Small Matrix Comparison (corrected)

[JarrettH]

Well I corrected the levels...I haven't even had a chance to look at them! The source is there along with: Flat16, EQM_AVC_HR, M4G High Detail V3.1 @ a bitrate of 1080. I used HQ-Slower with No Fast P-Skip and No DCT checked. The values suggested for deblocking by Sharktooth and Mp4 Guy were used too.

Here's all the shots rar'd: http://rapidshare.com/files/8471955/Shots.rar.html


SOURCE
http://img80.imageshack.us/img80/1491/source1zz6.png
A
http://img237.imageshack.us/img237/8499/1fa3.png
B
http://img237.imageshack.us/img237/3932/2xd2.png
C
http://img155.imageshack.us/img155/9629/3pm8.png


SOURCE
http://img147.imageshack.us/img147/6356/source2gw8.png
A
http://img151.imageshack.us/img151/8453/1uz3.png
B
http://img147.imageshack.us/img147/3818/2fb9.png
C
http://img151.imageshack.us/img151/557/3ys1.png


(No Source)

A
http://img367.imageshack.us/img367/5259/1jw8.png
B
http://img220.imageshack.us/img220/2397/2jd7.png
C
http://img48.imageshack.us/img48/137/3wu5.png


SOURCE
http://img48.imageshack.us/img48/8058/source4dp7.png
A
http://img48.imageshack.us/img48/8884/1yx6.png
B
http://img246.imageshack.us/img246/158/2op8.png
C
http://img223.imageshack.us/img223/5655/3nt1.png


SOURCE
http://img176.imageshack.us/img176/5823/source5vk7.png
A
http://img185.imageshack.us/img185/4016/1lq1.png
B
http://img176.imageshack.us/img176/8599/2fn2.png
C
http://img337.imageshack.us/img337/2281/3an3.png

[Sharktooth]

Yeah, now the colorimetry is good ;)

[JarrettH]

I know you guys hate rapidshare but I want to upload them somewhere rar'd. :D

[Sharktooth]

In the sencond and fourth series of SS the source frames are wrong... (hard to spot but that's it...;))

[CruNcher]

The second looks bad to me because it uses inloop i would rather suggest --no-deblock and upping the bitrate or even kill some source details but inloop wish wash i hate it and will hate it forever (yes even in motion it hurts my eyes) :P
To a degree wish wash encoding is acceptable for internet transmission (small timestamp video services like youtube and company) but for home use i see no reason with the compression power of h.264, ever be forced to see it again.
But it's hard to judge the quality of the encode without seeing it fully on screen, but definately you can see the power of H.264 in detail preservation with High Profile in those screens, very nice facial textures (High frequency) try to preserve as much with ASP @ that res and bitrate, anyway users should be carefull with High Profile usage if they plan to playback this on Next-Gen CE Devices.
Sony for example restricted the PS3 not to play High Profile customer created content (and it's unlikely to change). And no Asia H.264 SAPs yet on the market or announced only the official boxes (HD-DVD,Blue-Ray) and those are still very restricted and it seems this won't change that fast either, but sorry for going oftopic.

hehe Sharktooth nah it's easy to spot that the Source frames are of , but he wrote that allready in the start post i think :)

[Sharktooth]

On the second the last SS (C) has clearly a lower deblocking.

P.S.: Even the fifth serie has a wrong Source frame (this one was pretty damn hard to spot;))

[CruNcher]

yes but i wouldn't say clearly i mean used is used only that you see some more High Frequency details doesn't mean it looks subjectively better especialy in motion and this seems to be a high motion scene, to me it looks the same wish wash wise as the others of that batch, bitrate is definately to low or ratecontroll decission bad.

[JarrettH]

Yeah the source are wrong for all actually but the motion is very still except for the bridge. Incrementing in DGIndex has much larger steps. I would have opened the VOB in MPC but they won't playback properly on their own.

[Audionut]

Why not use virtualdubmod.

[JarrettH]

In each series I end up liking shots A four out of five times. C seems to look consistently more dull in appearance and slightly more washed out than the other two.

I still have the clips if you want to download them?? :cool:
They are about 5-13 seconds each in length.

[Caroliano]

1: I prefer A and B over C, but the diference is almost none. I can't see any diference between A and B.

2: C is much sharper than A and B. It has a bit more of ringing, but it shoud be better than a bit more of blocking that A and B has. I can't see any diference between A and B.

3: C is again much better. I can see more and better flying objects, and is less blocky than A and B. Again, no diference between these two.

4: In this scene A and B are better than C. They have a bit more of detail, and less ringing.

5: A and B has a bit less ringing, but other than that, no significative diference.

There is something wrong, because A and B are always identical (at least in the rar files, where they are named 1 and 2)

[JarrettH]

Gah, megui is buggy as hell right now. Wouldn't be surprised if it ended up using the same matrix twice.

[nurbs]

Well, you could always look at the log to make sure it used the right matrix.