Capturing a lot of moving objects and camera action games

[Blizzardstar]

What is better for capturing console games where there are a lot of movement? Bitrate or Quantizer? Bitrate makes it easier to calculate for a target size since it seems my segment output file feature seems to be broken, while in quantizer it takes low size with low movement and a lot when there is more movement in the screen making it a little harder to guess and you have to keep adjusting. I don't know about the quality.
Also are there any recommended filters?

[setarip_old]

Hi!

It might be beneficial if you stated what software and methodology you're using...

[Blizzardstar]

Oh sorry, thanks.

I'm using VirtualDub and XviD MPEG-4 codec. I just have the filters that are pre-installed.

[Blue_MiSfit]

:search:.

If you use a fixed bitrate, your quality won't be that great, as XviD is optimized for 2 pass VBR encodes IMHO. Capture to a lossless file first (if you can), and then re-encode with proper 2 pass. Otherwise constant quant 3 or 4 generally gives good quality and manageable filesizes.

[Blizzardstar]

Thanks, but what do you mean by 2 pass? I'm kinda new and ignorant on terms.

[Blue_MiSfit]

again, :search: :)

But, since you're new I will indulge you.

Basically, when you are capturing something using VirtualDub, your codec (xvid) only gets to see the video once - hence 1 pass.

A 2 pass encode is when you have the entire source on your hard drive (like a DVD for example), and therefore can access it multiple times. This is generally the "best" way of doing things (although the word best is somewhat frowned upon here ;)). Basically, the first pass is just analysis and decision making, and then the second pass looks at all the recorded information that was decided upon with the first pass, and then makes more accurate decisions for the final output file. This usually results in more even, and overall higher quality.

In order to facilitate a 2 pass encode from a capture, you first capture to a so-called lossless codec, like huffyuv, lagarith, or ffv1. These codecs do not need 2 passes, because their output when decoded is mathematically identical to the input. They produce gigantic files (smaller than uncompressed at any rate), but allow you to later do a proper 2 pass encode using a more efficient codec like XviD.

So the methodology is:
1) Capture from source using VirtualDub to an AVI containing HuffYUV or Lagarith lossless video.

2) Load that AVI into VirtualDub or other encoding program, possibly through AviSynth to facilitate filtering, deinterlacing, resizing etc

3) Set up an execute a 2 pass encode with whatever settings you want

I don't know how clear that is, but again :search: and you will get the idea!

~MiSfit

[Blizzardstar]

OMG Thank you! I can see the difference now!
The motion blur is 95-99% gone and takes away 40% of file size with the same quality (well it's better in quality since it doesn't have the blurr:) )
So does that means I can have an even better quality with the same old size?
I don't know what do you mean by the different types of 2pass, I used the one on the MPEG-4 codec and it made it not possible to use the quantizer, just bitrate but it still made it way better, almost at the same quality than quantizer (almost not noticable) and with almost half the size.
Are there others in VirtualDub?