CoreCodec/H.264 Codec "CoreAVC"

[BetaBoy]

We have talked about a "6-pack Bundle"...

1) CoreAAC v2.0 (w/DSFilter) will be out around the time TCPMP 1.0 is released based on our internal devel'd HE-AAC decoder for TCPMP.
2) CoreASP 1.0 (same as above)
3) CoreMP3 1.0 (same as above)
4) CoreAVC 1.0 (soon)
5) CoreFLAC
6) CoreVORBIS

But I despise the term 'Codec Pack' as most are 'codec hacks'. With layer after layer of codecs and filters that do not act in a cohesive manner.

/me hugs Haali's splitter

[Caroliano]

Sorry to ask, but what would be the advantage of CoreMP3 over the native mp3 decoder of windows?

[CobraX]

Howly cow :eek:

Let me straighten out things. I recently downloaded some HD trailers from the apple site. I didn't have any AVC decoders installed back then (cause I'm using the koepi XviD exe to decode my movies, so no ffdshow for me). On my rig (Athlon64 3000+@3200+, 1 GB ram) most of the movies didn't play fluently at all. They played fluently during 50% of the clip. It got me down a bit cause I couldn't play 720p trailers on my (pretty) fast rig. So I tried out CoreAVC 0.0.0.4. What the hell happened??? My trailers are now playing fluently during 90% of the clip. Simply amazing! Now I got down again when I read this thread and heard that CoreAVC is not going to be free :(. Can't you guys make it free for personal use (just like xvid)? Pleeeeease you would make my day (and even year).

[BetaBoy]

Sorry to ask, but what would be the advantage of CoreMP3 over the native mp3 decoder of windows?
- Speed savings
- Quality sound
When its ready (the DS filter that is.. the decoder has been done for months now) i'll submit it to HA for some testing.

[lexor]

- Quality sound
I wouldn't claim that, especially at HA, if I were you. All decoders must produce identical output (it's a finite state machine after all), otherwise there is a bug that needs to be fixed. If there was a bug that affected default decoder, it would have already been found by now, especially by the guys at AH, I mean they found a bit missallocation in Nero AAC encoder by ear, even though betrate analysis revealed no faul play. So you can claim speed, but no way you can claim quality improvement.

[Sharktooth]

quality improvment can be there thru a better interpolation and post processing.
it has been already proven there are better mp3 decoders than the reference decoder and the fraunhofer one included in windows.

[lexor]

quality improvment can be there thru a better interpolation and post processing.
it has been already proven there are better mp3 decoders than the reference decoder and the fraunhofer one included in windows.
What now? Where is this "proof" you speak of? All I see is claims being made by the same people that claim snake oil improves sound (these guys love MAD mp3 decoder), never seen a proof. Plus post processing is not decoding, and what interpolation? you mean dithering from 44.1 to 48 Khz (or from 16 to 24bits) or something like that? that's not decoding either.

[dimzon]

What now? Where is this "proof" you speak of? All I see is claims being made by the same people that claim snake oil improves sound (these guys love MAD mp3 decoder), never seen a proof. Plus post processing is not decoding, and what interpolation? you mean dithering from 44.1 to 48 Khz (or from 16 to 24bits) or something like that? that's not decoding either.
http://www.hydrogenaudio.org/forums/index.php?showtopic=17728

[Caroliano]

The diferences are so subtle that even Guruboolez had an hard time to find them. And audio decoding speed is not so important if we compare with video decoding speed... but anyway, good work.

[Sharktooth]

What now? Where is this "proof" you speak of? All I see is claims being made by the same people that claim snake oil improves sound (these guys love MAD mp3 decoder), never seen a proof. Plus post processing is not decoding, and what interpolation? you mean dithering from 44.1 to 48 Khz (or from 16 to 24bits) or something like that? that's not decoding either.
dimon provided the link. post processing (dct noise reduction, frequencies restoration, etc) is part of the decoding as well as interpolation between samples.
maybe you should get a clue of what you're talking about before posting.

[BetaBoy]

lexor... chill... all I said was quality sound. I did not say "IT IS THE BEST", "IT'S BETTER THEN...", etc. I'd never do that in a non-competitive manner.

[dimzon]

2 lexor && Sharktooth
The diferences are so subtle that even Guruboolez had an hard time to find them. And audio decoding speed is not so important if we compare with video decoding speed... but anyway, good work.

[lexor]

dimon provided the link. post processing (dct noise reduction, frequencies restoration, etc) is part of the decoding as well as interpolation between samples.
maybe you should get a clue of what you're talking about before posting.
ok... so you threw terms into your first post without explaining and I misinterpreted, amazing (I can see how that warrants an insult of my intelligence). But you go ahead, post this quality claim at HA, they gonna looooove you long time there :) don't let little old me stop you.

@betaboy, yeah that is a good point, just your original post seemed like you were gonna claim as advantages speed and quality (so in effect you did claim "better than..."), problem is that if you have a decoder you gotta have quality or you don't have a proper decoder. It's just different when you said speed, you guys got kick ass product in that respect, putting quality into same list would suggest equal level of superiority, which I don't believe can be produced (at all, not just by you).

My point wasn't that you can't make a top notch decoder in terms of quality, I merely doubt that you can achieve superiority of the same level as you do with speed, I seriously doubt anyone can compete there with you.

@dimzon sorry you posted while I was typing, since there is no link, can you clarify, the quote says "difference", but did he found an improvement? or is it just difference, as in, different people like different sound?

[Sharktooth]

you asked for proofs... and the proofs are there... by the same HA guys you mention in your posts.
obviously you still dont have a clue of what are you talking about.
discussing with arrogant and ignorant ppl drives to nowhere, so this is my last post on the subject.

[dimzon]

@dimzon sorry you posted while I was typing, since there is no link, can you clarify, the quote says "difference", but did he found an improvement? or is it just difference, as in, different people like different sound?


as you can read from linked post:
guruboolezwrote:

It's not a MAD CHALLENGE. I repeat again that the listening conditions of the test are not reproductible on daily music listening. I've just put the decoder behind a magnifying glass, by using low volume samples.

CONCLUSIONS:

MAD isn't impressive. On four samples, result was worse than simplest decoders: increase in noise, without audible benefit on other points of the signal (aliasing, higher definition, etc...). But on some other samples, real improvements were noticed (less aliasing, less naturel and less synthetic sound)

LAME and Fraunhofer (WA5) are close, but not totally identical. Both are noisy, but less than MAD (good point). With some samples, sound was irregular, unatural... MAD partailly correct this feeling, and foobar2000 was perfect in comparison. Something like aliasing or moiré was introduced by the decoding stage. Slight difference between LAME and Fraunhofer (hard to ABX : up to 30...50 trials), in favour of Fraunhofer, less synthetic and maybe slightly more noisy

foobar2000 : each time winner. The less noisy of all challengers on all samples. The only additionnal noise audible was a very high-pitched band, very typical, and audible on extreme conditions (near silence sound, and with my amp pump to the maximum). All aliasing, irregularities... were gone: instruments are smooth, naturel. Nothing synthetic at all. With one sample (the #7), the sound was clearly better than the reference file (CD!). The lack of dithering on mastering seem to be prejudicial on very low volume part, and foobar2000 decoding correct the aliasing present on the CD. Simply amazing conclusion....

[lexor]

you asked for proofs... and the proofs are there... by the same HA guys you mention in your posts.
obviously you still dont have a clue of what are you talking about.
discussing with arrogant and ignorant ppl drives to nowhere, so this is my last post on the subject.
And still you show no proof (dimzon bolded important an sentance)... how amusing. I agree with you on your last point then.

@betaboy: sorry I derailed the discussion, I promise to focus on the subject matter of the thread from now on.

[Episodio1]

Sharktooth, I see lexor here not being ignorant: "Plus post processing is not decoding, and what interpolation? you mean dithering from 44.1 to 48 Khz (or from 16 to 24bits) or something like that? that's not decoding either." Besides he is explaining his position; not same as you. :(

Having read the link... Im with lex0r.

[Kurtnoise]

We have talked about a "6-pack Bundle"...

1) CoreAAC v2.0 (w/DSFilter) will be out around the time TCPMP 1.0 is released based on our internal devel'd HE-AAC decoder for TCPMP.
2) CoreASP 1.0 (same as above)
3) CoreMP3 1.0 (same as above)
4) CoreAVC 1.0 (soon)
5) CoreFLAC
6) CoreVORBIS
7) CoreWavPack
8) CoreTTA
......:p :D

[lexor]

btw betaboy, I've been meaning to ask about 1.0 features, last we spoke you were saying that there is going to be a player (somewhat like the ActiveX preview from CES) that can be integrated into anything and works with any format. I was wondering if 1.0 gonna have browser integration or will it still be purely stand alone. (I'm still grasping at the hope that I can one day be rid of QT)

[Sharktooth]

Sharktooth, I see lexor here not being ignorant: "Plus post processing is not decoding, and what interpolation? you mean dithering from 44.1 to 48 Khz (or from 16 to 24bits) or something like that? that's not decoding either." Besides he is explaining his position; not same as you. :(

Having read the link... Im with lex0r.
well, i only said there are techniques (postprocessing - and yes it's part of the decoding process) to improve the audio output of a decoder (as well as post processing on mpeg4 asp improves video).
is it so hard to understand?
the proof is there are better decoders than the fraunhofer reference decoder (read the dimzon post and follow the link on HA).
sure the difference is not that big but there's indeed a difference and there's still space for improvment.
hope it's all clear now.

[BetaBoy]

CoreWavPack needs work... CoreTTA.. Well... we stopped supporting Alexander's efforts after he used CoreCodec... but that is a whole other subject.

[BetaBoy]

So... back on track all... Since we are close to releasing CoreAVC. I want to discuss with the community our current CoreAVC versions... the only thing in question atm is the 'Extended' AVC profile and how best to fit this in.

[[ Standard Version ]]
- Baseline, Main, High profile support
- No, Interlaced
- No, SMP
- No, GPU

[[ Professional Version ]]
- Baseline, Main, High profile support
- Yes, Interlaced support
- Yes, SMP support (limit 2)
- Yes, GPU support

[[ Enterprise Version ]]
- Baseline, Main, Extended, High, High10, High 4:2:2, High 4:4:4 profile support
- Yes, Interlaced support
- Yes, SMP support (unlimited)
- Yes, GPU support

Thoughts... Suggestions.

[ChronoCross]

Add interlaced support to the standard version. SMP and GPU should be the upgrade to professional.

[SeeMoreDigital]

Add interlaced support to the standard version. SMP and GPU should be the upgrade to professional.Agreed :)

Do we really need three versions. Would two not do?


Cheers

[tomos]

personally consider myself a standard user, and the last test version i downloaded does all i need minus SMP (only cos i have an X2), although interlace support would be useful for supporting video content i havent encoded myself

GPU support i would think belongs in the professional one, along with other features :)

[[)370|\|470!2]

(postprocessing - and yes it's part of the decoding process) LOL?
to improve the audio output of a decoder Huh? In which manner exactly it improves audio quality?
Haven't you ever heard an expression that there's no way of getting a candy from a crap?
(as well aspost processing on mpeg4 asp improves video). Is this a joke?
is it so hard to understand? Well, yes. For me at least.
Though working as an asm coder for a several years, i simply don't get it,
how you could gain anything else from a plain stream
except for stuff that is initially there and already extracted/dequantized.
the proof is there are better decoders than the fraunhofer reference decoder
(read the dimzon post and follow the link on HA).
sure the difference is not that big but there's indeed a difference and there's still space for improvment.
hope it's all clear now.
Yea, right. But if you look bit more closely, you'll see, that all he talking about is dithering there.
Or is it a part of decoding process too? :D

[jjesusfreak01]

Add interlaced support to the standard version. SMP and GPU should be the upgrade to professional.
Yeah, this interlaced support is the only one of these features that standard users really need. Seriously, it would be a significant impediment if interlace support was missing.

[GhengisKhan]

Alright, maybe it's me, but these versions seem really targeted at x264 users... I use Nero Recode and I don't have any of those options. I currently use the Maximum Definition - AVC profile with Nero Recode. Could anyone tell me what version/features I would be using then... Betaboy, will this type of encoding even play on my Axim X50v under TCPMP?

I'm a little confused with this whole splitting up the features thing... maybe someone can help me out...

[ChronoCross]

Alright, maybe it's me, but these versions seem really targeted at x264 users... I use Nero Recode and I don't have any of those options. I currently use the Maximum Definition - AVC profile with Nero Recode. Could anyone tell me what version/features I would be using then... Betaboy, will this type of encoding even play on my Axim X50v under TCPMP?

I'm a little confused with this whole splitting up the features thing... maybe someone can help me out...

This is a decoder. not an encoder.

[Sirber]

What would be the prices for that?

[foxyshadis]

Noise shaping and dithering are psychoaudio enhancements, the same way that EQ and DSP are enhancements, would you argue there is no way you can imagine the sound being improved by them? The former are much more subtle, though. As a coder you should know that the quantized fft stores information at a higher resolution than the source, even if it may not be entirely meaningful information, and using that extra resolution properly can improve the output quality slightly.

It won't make a 96 kbps mp3 sound like a 160, but no one's arguing that; though Core may have some magic artifact reduction algorithms, who knows.

[Kostarum Rex Persia]

BetaBoy, hi, your idea about three versions looks fine, but you should add Interlace support in Standard version.

About the price, Standard version should be FREE, Pro and Enterprise versions should could cost about 25 $ for Pro, and 50-60 $ for Enterprise version.

[BetaBoy]

SMD... Interlaced in the pro version based on what is currently available in AVC content. This is the basis for the three CoreAVC versions and why I asked about 'Enhanced' profile comments, because now we are leaning for it to be a Enterprise feature but for some streamed/mobile content (down the road) it will be essential. We could move this to the Pro version in a 1.xx version if the feedback and needs warrant it.

ChronoCross... noted.... but as I stated to SMD.

jjesusfreak01... the issue is time and investment it has taken to properly devel interlaced... but why needed in a standard version? (Other then SAT content).

Sirber... tbd... as this is one of the reasons why i posted for all of your thoughts on what I had proposed.

[BetaBoy]

CoreWavPack needs work... CoreTTA.. Well... we stopped supporting Alexander's efforts after he used CoreCodec... but that is a whole other subject.
Toff informed me I was wrong on CoreWavPack... I thought it was based on a beta library (sorry David) so add it to the list.

[nm]

SMD... Interlaced in the pro version based on what is currently available in AVC content.
We already have DVB satellite channels broadcasting interlaced H.264 video here in Europe and for many people that is their first encounter with H.264. CoreAVC could have a large market here just because of this transition to HDTV that is just beginning (and because other decoders are painfully slow).

[Kostarum Rex Persia]

BetaBoy, you didn't answer on my question(comment).

[BetaBoy]

If you read the thread earlier I commented on why it cannot be free... at least initially. It is a goal of ours... but the CAP for AVC is 1 1/2 times that of ASP, so... I can say that our PRO version will be well under the $25 you stated (we are thinking a discounted $14.95) and our Enterprise version is tbd as it not be out till Q2 anyway.... but maybe sooner.. you never know ;-)

Also... we will have a vertical upgrade path for those who have purchased the CoreAVC decoder for when our encoder is released.

[leowai]

Also... we will have a vertical upgrade path for those who have purchased the CoreAVC decoder for when our encoder is released.CoreAVC Encoder? After the experience on the ultra fast decoder. I'm wondering whether it encodes AVC DVD resolution (720x480) or even HD resolution with AMD 3200++ in real time? If yes, it's really an evil like this :devil:.

[Sharktooth]

@betaboy: i just want one thing clarified: are the encoder and decoder features the same in what you listed above?
I mean, SMP could be a feature of the "standard" decoder but not of the "standard" encoder... and interlacing support too, etc...
or the decoder will be the same for the 3 packages?

[Sharktooth]

LOL?
Huh? In which manner exactly it improves audio quality?
Haven't you ever heard an expression that there's no way of getting a candy from a crap?
Ah yes... tell that to the movie studios... LOL...
Post processing is there to improve perceived quality removing compression artifacts or modifing the output to be "easier" to perceive.
Well, we can get rid of ffdshow PP or loop deblocking or divx/xvid decoder PP if they are useless... dont we?
. Is this a joke?
Well, yes. For me at least.
Though working as an asm coder for a several years, i simply don't get it,
how you could gain anything else from a plain stream
except for stuff that is initially there and already extracted/dequantized.

Yea, right. But if you look bit more closely, you'll see, that all he talking about is dithering there.
Or is it a part of decoding process too? :D
everything done by the decoder is part of the decoding process... or it would have been done in another stage...
however let's stop discussing about those things in this thread. just open another one if we want to continue coz we're getting OT.

[pest]

Sorry for posting offtopic, but this is making me a bit angry


Though working as an asm coder for a several years, i simply don't get it,
how you could gain anything else from a plain stream
except for stuff that is initially there and already extracted/dequantized.


A audio coder is a bit different.
There are numerous ways to improve rounding even on the decoding side.
Every stage depends on the accuracy of a prior stage and yes there are
very subtle differences. If you could hear them is another story.

[SeeMoreDigital]

SMD... Interlaced in the pro version based on what is currently available in AVC content. .Hmmm,

Currently I've only been able to get my hands on progressive AVC sources but given MPEG-4 AVC has just started being broadcast in Euroland, there must be a way of getting hold of some PAL 1080i (@ 50 fields per second) test samples.

I guess it's too early to ascertain, which European countries will be using what output resolution, which in-turn will determine whether it will be progressive or interlaced....

From what I understand, Sky in the UK will be trying-out, 1280x720 @ 50 frames per second and 1920x1080 @ 50 fields per second.


Cheers

[Sharktooth]

1920x1080i @25fps (50 fields per second) looks crappy...

[videomixer9]

hm, Pro7 Germany will air "Hero" this Sunday ... specs for the station are named as MPEG4 / AVC - 1080i/25 (25 Vollbilder/Sekunde). Previous Movies like Star Wars were only aired in 720 by them in MPEG2.

If anyone got a DVB-S2 receiver, it airs on Sunday 20:15 CET via ASTRA 1E, 19,2° Ost, Transponder 102, 12,441 GHz vertical. It's Free to Air without any encryption.

However at least on Astra I know that there are several Demostations too airing HD content in MPEG4 AVC over DVB-S2.

[pankov]

http://rapidshare.de/files/13094302/TPS_HD__1920x1080i_.avc-ts.html
it's a little harder to get to the file but it's there
if you read good you will get it
;)
here and in one of my previous posts you can find some interlaced MPEG4-AVC content captured from sats over Europe

[blubberbirne]

*hm* ok, how can i get the pro Version. I'm from germany.

Paypal, Creditcard or what? I have nothing of them ;)

[SeeMoreDigital]

here and in one of my previous posts you can find some interlaced MPEG4-AVC content captured from sats over EuropeSadly I am not always able to download "file samples" from those kind of hosting services :(

[pankov]

if ftp://avisynth.org/ can handle the 42MB file I can upload it there too

[asdfsauce]

I know it might be a bit too early to ask, but are you planning to add hardware acceleration to your encoder as well?

And also, I don't know if it's already been posted in this thread, didn't see it - does anyone know of a dshow splitter for that AVC-TS clip? Tried using VLC, but my computer can't play 1080 AVC without CoreAVC.

[videomixer9]

it's regular ts, MPC internal ts splitter works fine with it