I don't know how many people use the GOP structuring and the quantize matrix, but i would like to learn how to so I can try to produce better vcds. After experimentation, I currently have my GOP structure as I,P only. I have not done anything to the quantize matrix. If anyone can explain or send me to a website, that would ver helpful.
GOP Structure and Quantize Matrix in TMPGenc
Collapse
X
-
A poster named "rocknroll" claims that the following settings will create the "ideal" VCD. Understand that I in no way am suggesting this, rather I'm merely passing it along:
"Do you need a great Matrix incl. GOP's for TMPGEnc ?
We have tried 1/2 year for the matrix and the quality is absolutly perfect for VCD and SVCD.
Under 'GOP structure' use 1,4,2,1,0
and for the matrix:
Intrablock: Non-intrablock
08,13,15,19,19,23,23,29 08,11,13,15,15,19,19,24
13,13,17,19,22,23,27,29 11,11,15,15,19,21,24,25
15,17,19,22,23,27,29,35 13,15,17,19,21,24,25,34
17,21,22,23,27,29,35,37 13,17,19,21,23,26,34,99
21,21,23,27,31,33,37,46 17,19,21,23,26,32,99,99
21,23,27,31,33,37,46,99 19,21,23,26,32,99,99,99
25,25,31,33,37,40,99,99 22,22,28,30,99,99,99,99
25,31,33,37,40,99,99,99 22,28,30,99,99,99,99,99
unmark: Output YUV data... & No motion search...
mark: Use floating point DCT...
then:
mark: soften block noise
Intra block=2 Non-intra block=4" -
I realize I am replying to a very old post but I was searching for information about the intra-block and non-intra block settings in TMPGEnc.
I have a pretty good feel for the use of the "soften block noise" setting but have found myself sacrificing forground (human faces, etc.) sharpness for the sake of reducing block noise in areas such as a backround wall or sky.
I have always used the "floating point" setting and the "soften block noise" setting when necessary but never messed with the matrix.
I decided to try the matrix that setarip passed along and I am very pleased with the results. As input, I used a high bitrate(1500 or so) Xvid DVD rip created by myself from a DVD I own.
Not only did the suggested matrix dramatically reduce that backround blockiness, but it did so without blurring the rest of the picture. Furthermore, it reduced the appearance of blocks during high-motion scenes, like when someone or something is moving against a backround, all the while keeping things looking nice and crisp.
It slowed down encoding time only very slightly.
In my opinion it is better than the default matrix and I plan to keep using it. I recommend giving it a try.Last edited by DrinkOrDie; 20 Jan 2005, 06:26 PM.My toy:
Custom Build PC Born on 03-08-08AMD Athlon64 X2 6400+ (3.21Ghz) (Black Edition) on overclock ready Asus M2R32-MVP Crossfire AMD chipset Mobo, 8-Channel HD Audio, Windows XP MCE, 2GB 800Mhz DDR2, 512MB NVIDIA GeForce 8400 GS, 500GB SATA-II HardDisks X 4 in RAID 4 mode,(1.5 TB storage capacity + 500GB eSata external) Sony MultiRec DVD-RW, PCI-HDTV Tuner, SOYO Topaz S 24" Wide LCD 1920X1200Comment
Comment