Nero/Ateme HP AVC Encoder Beta : Call For Testers

[CyberGuy]

Originally posted by acidsex
I dont have time to fire up a campaign. I am too busy teaching public school during the day and college courses at night. Once school lets out for the summer, I will have nothing to do for a month or two.

Ok, Ill make a pitch. If anyone would like to sponsor a high school teacher that wants to use H.264 HP/lossless for archival and delviery of the school news show, please sposnor me.
Well, I'm the Technology Coordinator for a school district and am interested for the same reason, but I don't think using a beta product for this is a good idea; you should probably wait until the final version comes out. With that said, I have 700+ computers available for the next couple of months that I could use to do multiple encodings with different settings at the same time, therefore finding possible bugs quicker. I've already got this setup using the JM Reference encoder so I only need to plug in the CLI Ateme encoder in place of the CLI JM encoder and I'm off testing. Would anybody like to nominate me?

[Sirber]

@acidsex

the codec is for testing only, not archiving or other things.

[acidsex]

The reason I would like to test this now is because our district HAS to make a decision in the next month on what codec we will officially adopt. Out of four people, one wants WM9, one wants QT H.264, and one wants Nero H.264 and that leaves me as the deciding vote. Thus far, Nero is the only one that doesnt currently offer a lossless mode and lossless is the most important factor for us.

Any beta testing would not be used in a production enviornment but simply as a test. But with the deadline we have in order to make the decision for budget reasons, I am not afforded the opportunity to wait a few months for the release to begin to make a decision.

With the way the current system is set up for selection, it really is nothing more than a popularity contest as to who gets selected regardless of who was in the last round of testing.

[bobololo]

Originally posted by acidsex
The reason I would like to test this now is because our district HAS to make a decision in the next month on what codec we will officially adopt. Out of four people, one wants WM9, one wants QT H.264, and one wants Nero H.264 and that leaves me as the deciding vote. Thus far, Nero is the only one that doesnt currently offer a lossless mode and lossless is the most important factor for us.

WM9 and QT H.264 offer lossless mode ? I didn't know that !? Ah maybe you're considering uncompressed RGB as lossless ?

Any beta testing would not be used in a production enviornment but simply as a test. But with the deadline we have in order to make the decision for budget reasons, I am not afforded the opportunity to wait a few months for the release to begin to make a decision.

I've nothing against you being a tester, but I really doubt about your argument justifying your need to evaluate the beta for taking your decision. Considering the codecs you have to choose, I don't see what the new beta would bring more to you. You should already be able to make your choice. It won't be very relevant from a budget point of view (we don't beta test the codec price ;)) and for the technical part, since there is no lossless mode for wm9 and qt7, there is no remaining choice ;).

ps: completely off topic, do your students know you're wandering around video forum using "acidsex" as nickname ? :D

[acidsex]

First, Acidsex is an old DJ name I used to spin an just never got rid of it.


QT does have a lossless mode as does WM9 (as close to lossless as it can get with Quality VBR.)Both file sizes are usually HUGE but still remain as close to lossless as possible. This is the major part of the beta that concerns me.

[IgorC]

Untill I know WM9 have a losseless audiocodecs but not video. Even if QT7 and WMV9 loseless exist it changes nothing. Nero AVC loseless will provide always superior quality (smaller size for the same quality)

[acidsex]

Originally posted by IgorC
Untill I know WM9 have a losseless audiocodecs but not video. Even if QT7 and WMV9 loseless exist it changes nothing. AVC loseless will provide always superior quality (smaller size for the same quality)

I dont disagree with your statements at all. However, I have the wonderful task of battling marketing hype that my bosses have got from M$ and Apple. But I have to have something to back up my decision other than my experiences with the main profile encodes I have done to validate my recommendation to my boss.

[jfehr]

'close to lossless as possible' isn't lossless. Almost any codec can be 'close to lossless as possible' if you throw enough bits at it. (Assuming there's room in the codec for that many bits, that is.)

lossless == exactly the same when decoded as the original source was. (not almost exactly the same.)


Originally posted by acidsex
First, Acidsex is an old DJ name I used to spin an just never got rid of it.


QT does have a lossless mode as does WM9 (as close to lossless as it can get with Quality VBR.)Both file sizes are usually HUGE but still remain as close to lossless as possible. This is the major part of the beta that concerns me.

[acidsex]

Well what I am saying is that those are the closest lossless mode that those codecs offers.

[CruNcher]

@acidsex
then guide your boss to this forum and let him see what the "real" advanced users say about this ;) i think every boss is smart enough not to take everything as the truth what marketing of a company (not important how big it is) is saying without actual proof and here you get that proof from differnt people from around the world :)

[acidsex]

Well I got one vote and much thanks to bonzi. Hopefully one other person will see me as trustworthy and then I can quit my whining :) Again, thanks bonzi.

[jfehr]

Anyone want to vote for me? I haven't had much experience on this end of the beta testing, but I've got plenty on the other side, so I know what information is useful to the developer when reporting bugs. (I used to port games for a living, if anyone wanted to know.)

As for testing, I have a large DVD collection to test with as well as Tivo content, and should be getting HD content soon as well.

[acidsex]

Again, my pleas and request for one more vote so I can join this beta test. The way it is looking after this weekend, I will fall short by one vote for this round. Surely someone here is like Bonzi and has a level of trust for me.

[LigH]

There are still a few days for bobo to update his list - some people still have some votes to spend.

[bobololo]

There are still a few days for bobo to update his list - some people still have some votes to spend.

It remains a few days to go before the official closing of testers recruitment :)

Btw after some internal discussions, we have conclude that ateme staff will have the possibility to submit some "golden votes" to allow volunteers they estimate valuable and trustworthy the chance to join the test. So even if you don't have any votes for now, you may be lucky enough to be directly promoted ;)

[acidsex]

Hopefully they will take into consideration thos of us who were on the first beta that didnt leak it and spare us with a vote to get us over the hump. :)

[Sergei_Esenin]

It remains a few days to go before the official closing of testers recruitment :)

Only a few days to go, and 0 votes for old Sergei. :scared: Ah well, not surprised since I rarely post here, and then mostly about analogue capture (laserdiscs, Betamax, VHS--fun films not available on DVD!). But, if anyone knows my daily posts to AVS Forum (http://www.avsforum.com/avs-vb/member.php?userid=7425165), or can tell I'm a trustworthy fellow from my 1,200+ posting history there, feel free to slip me a vote. :sly: I have much unique material to test with, from analogue caps to ATSC HDTV, and use a 100" Infocus X1 projector for my display so I can very easily detect flaws many people with smaller displays would miss.

Well, that's my last pitch. [fingers crossed] :D

[JasonFly]

OhOhOh!!! Thank you very much LigH . I 've just saw that you gave me a vote in the "golden list". I'm going to prepare some samples for this beta test.

Bye :thanks:

[acidsex]

OhOhOh!!! Thank you very much LigH . I 've just saw that you gave me a vote in the "golden list". I'm going to prepare some samples for this beta test.

Bye :thanks:

Ditto. Ligh came through for me as well. Kudos :)

[DeeGee]

I guess I should ask for someone to vote for me to have any hope of getting into the beta... :p
So if you happen to be one of the trusted people and think I'm worthy, could you please vote for me. :thanks:

[acidsex]

I am surprised ChilledoutUK doesnt have two votes yet. I always thought Chilled contributed greatly to threads but then again, I dont have a vote.

[SeeMoreDigital]

I am surprised ChilledoutUK doesnt have two votes yet. I always thought Chilled contributed greatly to threads but then again, I dont have a vote.Agreed... I must admit, if I was elegible to vote, I would have passed one Chille's way ;)


Cheers

[DeathTheSheep]

Perhaps if enough non-trusted people vote, a vote should be allowed him. I mean, if a person gets 3 or more votes from other candidates who were voted for, shouldn't he be allowed? Just some food for thought.

[acidsex]

I havent seen anyone with three votes yet but if so that would definitely make sense. I suspect that chilled will end up with one of those developer votes. He really has been invaluable over various subjects here.

[LigH]

# testers who did not yet vote:

- andrey
- bill_baroud
- bulletproof
- latexxx
- mr_schizo
- plonk420
- spyder
- thegeby
So - no reason for panic. ;)

[jfehr]

Doesn't look like I'll get a vote. (Didn't really expect to, as I'm pretty new here.) Hopefully they're still considering allowing testers in that are willing to fax them some sort of picture ID.

[Andrey]

>># testers who did not yet vote:
>>- andrey
>>...
Crunch. Sorry.
And something f%$^ed up my internet connection - it is dropped every 5 mins. :devil:

P.S. Prepared StarWars 4 form laserdisk (copy form ld to DVD) for test. Very noisy... :)

[acidsex]

woohoo, chilled got his second vote. good to see that it worked out thus far.

[LigH]

I am rendering a video using Terragen - even the unregistered copy is already capable of up to 1280x960, this will be enough for HDTV-720p (1280x720).

I just wonder if it is easily possible to make this terrain, world, and script available so that other people may reliably reproduce the result and use it for own tests.

(Water ripples, the sun, and clouds are animated.)

[Prettz]

bobololo: I can't tell if your first post mentions this or not, but does this HP AVC encoder now support custom quantization matrices? I always felt that this was the most important thing missing from the MP AVC encoder.

[LiFe]

Good to see I'm as popular as ever : )

I'm hating this new bloated GUI, which has provided few advances considering the real estate stolen.

[Sirber]

Good to see I'm as popular as ever : )

I'm hating this new bloated GUI, which has provided few advances considering the real estate stolen.
Talking about recode?

[Manao]

Prettz : yes, the HP encoder will support custom matrices ( they are walled 'custom scale matrices' in the original post of bobololo ). The MP encoder wasn't supporting them simply because custom amtrices are part of the high profile. However, do not think so quickly that custom matrices are the holy grail of h264. 8x8 dct transform is imho far more interesting.

[LiFe]

Talking about recode?
No, bit OT, talking about doom9.

[Sirber]

No, bit OT, talking about doom9.
MeGUI? :confused:

[LigH]

No, it sounds like LiFe critizises the new look of the doom9 board. But this does absolutely not belong to the topic of this thread! :rolleyes:

[DeathTheSheep]

[O/OT, not real sure anymore]
Woah, I'm likin' it. Can't see what you got against it. I especially like how the messages are emailed to you so that you can actually see the new post in your inbox. Pretty cool. And the quick reply feature... very nice indeed. The whole thing is highly upgraded and is much faster on my dial-up machine.

[LigH]

Dear bobo,

in your "schedule" you wrote that the "recruitment" will be closed today. It seems, though, that not all votes have arrived in time. So I hope that you may consider some kind of "wildcards" for testers (e.g. which did not yet receive the required count of votes, but seem to be worth a try - concerning their overall board activity from your point of view, or several "nice words" from other volunteers). Of course, this will depend on the amount of beta testers you would call "optimal".

[FXsan]

So it will do monochrome, now, thus 4:0:0 ?
But what about 4:2:2, which is also defined in High profile.
This is definitively high quality (yet reserved for slightly higher bitrates, as for lower bitrates, it is better to first decimate chroma).
Is it hidden feature/parameter somewhere, or planned for next major release (you speak of a 6 months rythm?...)

Currently I'm still with Dv videos as source (4:2:0), but I was thinking of a 3CCD camera with digital recording, and there 4:2:2 does matter for final encoding. The new HD corder (JVC and Sony) seems pretty nice ! But I'm getting off topic, just for context :)

BTW a major feature could be with MPEG source file an option to keep the source GOP. It may not be optimum if the target bitrate needs a long GOP, but if it is to stay with a GOP aroung 15, it may be worth it. This way, we may have less artefacts due to movement estimation of the initial encoding (re-encoding transforming I frames to I frames, and no B or P frame to I frame) ?

[Prettz]

Prettz : yes, the HP encoder will support custom matrices ( they are walled 'custom scale matrices' in the original post of bobololo ). The MP encoder wasn't supporting them simply because custom amtrices are part of the high profile. However, do not think so quickly that custom matrices are the holy grail of h264. 8x8 dct transform is imho far more interesting.
I did some encode tests with the MP AVC encoder and, compared to an Xvid encode that used the hvs-better matrix, the AVC encode had way too much ringing all over. If I could have used hvs-better, most of the ringing in the easy to compress sections of the movie would no doubt have disappeared.

I would say that forcing the MPEG matrix set an upper limit on quality for animation, no matter what other advanced tools the codec had. So this is great news that the HP profile supports custom matrices.

[Kostarum Rex Persia]

Nero Recode H.264 encoder is on my machine(Athlon XP 3200+,1 GB RAM DDR400 CL 2.5) is too slow,about 5.5-7 fps.

When will Ateme improve speed of H.264 encoding?

[berrinam]

:search: This has been answered earlier in this thread. Ateme have improved speed on the new version of their codec for the same features, but the extra features make it slower overall.

The current version can also be sped up by shortening the feature list.

[baer999]

Well better speed at the same settings... thats great ! How long does the beta test approximately go ? The difference to x264 will be much bigger than it is now, or isn't it ?

[ChronoCross]

Nero Recode H.264 encoder is on my machine(Athlon XP 3200+,1 GB RAM DDR400 CL 2.5) is too slow,about 5.5-7 fps.

When will Ateme improve speed of H.264 encoding?

your pc seems wierd. I can run it at 4.5fps on my p3 1.13 ghz machine with 256MB ram.......

[Sharktooth]

Nero Recode H.264 encoder is on my machine(Athlon XP 3200+,1 GB RAM DDR400 CL 2.5) is too slow,about 5.5-7 fps.

When will Ateme improve speed of H.264 encoding?
Please.... please... :readrule: :search: and stay in topic.
Recode, given its quality, is already the fastest encoder available.
If you have speed problems tune down the codec settings.

[peteag]

when will this version be avaible in recode?

[Sharktooth]

... after the beta test ...

[bobololo]

Dear bobo,
in your "schedule" you wrote that the "recruitment" will be closed today. It seems, though, that not all votes have arrived in time. So I hope that you may consider some kind of "wildcards" for testers (e.g. which did not yet receive the required count of votes, but seem to be worth a try - concerning their overall board activity from your point of view, or several "nice words" from other volunteers). Of course, this will depend on the amount of beta testers you would call "optimal".

While we'll start sending the terms on monday, I think we weill still allow late votes.

[bobololo]

So it will do monochrome, now, thus 4:0:0 ?
But what about 4:2:2, which is also defined in High profile. This is definitively high quality (yet reserved for slightly higher bitrates, as for lower bitrates, it is better to first decimate chroma).
Is it hidden feature/parameter somewhere, or planned for next major release (you speak of a 6 months rythm?...)

4:2:2 color space is supported in High 4:2:2 profile only so we won't support it for this beta. 4:2:2 is mainly used in very specific & professional use we don't target for now but that could change in the future.


Currently I'm still with Dv videos as source (4:2:0), but I was thinking of a 3CCD camera with digital recording, and there 4:2:2 does matter for final encoding. The new HD corder (JVC and Sony) seems pretty nice ! But I'm getting off topic, just for context :)

BTW a major feature could be with MPEG source file an option to keep the source GOP. It may not be optimum if the target bitrate needs a long GOP, but if it is to stay with a GOP aroung 15, it may be worth it. This way, we may have less artefacts due to movement estimation of the initial encoding (re-encoding transforming I frames to I frames, and no B or P frame to I frame) ?

The use of short GOP doesn't affect the quality the way your describe it. In the opposite it tends to reduce the whole quality of the clip since it involves the coding of I frames (which are also bigger) where they aren't useful wasting precious bitrate for nothing. Short GOPs are mainly used in streaming application (DVB, IPTV, etc) in order to limit channel switch delay.

[bobololo]

I did some encode tests with the MP AVC encoder and, compared to an Xvid encode that used the hvs-better matrix, the AVC encode had way too much ringing all over. If I could have used hvs-better, most of the ringing in the easy to compress sections of the movie would no doubt have disappeared.

I would say that forcing the MPEG matrix set an upper limit on quality for animation, no matter what other advanced tools the codec had. So this is great news that the HP profile supports custom matrices.

Your remark is quite interesting since it's a rather well adopted recognition that AVC considerably reduce ringing artefact over mpeg-4 asp. Could you show some samples where this case occurs (provided your settings) ?