Is the bluesky clip blocking issue solved?

[Chainmax]

If not, has the malfunction been identified? When could a fix be expected?

[Manao]

Did you try the AQ patch ? Because it's not a malfunction, and it can't be solved without AQ...

[Chainmax]

I'm using x264 rev351, does that include this patch you're talking about?

[IgorC]

:search: . Sharktooth compilation has aq and other patches. It's sticky.
You can perform test yorself . Here is a sample of a RAW HD 1920x1080 ftp://ftp.ldv.e-technik.tu-muenchen.de/dist/test_sequences/1080p/blue_sky.yuv

It's simple to see on this sample the difference.

[akupenguin]

The bluesky malfunction is not just blocking. There are also incorrect MVs used, which (in high QP B-frames) is even uglier than blocking. Sufficiently stong AQ will hide that too, but it's not the best way.

[Chainmax]

IgorC: I visit the sticky many times to download the latest version, but since I never saw the "applied" word in the patches section I thought they had to be downloaded separately :o.

akupenguin:what would be the right way to solve the issue?


Is AQ applied automatically then? I don't see a switch for it in MeGUI...

[foxyshadis]

Is AQ applied automatically then? I don't see a switch for it in MeGUI...
It's under advanced, with the custom matrix and zone options. I guess since it's still somewhat experimental.

[Chainmax]

I see, thanks. All the posts I read about it mention only one setting for it, I assume it's strength and they leave sensitivity untouched. I will try strength 0.3 and work my way up from there.

[bond]

I see, thanks. All the posts I read about it mention only one setting for it, I assume it's strength and they leave sensitivity untouched. I will try strength 0.3 and work my way up from there.read again:
http://forum.doom9.org/showthread.php?p=730966#post730966

[Chainmax]

I see, so no AQ for the time being then (by the way, the default is strength=0). Thanks for the heads-up.

[bond]

I see, so no AQ for the time being then (by the way, the default is strength=0). Thanks for the heads-up.yeah the default is 0, which means its disabled

i refered to sensitivity when talking about the default

[Chainmax]

Is AQ not recommended for animation, like its Xvid implementation?

[Sharktooth]

it's completely different from xvid AQ.

[Chainmax]

So it's safe for all kinds of content? Where can I find more info about it?

[foxyshadis]

The source, akupenguin, and test encodes, I think. (It definitely does an excellent job in high dynamic range anime, including day and night scenes as well as fades. I haven't tried it on other sources yet.)

The only thread I've seen discussing it is this one (http://forum.doom9.org/showthread.php?t=101759), and I haven't seen anything pop up on the mailing list. Bond says that it hurts more than it helps on high-bitrate encodes with high strength and sensitivity, so it's most likely more tuned to helping mid/low bitrate encodes.

[bond]

The only thread I've seen discussing it is this one (http://forum.doom9.org/showthread.php?t=101759), and I haven't seen anything pop up on the mailing list. Bond says that it hurts more than it helps on high-bitrate encodes with high strength and sensitivity, so it's most likely more tuned to helping mid/low bitrate encodes.i encoded at 700kbps which i see as a mid bitrate and as the bitrate you will use at a 1cd encode

also i did use strenght 1.0 which is indeed not a low setting but together with sensitivity 10 which is the normal setting

and i have to say that it removed details clearly overall on my test clip when comparing to a clip without aq

also i dont think aq was even written for enhancing overall quality/detail level. it was written for placing more bits in darker flat areas

[foxyshadis]

i encoded at 700kbps which i see as a mid bitrate and as the bitrate you will use at a 1cd encode

also i did use strenght 1.0 which is indeed not a low setting but together with sensitivity 10 which is the normal setting

and i have to say that it removed details clearly overall on my test clip when comparing to a clip without aq

also i dont think aq was even written for enhancing overall quality/detail level. it was written for placing more bits in darker flat areas
Gotcha, I thought you were testing around ~1100 for some reason. Must have confused it with another thread.

[Chainmax]

I made a test encode with AQ strength=0.5 and sensitivity at default (15) and *.mp4 guy's low bitrate matrix. I'd appreciate some feedback. You can download it from [link deleted].

[stephanV]

I'm not sure what comments you are looking for, but if this was meant to decrease the blocking discussed here, it didn't really work. :(

[Chainmax]

Should I use stronger inloop (currently at 2/2), stronger AQ or add slight denoising to the filterchain?

[foxyshadis]

Stronger inloop will just munge texture. Playing with AQ sensitivity might help, but maybe the bright whites are causing Vash's costume around ~0:50 to block up. Unless that's in the source?

My processor almost redlines playing that. =x I think it's time for an upgrade.

[DryFire]

The guns also look rather blocky/blurry. Though it is still better looking then an old 233mb wmv3 backup I have.

[Chainmax]

The source gets horribly blocky in spots, especially in Vash's costume. I have a thread about SmartSSIQ and Bifrost in the avisynth forum where I posted some pictures. I'll try using BicubicResize(640,480,0,0.5) and add a RemoveGrain(mode=5), maybe that will be enough.