ShiZZZoN- couple things not to do

[sean madison]

I just wanted everyone to know a few things you should know about tmpgenc so that if you ever had these probelms, you know they can be easily avoided-

1- if you have encoded and burned a movie and when played back on tv, the vid is shoved way up in top left corner very small even though u encoded at 352x240 or any other standard res
ANSWER- when you encode to vcd or svcd, make sure in Nero, you choose the same type of movie that u need to burn(if u encoded a vcd, burn as vcd, if svcd, burn as svcd)

2- If you are unable to import an avi into tmpgenc with the following error message cannot open or unsupported, do this-
ANSWER- go to options, environmental settings, then go to the VFAPI tab. In there, I would presume you already have Direct X 8.0a, if you do not or have restored your registry, you will need to redl direct x 8.0a. Now in this tab, make sure all priorities are set to or below and move the directshow avi plugin to a priority to 5. This will save u time for having to re-encode it in virtual dub.

I need a question answered too for all of this-
In quantize matrix, where it lists for intra and non intra block, what exactly is that, what happens when you customly change it, what are good settings, and can i change it to whatever i want?

[KoalaBear]

Short answer: unless you're working towards a degree in mathematics and/or perceptual psychology, you should probably leave them alone.

Long answer: When a macroblock undergoes DCT, the values representing the pels of the original matrix get squeezed to an 8x8 block of coefficients. This block has the interesting property that values representing low spatial frequencies aggregate in the upper left hand side of the block while high spatial freqencies aggregate in the lower right side of the block.

The coefficients are then scaled (divided) by the quantization matrix which is designed to truncate those values in proportion to their spatial frequency. This is where the "lossy" part of MPEG compression occurs. The higher the spatial frequency, the lower the probability your eye will notice a loss, the more of those values will end up being truncated to zero.

Both the luma and chroma portions of the macroblock are quantized using standard tables optimized for those purposes. Unless you know a great deal about the way the system is designed to work there is essentially no way to improve upon them.

However, some encoders can generate optimized quantization tables on-the-fly based on the properties of the material being encoded (the so-called "Q Factor"). These encoders are your best shot at improving over the standard because they allow you to prioritize picture detail over compression efficiency without having to worry about the underlying details.

<font size=-1>[ This Message was edited by: KoalaBear on 2001-11-06 11:07:43 ]</font>