CoreCodec / CoreAVC ranting (un)official thread

[BetaBoy]

We are working on many of the concerns being adressed atm... including an Activation Wizard that is installed on the users PC to self-serve activation/de-activation without any interaction of the Core Account system.

Once this is completed we will move it to our services home and release the new update with the latest fixes.

[_xxl]

"dev team here" did nothing about h264 decoding speedup in ffdshow. Read teh changelog. You're probably confusing the ffmpeg devs that make all the decoder improvements with the people that merge those changes into ffdshow. Let's give credit where it's due, shall we ? And ffdshow still uses only one core for decoding, not to mention interlaced decoding is a pain, but then it's sometimes a pain in CoreAVC too.
Please, don't forget that FFDSHOW is released as an open source software under GPL license agreement..We do it for free.If you don't like ffdshow don't use it!

[Manao]

I don't think KoD's statement was intended as a critic of the dev team. Crypto had just said : "ffdshow has gotten better and better and performs now in the core range, with nearly its speed but more precise decoding", which is just not true (yet), so KoD corrected it.

[KoD]

I appreciate the work haruhiko, foxyshadis, clsid and drevil do to keep ffdshow updated and to fix some of its annoyances. Truth be told, I think haruhiko needs more recognition for his work than he currently receives in the ffdshow thread here on the forum as he is the main force behind significant code changes.

But let's be honest, the merit for decoder improvements go to the ffmpeg devs as ffdshow simply merges those changes into its source tree. And the current speed of h264 decoding in libavcodec is not near that of coreavc. We may speak what we want, but tests prove how things really are and not how we want them to be.

Please also note that some of the people working on ffmpeg as well as other free video and audio related projects are also those that during "job hours" work for the companies that receive so much flack here. You see, open source does not put bread on the table, at least as long as it doesn't "fight" against a widely perceived evil (I'm watching at you, Firefox & IE !) and does not have to breach patents (since almost all the tech in the video and audio realm is covered by patents).

[f@chance]

For me it is finallythe end of the line. After purchasing CAVC 1.2 on the day of its rlease I Never got the download link, Never was able to open my account because of Bad User Name or Password.

Dec 6th PayPal issued a refund for a product that I never downloaded nor installed. After reading these pages and pages about the Activaton PITA how glad that I will n ot use CAVC 1.2.

As it was stated in this thread many times this business decision will hurt the end user and they do not want the hassle of this activation scheme.

[Seb.26]

[...] with the latest fixes.
Is it possible for you to tell a little bit more about this "latest fixes" please ?
( maybe fingerprints+eye-iris activation ?! :D ... sorry ... :p )

[TheFluff]

I use riverbed (ftp://ftp.ldv.e-technik.tu-muenchen.de/dist/test_sequences/1080p/riverbed.yuv) encoded lossless with CABAC and CAVLC to test against. I get:

CABAC (23Mbps):
ffdshow - 16.1 dfps
coreavc 0.4 - 19.1 dfps
coreavc pre-1.0 - 37.3 dfps

CAVLC (26Mbps):
ffdshow - 19.1 dfps
coreavc 0.4 - 22.5 dfps
coreavc 1.0-pre - 44.5 dfps
What did you measure with? Haali's timecodec isn't accurate at all for multithreaded decoders, AFAIK.

Also, h264 support in ffdshow hasn't really changed since the big speedup push a few months back.
If you by "a few months" mean "about 5 weeks", then OK. But there was also another, bigger optimization run about 7 weeks ago, I guess you might mean that, too...

[foxyshadis]

I should check times more often, because relying on my memory is fraught with danger. :p I meant the spring and october flurries of patches.

Tested with timecodec, null renderer. I realize the fps value is off, but dfps should be correct, since it measures total time to decode the full stream (and thus is rather sensitive to having other things going on in your system). Unless there's more wrong with it.

[HookedOnTV]

My concern with the activation issue is what happens 6 months from now when my hard drive takes a dump and there is noone on the other end of email requests for a replacement activation.

[BetaBoy]

My concern with the activation issue is what happens 6 months from now when my hard drive takes a dump and there is noone on the other end of email requests for a replacement activation.

if you dont change your Windows serial number then there is no need to worry... but this is the reason why we have created the Core Activation Wizard that will be bundled with all our our products.

[BetaBoy]

Is it possible for you to tell a little bit more about this "latest fixes" please ?
( maybe fingerprints+eye-iris activation ?! :D ... sorry ... :p )

Yes, in the next version we will make you stare at the flashing HD light to read the 'morse code' serial number ;-) . . . We have fixed the issue related to TheFluff's posting and are working on the interlacing problem as well. I could also tell you all about GPU but I feel it best to fix the last few reported issues before we start talking about that.

[DeepBeepMeep]

Talking about the GPU support might not be a good idea ... it will best next time simply to show it :).

However, now that you are talking about fixing bugs, are you going to fix the incorrect color ranges for some H264 streams (black level is at 32 instead of being at 16, same problem for the whites I think). It is not clear based of your posts that you are aware of this bug. I could provide a sample file if this help.

[BetaBoy]

Talking about the GPU support might not be a good idea ... it will best next time simply to show it :).

However, now that you are talking about fixing bugs, are you going to fix the incorrect color ranges for some H264 streams (black level is at 32 instead of being at 16, same problem for the whites I think). It is not clear based of your posts that you are aware of this bug. I could provide a sample file if this help.

Pls do!

[Foreigner999]

How long has coreavc 1.2 been out so far? ;)

Seems it's already been cracked. :devil:

Betaboy any news on CoreAAC 2.0 that was planned to be included?

[Seb.26]

Yes, in the next version we will make you stare at the flashing HD light to read the 'morse code' serial number ;-) . . .
Happy to see you still have humour ... ;)
( And I think you need it everyday with us ! )

We have fixed the issue related to TheFluff's posting and are working on the interlacing problem as well. I could also tell you all about GPU but I feel it best to fix the last few reported issues before we start talking about that.
OK ... you are right : first stabilize actual release before add new features ;)
( but release GPU fast too !!! :D :D :D )

<< May the force B with U .. >>

However, now that you are talking about fixing bugs, are you going to fix the incorrect color ranges for some H264 streams (black level is at 32 instead of being at 16, same problem for the whites I think). It is not clear based of your posts that you are aware of this bug. I could provide a sample file if this help.
Have you a link about this point ?!
( I haven't seen any problem about that ?! )

[Inventive Software]

I must quote this from the other thread about the activation. Thanks clsid for a good laugh. ;)

A legal copy of CoreAVC Pro 1.2: $14.95
3 bottles of coca cola regular: $4
2 bags of potato chips: $2
Spending 5 hours to crack the activation: $0

Uploading a torrent to piratebay: PRICELESS

[DeepBeepMeep]

Pls do!


Ok I have sent a 1Mb sample. I have tried a couple of Email addresses I found. Please let me know if you got the Email. For those who are interested in this issue, here is the content of the Email:

"Hi,

Following my post on Doom9, please find a very short TS sample of 1080p film which has incorrect black and white levels (here black looks grey). It seems this problem is typical on everything capped on SkyHD in UK and PremiereHD in Germany.

In fact everything happens as if the option 'Fix VMR9 Level' was not checked. After a deeper study it appears this problem is visible if the CoreAVC IS NOT the last filter in the chain even if the video renderer is VMR9. In fact I tend to use ffdshow as the last filter of the chain for postprocessing.

If now the renderer is the Overlay mixer, even if CoreAvc is at the end of the chain the black level will be wrong since it will look grey.

My assumption is that you rely on the DirectShow (hardware?) system filters to fix the color range and that this works only if CoreAVC is the filter before the VMR9 renderer.

It would nice to fix this issue since I need to use a filter in ffdshow to fix the color range which consumes precious CPU. In addition I am pretty sure this color conversions may cause the picture to lose quality."

[oddball]

There is a crack out for 1.2 just so you know. Not gonna say the group name or anything. I paid for my copy and only need to run it on a single system so not too bothered really. But just so you guys know this Activation stuff is a waste of time and only penalises legitimate users.

[clsid]

Too bad that it requires Windows XP :( It uses some kind of suspicious packer as well.

[sillKotscha]

It uses some kind of suspicious packer as well.

found a torrent with just the *.ax file and regdrop included in the folder...

... searched around just out of curiosity :)

[ChronoCross]

hopefully there will be some DMCA pulldowns on anyone offering it.

[foxyshadis]

This thread is so going to get rule 6'd if this conversation continues.

[Sharktooth]

Removing DRM is a right (if the product was legally obtained)... otherwise FairUse4WM would be considered a rule 6 violation as well...
However i was able to find even a keygen, a patched .ax and a crack... in less than 5 minutes.
Obviously i dont support warez (just had some fun searching usenet and p2p channels) but this should make companies think twice or even more before investing time and money into DRM and protection schemes. They only create problems (meaning more time and money investment) and more unhappy customers that will go the "illegal" way to get the same product for free and without hassles...
Be nice and honest with your customers and they will be happy to buy your product...

[dragongodz]

guys please lets try and not get this thread closed.

also
Is it possible for you to tell a little bit more about this "latest fixes" please ?
questions like this should go in the normal CAVC thread, not this one.

I could also tell you all about GPU but I feel it best to fix the last few reported issues before we start talking about that.
unless you are getting ready to release a version that has GPU support all you are doing is marketing vaporware with that statement. you should already know people were not impressed about you hinting you were working on GPU a LONG time ago for nothing to appear. if you keep mentioning it in this way you deserve any and all criticism you get for it not being implamented whenever a new version is released.

[Seb.26]

guys please lets try and not get this thread closed.
questions like this should go in the normal CAVC thread, not this one.

<< SIR ! YES ! SIR ! >>

[f@chance]

This was a no brainer, didn't even have time to make a latte before finding it.

As a former ISV it pains me though to see hard work down the drain, but you reap what you sow. We wouldn't have this discussion unless the paying public got pi**ed of by the activation scheme. It is a crying shame that the resources spent on the core release system were not spent on the product enhancement.

[Rumbah]

But let's be honest, the merit for decoder improvements go to the ffmpeg devs as ffdshow simply merges those changes into its source tree. And the current speed of h264 decoding in libavcodec is not near that of coreavc. We may speak what we want, but tests prove how things really are and not how we want them to be.

I just tried the new libavcodec in VLC 0.8.6 against CoreAVC 1.2 with the Spiderman 3 720p trailer on a P4 3,2 Ghz HT processor. And I have to say I'm quite impressed with the new VLC. With CoreAVC the processor usage while playing it is between 10% and 18% while in VLC it is between 0% and 8%.

[Sirber]

I just tried the new libavcodec in VLC 0.8.6 against CoreAVC 1.2 with the Spiderman 3 720p trailer on a P4 3,2 Ghz HT processor. And I have to say I'm quite impressed with the new VLC. With CoreAVC the processor usage while playing it is between 10% and 18% while in VLC it is between 0% and 8%.That's great! :D

[SeeMoreDigital]

I just tried the new libavcodec in VLC 0.8.6 against CoreAVC 1.2 with the Spiderman 3 720p trailer on a P4 3,2 Ghz HT processor. And I have to say I'm quite impressed with the new VLC. With CoreAVC the processor usage while playing it is between 10% and 18% while in VLC it is between 0% and 8%.How does the speed compare to FFdshow's libavcodec H.264 decoder...

[Sirber]

(just had some fun searching usenet and p2p channels)It's easier than you think...

[Sharktooth]

Oh, well... even on the web...

[Rectal Prolapse]

But does VLC do deblocking and deinterlacing?

Oh wait - you said 720p. Was deinterlacing disabled in VLC and CoreAVC? I'm not sure if CoreAVC will try to "deinterlace" a progressive stream if that is enabled - if so it would chew up more CPU cycles.

Does VLC use Overlay or VMR9? If the CoreAVC test was in VMR9 and VLC is in Overlay, you should see a big difference.

[Sharktooth]

Deinterlacing is triggered only on interlaced sources.
However VLC uses libavcodec that has a more "compliant" output than coreavc (due to the blocking bug in coreavc).
So basically if Rumbah results are correct (but i doubt it - maybe it was a VMR vs Overlay comparison), libavcodec looks better and is faster than coreavc.

[Rectal Prolapse]

Thanks for the clarification.

[KoD]

I should say that these kinds of low VLC cpu usage when decoding avc is not something new. It was reported months ago, but the general consensus was that the real cpu usage of VLC is higher than what taskmanager shows.

[Sharktooth]

that's a nonsense...

[Rectal Prolapse]

That could be possible if a lot of work is done by worker threads not owned by VLC. However, overall CPU usage should be mostly accurate, even if CPU usage of VLC itself is low. Right? :)

[Sirber]

right :)
you cannot hide CPU power! mouhahahahaha!

[KoD]

What can I say ? Please enjoy avc playback with 0-8% cpu usage, then. As for me, I have higher cpu usage while watching a simple dvd.

[Rumbah]

How does the speed compare to FFdshow's libavcodec H.264 decoder...

I installed ffdshow tryouts revision 684 Dec 11 2006, and I got a cpu usage playing that file of about 10% - 23%. What I noticed is that CoraAVC uses at least two threads for decoding (both HT "cpus" work) while ffdshow uses only one(CoreAVC and ffdshow playback with Media Player Classic).


Does VLC use Overlay or VMR9? If the CoreAVC test was in VMR9 and VLC is in Overlay, you should see a big difference.
I think VLC uses Overlay, at least it works with the ATI theater mode. I set MPC output to Overlay mixer, so it should use Overlay, too. (I think VLC does not use DirectShow, perhaps that's a big advantage regarding speed)

(but i doubt it - maybe it was a VMR vs Overlay comparison)
As mentioned above, I think both used Overlay.

As for me, I have higher cpu usage while watching a simple dvd.
Watching DVD I have a cpu usage of about 2% in VLC and 10% in MPC with the internal decoders.

[bob0r]

http://news.bbc.co.uk/2/hi/technology/6182657.stm

:devil:

[Foreigner999]

Hmm seems the coreavc rant thread has turned pretty luke warm after the cracks were made public. I wonder if this was just a coincidence or the maybe just the free market speaking?

*****************************************************
Gates said that no one is satisfied with the current state of DRM, which “causes too much pain for legitmate buyers” while trying to distinguish between legal and illegal uses.
*****************************************************
From: http://www.techcrunch.com/2006/12/14/bill-gates-on-the-future-of-drm/

[foxyshadis]

There've been more posts the last 3 days than the 3 days before the crack was released, so.... It just means there's not much more to say, since it's up to Core to make the decision. Until a new version comes out and dozens more show up to rant about Core's lack of purchase automation, that is.

[JohnnyFu]

Hmm seems the coreavc rant thread has turned pretty luke warm after the cracks were made public.

thanks for the hint, that means i can finally reinstall my windows without loosing well paid software.

[3ngel]

including an Activation Wizard that is installed on the users PC to self-serve activation/de-activation without any interaction of the Core Account system

That's the FIRST INTELLIGENT THING i hear from Betaboy.
In this case, if this situation will REALLY be implemented WITH NO INTERACTION WITH SERVER, and NO NEED FOR CONNECTION, then i think it will be a good thing for you, for the users on this forum.
You will regain ALL the lost respect.

[Px]

that's a nonsense...
No, programs can cheat so-called "OS CPU Usage" timer, to avoid that some program read cpu usage from internal cpu sources. For example - RMClock (cpu.rightmark.org). Some time ago I made tests with different video decoders, with apple, powerdvd and coreavc average cpu usage were similar, but with tcpmp "OS usage" and "CPU usage" differs for near 20%, so "OS usage" shows lower cpu usage, than it really was....

[oddball]

Latest VLC runs like poo on my setup. I use Zoom Player beta 5 RC1 and overlay for speed. Beats all other players and decoders with CoreAVC 1.2 installed. I can just about playback 1080P x264 video on a AMD 3200 XP. VLC or ffdshow just stops dead almost on playback and seizes all CPU so much that ending the player takes ages.

[Inventive Software]

*Bump*, just in case people are still feeling rantish before Xmas. ;)

[bob0r]

I have given up on coreavc, quicker updates my shiny white ass. (If DVBviewer deinterlacing mode, display widescreen as 4:3 is suppost to be fixed, why not release a minor update just for that? Hell SVN like updates would be perfect for me, i dont mind updating once a day/week/month)

I am beginning to like Cyberlink H.264 decoder in software mode more and more.
And seeing new CPU prices drop like flies, decoding 1080p_x264_all_cpu_load_costing_high_profile_settings would be easy with ffdshow even on half a core.

I always had great respect for coders who code something great from the ground up, but when commerialism is holding them back to produce more updates, they get grumpy when bugs are found, and they are not in the mood to continue, it pisses me off :/

Products worth my money(buying or donating) due to proper coding and bug fixing: 2
- ioFTPD
- FlashFXP
- <strike> something avcish here</strike>

[rura]

I'll post and rant a bit. I bought 1.0 the day it came out and they never emailed me about 1.2. I've tried emailing but I have not received a response and it doesn't look like I ever will. Activation (while disgusting) is the least of my concern when I can't even gain access to my legitimate copy... Suffice it to say I've used alternate means to get 1.2 and don't regret it in the least. CoreCodec is bar none the worst software company (from a support standpoint) that I have ever dealt with in the 17 years since I got a computer.