View Full Version : VC-1 encoding problem
Dark Shikari
18th August 2007, 06:29
In the linked tar file are two videos; a WMV encoded with the "insane" preset with wmcmd.vbs using unrestricted 2pass mode at 2 megabits per second ABR, with the WVC1 codec.
The other video is a very high quality (CQ2) Xvid file of the original source. The original source itself is about 370MB so I'd rather not post it.
Now go subtract them in AviSynth... and you'll find that many of the frames are totally out of sync, for some strange reason; but then they go back in sync later, and so forth. Its very obvious the frames are not aligned properly with each other when doing the subtraction, but I don't know why, and worse, its making it impossible to do any sort of quality test at all.
What's the problem?
Download link for videos (http://www.mediafire.com/download.php?8pioongkyxd)
Inventive Software
18th August 2007, 17:13
Did you do a CBR encode? It's possible it dropped frames.
buzzqw
18th August 2007, 17:16
the insane profile is 2 pass encoding
BHH
Inventive Software
18th August 2007, 17:28
Whoops. My bad. :o
buzzqw
18th August 2007, 19:17
i think is the b frames encoding and decoding
i have a similar problem a mounth ago.. and the problem was the b frames encoding/decoding in xvid when there aren't insufficent ref frames
try with max 1 bframes
BHH
Dark Shikari
19th August 2007, 01:02
:rolleyes:
I can't get this to work, even by re-converting the files losslessly back to AVI or anything.
I really don't want to post my next codec comparison with a bunch of 0.92 SSIMs for VC-1 just because it can't seem to order its frames correctly.
Sagittaire
19th August 2007, 10:08
:rolleyes:
I can't get this to work, even by re-converting the files losslessly back to AVI or anything.
I really don't want to post my next codec comparison with a bunch of 0.92 SSIMs for VC-1 just because it can't seem to order its frames correctly.
VC1 with WME9 don't use adaptative bframe. Use VC1 without bframe. Your result will be certainely better.
Dark Shikari
19th August 2007, 12:36
VC1 with WME9 don't use adaptative bframe. Use VC1 without bframe. Your result will be certainely better.
But won't that decrease quality overall, making VC-1 worse?
And I used WME9 and the exact same problem occurred :confused:
Sagittaire
19th August 2007, 14:57
But won't that decrease quality overall, making VC-1 worse?
And I used WME9 and the exact same problem occurred :confused:
In general case no.
But your source is perhaps exception and you have really more exception without adaptative bframe placement (fade, high motion ... etc).
CruNcher
19th August 2007, 15:35
B-frames can still hurt more then help (HVS wise) not as bad as in Mpeg-4 ASP for sure tough, im sure Microsoft uses a more H.264 kind of B-frame aproach but nevertheless it can hurt especialy in combination with RC ;)
benwaggoner
19th August 2007, 15:45
Generally speaking, 1-B frame is the optimum choice for most content with VC-1 for both objective and subjective ratings.
Dark Shikari
19th August 2007, 16:05
Generally speaking, 1-B frame is the optimum choice for most content with VC-1 for both objective and subjective ratings.
1 b-frame still causes the same problems...
Sagittaire
19th August 2007, 16:07
1 b-frame still causes the same problems...
Try without ...
I think that you can tweak the bframe ratio for WMV9
http://www.microsoft.com/windows/windowsmedia/howto/articles/codecadvancedsettings.aspx#ForceBFrameDeltaQP
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.