french! Barely.
----
That's good, because I do not speak french! Barely.

IgorC

25th December 2005, 23:31

digigami

25th December 2005, 23:34

Read the rules http://forum.doom9.org/forum-rules.htm

You are violating folowing rules
3. Keep the focus: Questions outside the scope of a certain forum will either be moved, locked or simply be deleted.

4. Be nice to each other

13. The official language is English

OK, I have translated my minor french statements. Also, you will see all my posts are related to the discussion...

Also, do you notice I am H.264 advocate now? We are all working on the same thing. Except I am mostly mac these days:
http://www.digigami.com/eurekagram/pdf/1993_11_SD_Hacking.pdf

digigami

25th December 2005, 23:53

and my best possible encoding at 3800 Kbps for libavcodec MPEG2 encoder
http://multimediacom.free.fr/Download/MPEG2-HQ5.m2v

Sagittaire - I believe this is not your _best_ possible. The source is 24fps and you have encoded at 25fps. Your encoder is time-resampling the frames which introduces extra motion blur. Encode again, everything the same except 24fps instead of 25 and you will have your best. Also, maybe enough to remove light blocking during fade in.

Sagittaire

26th December 2005, 00:25

Sagittaire - I believe this is not your _best_ possible. The source is 24fps and you have encoded at 25fps. Your encoder is time-resampling the frames which introduces extra motion blur. Encode again, everything the same except 24fps instead of 25 and you will have your best. Also, maybe enough to remove light blocking during fade in.

don't change anything for quality (except bitrate -> 3800 Kbps for 25 fps and 3648 bps for 24 fps) ... it's simply a speed frame conversion 24@25 fps in avisynth.

bond

26th December 2005, 00:41

guys, keep cool. no french plz

digigami

26th December 2005, 00:54

don't change anything for quality (except bitrate -> 3800 Kbps for 25 fps and 3648 bps for 24 fps) ... it's simply a speed frame conversion 24@25 fps in avisynth.

OK . 24@25 I understand. Any hits on using Recode?

I have no idea what I am doing wrong with Recode:
1. Import 720p trailer from Apple.
2. Select Cinema AVC profile
3. Clear 'fit to size of media'.
4. Click next.
5. Click burn or set to 1-pass. Always fails.

Project type: Nero Digital
Target size: CD80 (354000)

15:50:29 File LangStreamPatcher.cpp, Line 56
Command table patching started...

15:50:29 File LangStreamPatcher.cpp, Line 133
All tables successfully patched!

Burn settings:
Target: Hard disk folder
Number of copies: 1
Target folder: C:\Documents and Settings\gen\My Documents\Nero Recode\New Folder (57338826752 bytes free)
NeroDigital profile: Cinema - AVC
Shutdown when finished: false
Advanced Analysis: false
Compress Adaptive: false

15:51:09 File burn_ndigital.cpp, Line 158
NeroDigital: init

15:51:09 File burn_ndigital.cpp, Line 269
NeroDigital: burning 42632 sectors (42632 required)

15:51:09 File burn_ndigital.cpp, Line 343
NeroDigital: beginning 1st-pass operation

15:51:09 File burn_ndigital.cpp, Line 697
NeroDigital: encode ice_age_2-tlrD_720p

Source:
Size (1280,688)
Display Aspect Ratio (80,43)
Deinterlace disabled
Cropping enabled (0, 0, 0, 0)
Resizing enabled (1280,688)
Pixel ratio (1,1)
Bitrate: 8000000 bps

15:51:09 File burn_ndigital.cpp, Line 722
Nth-pass file = C:\DOCUME~1\gen\LOCALS~1\Temp\\ice_age_2-tlrD_720pa01908.bin

ERROR in file:E:\Sources\Current\Recode 2.2 Compare\Nero Recode II\burning\burn_ndigital.cpp, line:1884, error:Unspecified error (0x80004005)
15:51:09 File burn_ndigital.cpp, Line 200
NeroDigital: exit

Sharktooth

26th December 2005, 04:48

Seems the problem is "burn". Set up recode to encode to a file instead of burning to a disc.

digigami

26th December 2005, 06:55

Seems the problem is "burn". Set up recode to encode to a file instead of burning to a disc.

I'm not sure what I have has that option. I have the Hard Disk Folder option set, rather than a physical burner, but I cannot see an option to do something other than make a CD.

I am using the 30-day trial of Recode CE, because although I purchased it, they have not sent me an email.

Right now the command line tool (x264) is looking quite appealing.

UPDATE: I looked at various ripping guides for using Nero for iPod video, PSP, etc and other than changing the profile/target, I am using Recode correctly. Burn is indeed the button. I'm not sure that failing with a MFC ASSERT statement is "production software".

Sharktooth

26th December 2005, 14:34

You may want to adapt MeGUI to Mono so it can run on OSX. It should be quite an easy task coz Mono supports C#.
You can also try updating to the lastest Nero 7.0.1.4b just released...

digigami

26th December 2005, 14:39

My codec comparison sources. Look at my review scenes of Matrix3.. the source simply isn't good enough.

Can you post the forum url for this, thanks?

Revgen

26th December 2005, 22:10

@Digigami

I think that MPEG-1 as an HD rez clip will be good because of the lack of licensing fees for mass production, and because people with slower processors would be able to watch them.

But to say that MPEG-1 or MPEG-2 is comparable to AVC is not an accurate statement.

If I were you, I'd promote your encoders (especially your MPEG-1 encoder) as a cheaper, more compatible, and more processor friendly alternative for mass distribution. You won't get trashed as much, and you might be taken more seriously.

Just my 2 cents.

digigami

26th December 2005, 23:15

You may want to adapt MeGUI to Mono so it can run on OSX. It should be quite an easy task coz Mono supports C#.
You can also try updating to the lastest Nero 7.0.1.4b just released...

Nah. If I need a gui, I can whip one up right quick like. The fact that x264 built without a problem with configure && make is a big plus for getting started. I tend to go about things a little differently so my gui probably won't look like the other ones....

digigami

26th December 2005, 23:17

If I were you, I'd promote your encoders (especially your MPEG-1 encoder) as a cheaper, more compatible, and more processor friendly alternative for mass distribution. You won't get trashed as much, and you might be taken more seriously.
Just my 2 cents.

You think? Go up the to the top of the thread and read the first page.

I've been reading the specs for Blu-Ray and HD-DVD and I can tell you, Blu-Ray has the bitrate ceiling, and HD-DVD supports a really nice wide spectrum of frame sizes, so MPEG-2 is going to work great for both of those. But like I said, the intelligent posters here have won me over to x264.

digigami

26th December 2005, 23:58

But to say that MPEG-1 or MPEG-2 is comparable to AVC is not an accurate statement.

Well, I'm still working on getting to the bottom of that. Using x264 I have generated another 720p test of Ice Age that is about 26MB. It still has some strobing effects, partially because there is banding in the Apple trailer, but also because video compression is about illusion, not about perfection. I think if Sagittaire adds another 1-2 mbit/s to his MPEG-2 clip it will be hard to improve. Which is why I want to generate an H.264 clip that is free of artifacting, or an accurate rendering of the source. Instead of trying to get the bitrate as low as possible I am interested in the least bits while still studio quality results. What we have done thus far is simply started a discussion.

I am going to answer the H.264 questions to my own satisfaction. Now that I know what the best codecs are considered to be, I can produce my own tests and publish my findings. And at the same time I have discovered some passionate people who have an excellent understanding of the codec.

H.264 may be a ratified standard in print, but I think it has a ways to go before that is mirrored in the implementations. All across the board, perhaps except for Ateme Pro, which is unobtainium. Even if Ateme is perfect, I think the decoders also have some maturing to do.

If you read every item above this post, you will not see Apple described as being incapable of making a MP4 container with a good audio stream; only the limitations on their use of AVC HP. So why, then, does Nero Showtime not play audio when it auditions HD H.264 content from the Apple site? This kind of inconsistency is everywhere. There is no question that H.264 is a revolutionary technology - I just think it is a bit raw at this point.

I just read that Microsoft's newest codec is basically a dumbing-down of H.264 so to be playable on a wider class of hardware? Does anyone know if this is factual? If it is true, I think Apple's choice of using a subset of the standard to accomplish the same goal is an example of supporting the standard, not undermining it.

bond

27th December 2005, 00:04

If you read every item above this post, you will not see Apple described as being incapable of making a MP4 container with a good audio stream; only the limitations on their use of AVC HP.apple uses not really main profile avc in their encoder. they are using baseline profile + 1 b-frame

So why, then, does Nero Showtime not play audio when it auditions HD H.264 content from the Apple site? This kind of inconsistency is everywhere. There is no question that H.264 is a revolutionary technology - I just think it is a bit raw at this point.audio has nothing to do with avc of course...

also aac-in-mov is not the same as aac-in-mp4 of course, so a mp4 player doesnt necessarily play mov (or mpg if you like)

digigami

27th December 2005, 00:10

apple uses not really main profile avc in their encoder. they are using baseline profile + 1 b-frame

Yes, I am clear on that now.

audio has nothing to do with avc of course...
It does have to do with H.264. George Lucas "Sound is 50% of the movie experience".

also aac-in-mov is not the same as aac-in-mp4 of course, so a mp4 player doesnt necessarily play mov (or mpg if you like)
Although I haven't been tracking things closely, my understanding is that MPEG-4 adopted the QT file format. In 97 I wrote a cross platform tool to reorganize the QT atoms for streaming. It looks to me like the container is the same, what is the difference and why would Apple not put AAC into the MOV in the way expected for .MP4? Seems odd. Also, since Nero plays other QT content, why would this MOV be any different.

bond

27th December 2005, 00:25

It does have to do with H.264. George Lucas "Sound is 50% of the movie experience".lol, now what? avc is a video standard, aac is an audio standard, both are independant

eg avc can be placed in .mp4 without needing aac and the other way round...

Although I haven't been tracking things closely, my understanding is that MPEG-4 adopted the QT file format. In 97 I wrote a cross platform tool to reorganize the QT atoms for streaming. It looks to me like the container is the same, what is the difference.mp4 and .mov is similar but not the same container

for whats different, read the specs... but how aac is placed in the file is one example

and why would Apple not put AAC into the MOV in the way expected for .MP4? Seems odd. because apple is braindead?

digigami

27th December 2005, 00:38

because apple is braindead?

Given your opinions, you have an unusual sig.

Sharktooth

27th December 2005, 04:56

It's the usual story. Apple does things how they like to do them with no respect for standards and compatibility.

digigami

27th December 2005, 15:16

I can make the same MPEG2 encoding in 1080p with libavcodec with 5 Mbps I think (same metric). IMO libavcodec is the best MPEG2 encoder. I think that if you can prove that your encoder is better than libavcodec could be a good publicity for your encoder ...

Sagittaire - since you seem to know the most about MPEG-2 compared with H.264, can you make a comparision of each H.264 profile/AVC level vs MPEG-2 libavcodec? It does not need to be perfect, for instance:

H.264 AVC L 5.1 - 2.5 x better than libavcodec
...
H.264 PSP Profile - 2.0x better than libavcodec
H.264 iPod Profile - 1.5 x better than libavcodec

Or anyone else that knows, approximately.

digigami

27th December 2005, 15:17

It's the usual story. Apple does things how they like to do them with no respect for standards and compatibility.

Sharktooth, in your posts about your custom quant matrices you refer to the recommendations for H.264 from the JVT - other than the standards document, is there a place I can read about this?

digigami

27th December 2005, 15:24

Here sample H264 High Profil
King Kong (http://multimediacom.free.fr/Download/x264HP_KingKong_720p_1500.mp4)
1280*720*25
1500 Kbps

John, Sagittaire and anyone else who created H.264 samples - please post the AVC level, encoder/parameters used to encode the streams so others can duplicate the results.

digigami

27th December 2005, 16:49

It's the usual story. Apple does things how they like to do them with no respect for standards and compatibility.

OK, I'm doing some 1080p comparisons right now.

Nero H.264 sample 1080p "The Greatest Game": P4HT 3.06 Gz
* VLC 082 - does not play - massive glitches
* Nero Showtime - drops frames/audio

QuickTime H.264 Sample 1080p "Ice Age"
* QT Player - plays perfectly.

"The Greatest Game" is drama, almost no motion and very slow fades through black.

slavickas

27th December 2005, 17:33

why using VLC 0.8.2? 0.8.4 official release exists for some time

CruNcher

27th December 2005, 17:46

Digigami in Nero7 their is the newest Nero Decoder Nevideo.ax 4.2.2.3 and it has HT support (dunno if its also in the other packages but Nero sometimes don't keep them sync look in C:\Program Files\Shared Files\Ahead\DSFilter) for your Version it could be older unfortunately. In case you really have older Filters you can just download and install the Nero7 Demo and the newest update and voila newest filter ;)

VLC Win32 has massive problems with HD H.264 Decoding whyever this is the case.
Mplayer Win32 would be another good solution same as FFDshow
Also a very good new Decoder on "the market" is CoreAVC (unfortunately its only available @ the moment as a plugin for The Core Pocket Media Player) you could try this too.

The Greatest Game Ever Played playbacks here fine useing CoreAVC on my 2.4 GHz Athlon 64 Venice. With all other players i have droped frames/audio

digigami

27th December 2005, 18:20

why using VLC 0.8.2? 0.8.4 official release exists for some time

Because earlier in the thread 084a was not working and someone suggested 082

digigami

27th December 2005, 18:29

Also a very good new Decoder on "the market" is CoreAVC (unfortunately its only available @ the moment as a plugin for The Core Pocket Media Player) you could try this too.

The Greatest Game Ever Played playbacks here fine useing CoreAVC on my 2.4 GHz Athlon 64 Venice. With all other players i have droped frames/audio

I found Core Pocket Media Player but it is for handhelds. Do you have an URL for the desktop software?

Sagittaire

27th December 2005, 18:32

Sagittaire - since you seem to know the most about MPEG-2 compared with H.264, can you make a comparision of each H.264 profile/AVC level vs MPEG-2 libavcodec? It does not need to be perfect, for instance:

Test in progress for my mega-test codec but only for HD ...

sillKotscha

27th December 2005, 19:17

Do you have an URL for the desktop software?

Win32 Player + Plugins (http://picard.exceed.hu/tcpmp/test/tcpmp.win32.0.71d.zip)

You're welcome!

CruNcher

27th December 2005, 21:41

http://picard.exceed.hu/tcpmp/test/avc.win32.0.71e.zip <- CoreAVC and AAC plugin just put it in the same folder as the Player extracted files ofcourse :)

digigami

28th December 2005, 04:55

Win32 Player + Plugins (http://picard.exceed.hu/tcpmp/test/tcpmp.win32.0.71d.zip)

Danke schoen.

PicardGK

28th December 2005, 11:51

I have uploaded a new version of TCPMP and CoreAVC. Please use that. The old player had a buffering problem. Additionally I actually invested a little time to optimize the AVC decoder. Multithread and CQM support is still missing (which I plan to support someday)

http://picard.exceed.hu/tcpmp/test/tcpmp.win32.0.71g.zip
http://picard.exceed.hu/tcpmp/test/avc.win32.0.71g.zip

CruNcher: could you test the new version and compare the speed with the "Greatest Game Ever Played" clip? Select options/video/null and options/audio/disabled and use the file/benchmark menu. I'am interested how these changes effected the speed on AMD (I have only a P4).

CruNcher

28th December 2005, 13:01

TCPMP Version 0.71d Benchmark Results + CoreAVC 0.71e

Average Speed 141.03%
Video Frames 3673
Amount of Data 115601 KB

Bench. Time 1:48.625
Bench. Frame Rate 33.81
Bench. Data Rate 8.7 Mbit/s

Original Time 2:33.194
Original Frame Rate 23.98
Original Data Rate 6.1 Mbit/s

URL \mplayer2\TheGreatestGame_HD_AVC.mp4
Size 118458379
Platform Windows
OS Version 5.01
Video output DirectDraw 1280x1024 32bits Lookup
Video zoom 1920x1080 -> 1280x720

TCPMP Version 0.71f Benchmark Results + CoreAVC 0.71g

Average Speed 158.62%
Video Frames 3673
Amount of Data 115601 KB

Bench. Time 1:36.578
Bench. Frame Rate 38.03
Bench. Data Rate 9.8 Mbit/s

Original Time 2:33.194
Original Frame Rate 23.98
Original Data Rate 6.1 Mbit/s

URL \mplayer2\TheGreatestGame_HD_AVC.mp4
Size 118458379
Platform Windows
OS Version 5.01
Video output DirectDraw 1280x1024 32bits Lookup
Video zoom 1920x1080 -> 1280x720

TCPMP Version 0.71d Benchmark Results + CoreAVC 0.71e

Average Speed 196.80%
Video Frames 3673
Amount of Data 115601 KB

Bench. Time 1:17.843
Bench. Frame Rate 47.18
Bench. Data Rate 12.1 Mbit/s

Original Time 2:33.194
Original Frame Rate 23.98
Original Data Rate 6.1 Mbit/s

URL \mplayer2\TheGreatestGame_HD_AVC.mp4
Size 118458379
Platform Windows
OS Version 5.01
Video output Null Video 1920x1080 0bits Lookup
Video zoom 1920x1080 -> 0x0

TCPMP Version 0.71f Benchmark Results + CoreAVC 0.71g

Average Speed 232.28%
Video Frames 3673
Amount of Data 115601 KB

Bench. Time 1:05.952
Bench. Frame Rate 55.69
Bench. Data Rate 14.3 Mbit/s

Original Time 2:33.194
Original Frame Rate 23.98
Original Data Rate 6.1 Mbit/s

URL \mplayer2\TheGreatestGame_HD_AVC.mp4
Size 118458379
Platform Windows
OS Version 5.01
Video output Null Video 1920x1080 0bits Lookup
Video zoom 1920x1080 -> 0x0

fantastic results keep it up picard :) or should i say Make it so ;)

the decoding problems also are gone
i see now that you downscale the picture to the nearest display res hehe so for me thats 720p :)

PicardGK

28th December 2005, 13:58

Thanks.

i see now that you downscale the picture to the nearest display res hehe so for me thats 720p :)
In the case of DirectDraw overlay mode the full sized YUV color space frames are transfered to the video card memory and it's downscaled by video card during displaying.

But the null vs directdraw speed difference is larger as mine. Do you have a 8x AGP video card? Or maybe it does vsync even in benchmark mode. Do you seen tearing during benchmark? (I do with my Radeon 9600)

Sorry for being offtopic. For further CoreAVC discussion: http://forum.doom9.org/showthread.php?t=104277

ps: ops. I forget to update the version number in the player to 0.71g

Sharktooth

28th December 2005, 15:45

im truely impressed!
i can play most of my HD stuff with less than 50% of CPU occupation (the remaining stuff has been ecoded with custom matrices...).

Sharktooth

28th December 2005, 15:54

back on topic (more or less)... that could be interesting: http://www.cooltechzone.com/index.php?option=content&task=view&id=2113
so, FVD is a third contender and IIRC it doesnt support mpeg codecs.

digigami

28th December 2005, 17:49

Test in progress for my mega-test codec but only for HD ...

Sounds great. I picked up a copy of the Ice Age (#1) DVD to take a look at their quant matrices. I will post them here in a bit, maybe you can re-do your test with their qmatrices.

Also, could you post your AVISynth script used to make your winning MPEG-2? I think there are a lot of lurkers who might want to duplicate your results.

Gen

digigami

28th December 2005, 21:02

i also redid Ben Waggoners 8 Mbps The Island Trailer with the Ateme Encoder and X264 alot more complex then his quicktime encode, unfortunately i don't have so much space available for upload i see that i can host it maybe on Rapidshare :)

I'm reworking Wagonner's 'HDTV MPEG-2 file. It's a 2.21:1 aspect ratio, and he used letterboxing. I switched to 1280x720 and am using aspect ratio signalling, and so far I am about 1/2 the size, average bitrate of 9.8 mbits/s.

If you want a place to host samples, I will create an account on our server.

digigami

28th December 2005, 21:15

back on topic (more or less)... that could be interesting: http://www.cooltechzone.com/index.php?option=content&task=view&id=2113
so, FVD is a third contender and IIRC it doesnt support mpeg codecs.

Yes, the chinese have been loathe to follow along. My guess is that MS gave them the codec royalty free. When MS is building market share they will do almost anything. I understand licensing/patents were prime motivators behind the development of SVCD vs DVD.

Sagittaire

28th December 2005, 21:54

Sounds great. I picked up a copy of the Ice Age (#1) DVD to take a look at their quant matrices. I will post them here in a bit, maybe you can re-do your test with their qmatrices.

Also, could you post your AVISynth script used to make your winning MPEG-2? I think there are a lot of lurkers who might want to duplicate your results.

Gen

My matrix (Sagittaire Matrix) seem to be the best for metric at this time. It's a HVS like matrix (HVS-good structure with lower coef for high frequencies intra/inter)

Avs script is very simple script without advanced tweak filter:
source=DirectShowSource("D:\Mes dossiers\High-Definition\ice_age_2-tlrD_h1080p.mov", fps=23.976)
source=Crop(source,12,4,-12,-4)
source=LanczosResize(source,1280,720)
source=AssumeFPS(source,25)
Return(source)

But before all it's a Libavcodec performance and not Matrix or Script performance.

First result: for same quality (metric) with HPIV at 1250 Kbps with the best AVC encoder I must use 2600 Kbps with Libavcodec MPEG2 encoder

digigami

28th December 2005, 21:55

no-one will encode professionally HD stuff with your or any other mpeg-2 at 3.3mbps or even at 8mbps as well...
however h.264 can be LOSSLESS too (that's different from transparent).
a professional h.264 encode with all the quality needed for distribution will be something around 3000kbps but you will not notice ANY kind of artifacts even with a frame by frame comparison unlike some actual DVDs.

Ok, I'm referring to this:
http://www.ti.com/asia/docs/india/tiidevconf2004/analog_symp/munish.pdf

They are quoting H.264 FRExt "Stunning Quality" 720p at 6.6 mbits/s. Does this sound right to everyone?

Also, having familiarized myself with the coding techniques in H.264, I think it is interesting that the post 03 extensions were primarily aimed at making H.264 more MPEG-2 like.

Regarding the film grain modelling, while an interesting addition, I think it will be tricky to implement automatically because if you look at the way modern cinematographers are using film stock, they are using it more as a way to tell a story, which means mixing different stock, with different grain characteristics intercut into the feature. Which means, if you want film grain modeling to work first you have to : 1) automatically detect the size of the grains and/or film stock in a scene. 2) remove the original grains and finally code the correct model. Does anyone know how Ateme handles this? You can bet this won't be in any single-pass hardware encoders anytime soon. Those will be the ones feeding your DVB in europe.

Kostarum Rex Persia

28th December 2005, 23:42

digigami

29th December 2005, 00:06

digigami, I don't know what are you trying to prove us? Even baby know that H.264 codec is much better then MPEG2 on same bitrates.

I begining to suspect that your idea is to block further improving H.264 in quality? It seems that Sony pays you to tell that MPEG2 is better or same than any H.264 codec.

What are you talking about? Read the thread please. I don't know about anyone else, but right now I am working on learning what the correct bitrate is for H.264 720p 'stunning quality'. I am using Nero Digital Pro, which I just bought, and x264.

Sharktooth

29th December 2005, 05:06

Ok, I'm referring to this:
http://www.ti.com/asia/docs/india/tiidevconf2004/analog_symp/munish.pdf

They are quoting H.264 FRExt "Stunning Quality" 720p at 6.6 mbits/s. Does this sound right to everyone?

Also, having familiarized myself with the coding techniques in H.264, I think it is interesting that the post 03 extensions were primarily aimed at making H.264 more MPEG-2 like.

Regarding the film grain modelling, while an interesting addition, I think it will be tricky to implement automatically because if you look at the way modern cinematographers are using film stock, they are using it more as a way to tell a story, which means mixing different stock, with different grain characteristics intercut into the feature. Which means, if you want film grain modeling to work first you have to : 1) automatically detect the size of the grains and/or film stock in a scene. 2) remove the original grains and finally code the correct model. Does anyone know how Ateme handles this? You can bet this won't be in any single-pass hardware encoders anytime soon. Those will be the ones feeding your DVB in europe.
IMHO 6.6mbps is a bit excessive, expecially if you look at how poorly some DVDs are encoded.
3mbps seems fair to me for almost 90% of movies around. Sure there's always the ultra-grainy movie that requires more bits...
For what concerns grain modelling, the ateme encoder already has that, but the decoder must support it (ateme decoder does) or no grain will be added while decoding.

digigami

29th December 2005, 05:22

But before all it's a Libavcodec performance and not Matrix or Script performance.
First result: for same quality (metric) with HPIV at 1250 Kbps with the best AVC encoder I must use 2600 Kbps with Libavcodec MPEG2 encoder

Ok, can you go with Sharktooth's recommendation for 3 mbits/s AVC and discover the bitrate for Libavcodec?

Also what commandline switches do you pass to Libavcodec MPEG-2 encoder?

Gen

CruNcher

29th December 2005, 08:11

munishs presentation is ok but their marketing is bad http://www.fastvdo.com/H.264.html their software decoder is way of to be one of the bests :P and also one thing is outdated in that presentation, also even if the Blu-Ray founders where happy with the results after the 8x8 transform was introduced to FreXt they still decided as we all know for Mpeg-2 as their first choice :P, but if you read something like this it's clear why this is the case ;)

Thank you for the chance to see our material after compression with the AVC
encoders. The butterfly images are a good way to compare compression
algorithms. We recognize that the material we provided was some of the most
challenging that can be presented to a compression scheme, but it does
provide indication of the performance of the compression.

You presented two image segments for our review. You stated that both were
encoded at roughly 15Mb/s bit rate.

We saw the two segments and we do not know the details of the encoding
process, that being said the resultant image from current
Main Profile AVC encoder was not acceptable for quality due to the 'dancing'
of the noise floor. The modification to the AVC algorithm that you showed
did solve this problem and was a better resultant image.

We don't know if a different application of the current AVC toolbox could
have been applied that resulted in the same image quality.

We are still very concerned with the terms and conditions for licensing of
the AVC algorithm, especially for streaming applications. This would need to
be fully addressed prior to our support of the approach.“

--
Jerry Pierce
Senior Vice President
Technology
Universal Pictures
818-777-6300 cell 818-749-7776
jerry@jerrypierce.org

'm reworking Wagonner's 'HDTV MPEG-2 file. It's a 2.21:1 aspect ratio, and he used letterboxing. I switched to 1280x720 and am using aspect ratio signalling, and so far I am about 1/2 the size, average bitrate of 9.8 mbits/s.

If you want a place to host samples, I will create an account on our server.

Thx for that offer but seems you missed that http://forum.doom9.org/showthread.php?p=756989#post756989 ;)

digigami

29th December 2005, 09:46

CruNcher

29th December 2005, 10:48

digigami

29th December 2005, 18:13

@digigami
look @ that thread their you find Hardware solutions
http://forum.doom9.org/showthread.php?t=95939
The Tandberg EN5990 Encoder is the most used here in Europe by Playout Stations for HD AVC
http://www.tandbergtv.com/productview.asp?n=68

Only supports up to 4.0 - I think this is the 2003 spec, right?

What hardware are they using downstream at the customers home?

Sagittaire

29th December 2005, 21:42

H264 MP@L4.0 or HP@L4.0

Level are simply RC specifications (vbv) and maximun vector range.

Level 4.0:
Max Frame Size (sample): ~ 2 Mbs
Max sample: ~ 60 Mbps
Max Vector Range: +/- 512
Resolution: ~1280*720*60 or ~1920*1088*30 max

For H264 it's a very high max quality ... certainely for MPEG2 too.

CruNcher

29th December 2005, 22:25

What hardware are they using downstream at the customers home?

Hardware like this and O wonder it's HP@L4.0 ;) but it's not widely available to customers yet just 1000 units or so used in Stores and by Special selected Persons.
http://www.pacemicro.com/Corporate/products/prodinfo.asp?PID=DS810

Im more for this solutions anyway we in europe also useing the new DVB-S2 standard so we have to update this too to be able to watch Free and Paytv in HD http://www.knc1.com/d/produkte/digital_dvb_s2_plus.htm

We certainly have one major economical important factor here and that is the Soccer World Championship in 2006, it's really pushing HD on all levels here.

russo121

2nd January 2006, 23:21

I can't see any movie in x264 mp4 form? I get an error in vlc media player version 0.8.4a - "mp4 error: MP4 plugin discarded (no moov box)" ?
I tried also nero showtime, quicktime player, wmplayer, tcpmp. I don't know how to see.. can somebody help?

Edit:
Here sample H264 High Profil

King Kong (http://multimediacom.free.fr/Download/x264HP_KingKong_720p_1500.mp4)
1280*720*25
1500 Kbps

Harry potter IV (http://multimediacom.free.fr/Download/x264HP_HPIV_720p_1250.mp4)
1280*720*25
1250 Kbps

Ice Age 2 (http://multimediacom.free.fr/Download/x264HP_IceAge_720p_1250.mp4)
1280*720*25
1250 Kbps

dbzgundam

4th January 2006, 15:31

My matrix (Sagittaire Matrix) seem to be the best for metric at this time. It's a HVS like matrix (HVS-good structure with lower coef for high frequencies intra/inter)

Avs script is very simple script without advanced tweak filter:
source=DirectShowSource("D:\Mes dossiers\High-Definition\ice_age_2-tlrD_h1080p.mov", fps=23.976)
source=Crop(source,12,4,-12,-4)
source=LanczosResize(source,1280,720)
source=AssumeFPS(source,25)
Return(source)

But before all it's a Libavcodec performance and not Matrix or Script performance.

First result: for same quality (metric) with HPIV at 1250 Kbps with the best AVC encoder I must use 2600 Kbps with Libavcodec MPEG2 encoder

I don't mean to go off topic, but what exactly is used to output with libav? Vdub can't do it through FFDShow for obvious reasons (This is MPEG2 afterall), so unless there's a CLI way, I don't know how. >>

On topic though, Sagittaire's MPEG2 clip certainly is amazing, for the bitrate of course... All these claims I hear about AVC being good with moderate amounts of grain at 3000kbps still seem a bit low to me... Though I am using DVD sources, not uncompressed or HD sources, 3Mbits seems to still smear at least a little bit with current implementations of x264.

MPEG2 does even worse, for obvious reasons! :D I have also tried (somewhat) to duplicate Sagittaire's results... I assume digigami's screenshot was showing Sagittaire's Matrix, so I used that in HC for encoding... Results were kind of crappy, even with the max bitrate allowed being at least 8Mbps (this time, the source was filtered anime though).

The point of all this being of course, MPEG2 is never gonna come close to getting AVC's results... Even that of well-developed ASP codecs as well (I've done HD XviD tests and they encode much better than MPEG2 equivalents). Still though... I'd like to apply Sagittaire's libavcodec settings to an SD res video... I imagine it would lead to very small (for MPEG2) DVDs. ;)