aq-strength - grainy/clean sources

[rahzel]

When is it better to use a high or low aq-strength?

I've encoded a grainy source where I found an aq-strength of 0.60 worked best. On a clean source, I found that an aq-strength of 0.70 worked best. From what I've heard, it's best to use a lower aq-strength on both clean and grainy sources. So when is it better to use a higher aq-strength; like say, 0.9-1.2?

[sneaker_ger]

From the tunings:
still image: 1.2
default and film: 1.0
animation: 0.6
grain: 0.5

These are the recommendations of the x264 developers, but everyone has his own opinion, as you've already "heard".

[MatLz]

I've encoded a grainy source where I found an aq-strength of 0.60 worked best. On a clean source, I found that an aq-strength of 0.70 worked best.Seems that 0.65 should be a good compromise for you. :D
(But I don't think AQ is that linear)

AQ is really a strong knob wich drasticly changes the perceived quality.
But since quality is relative to the user, you have to choose what it is looking best for you.

To SHORTLY resume :
It is Ringing versus Banding.
High AQ strength retains more gradients but produces ringing.
Low AQ strength preserves edges but produces banding.

But dont forget psy-rd ;)

[burfadel]

Depends on whether you are using Variance (mode 1, default) or Auto-Variance (mode 2).

[creamyhorror]

AQ makes a lot of difference under certain conditions.

From an old post on Videohelp (http://forum.videohelp.com/threads/308482-high-quality-mpeg-2-encoding?p=1907070&viewfull=1#post1907070) (by poisondeathray), about the difference that aq-strength (under mode 1) makes:
This is a short AVCHD 1080i30 camcorder clip, yadif, lanczosresized 720x400, encoded at 1000kbps. You can find the original here for your own testing and learning: http://www.megaupload.com/?d=9B84U1II

It's a good clip to illustrate what x264's AQ does. Now there is no model here to distract you :) . Pay attention to the shadow detail, and details in the water (and under the water). After examining this clip, go back to the FFV1 clip and re-examine that one. If you still can't see the differences clearly, I'll post cropped image to highlight some of the bigger differences.

Not to sound like a broken record, but again this is a single frame just to illlustrate a point, and you should be looking temporally at the whole clip, different frames, different quality measures, etc. etc....blah blah blah....you get the point. Note I only posted AME , but every encoder that doesn't have AQ like functions has similar results (I tested about 20 different ones)

Original
http://i28.tinypic.com/f1z8tz.png

Adobe Media Encoder CS4 (based on MC)
http://i28.tinypic.com/e684ub.png

x264 AQ=0
http://i32.tinypic.com/2s85r3m.png

x264 AQ=0.5
http://i27.tinypic.com/30ue25j.png

x264 AQ=1.0
http://i26.tinypic.com/23sw4sy.png

x264 AQ=1.5
http://i32.tinypic.com/2w4mx6s.png

As poisondeathray showed, at low bitrates and under different aq-strengths, large differences in bit distribution can become apparent.

[rahzel]

Seems that 0.65 should be a good compromise for you. :D
(But I don't think AQ is that linear)

AQ is really a strong knob wich drasticly changes the perceived quality.
But since quality is relative to the user, you have to choose what it is looking best for you.

To SHORTLY resume :
It is Ringing versus Banding.
High AQ strength retains more gradients but produces ringing.
Low AQ strength preserves edges but produces banding.

But dont forget psy-rd ;)
Hehe.. I've used more than 0.7 and 0.6... I've used anywhere from 0.6-0.9 (depending on the source). I know aq-strength greatly affects the picture, and I do adjust it based on the source. I guess I'm just wondering because I personally never found that a setting of 1.0 or higher to produce a good result. I usually find that it produces too many artifacts on faces/edges. And it doesn't make sense to me that a lower aq-strength is recommended for clean AND grainy sources, but a higher aq strength is recommended for "film". What exactly is considered "film"? Maybe I just haven't done enough encodes yet, I dunno.

BTW, I've used both aq mode 1 and 2. I'm not 100% sure of the differences. I just used whatever works best to my eyes. I do usually find that my QP averages are higher if I use mode 2, but sometimes it still looks better despite the higher QP averages.

Regarding psy-rd, my understanding is that it's a sharpness retention by applying less compression to edges/grain and that psy-trellis takes bits from flat/low detail areas and applies it to edges/grain. So with that said, would it be good to decrease it if you do notice some blocking or gradation issues in dark scenes and/or for clean sources? If it's a grainy film, is it better to increase psy to help retain grain?

[nibus]

I think sources are so different that there is no "better" with AQ, even with sources that appear similar at first glance. You really just have to experiment with every encode if you want the optimal settings. And MatLz is right, I've found the best results come from tweaking psy-rd along with AQ. It also depends on the resolution you are encoding to - I've found HD material looks better with a lower aq while SD material needs a little higher.