Oops. I accidently didn't apply a variable name change in the VAQ-patch. I reviewed the changes, compiled a new build and tested it. Now Variable Adaptive Quantization (VAQ) really works in the VAQ-build. If you want to use it, please download the installer again and update your installation with it.

Oh. And of course: A Happy New Year!

The Directshow-filter in the first version of the recent installers didn't work properly. I finally managed to properly compile the filter, now it works perfectly. I silently updated the installers. If you already have downloaded the build and don't use for example ffdshow, please re-download and install the (silently) updated version. Sorry for the fuzz!

Isibaar published Xvid-1.2.0 earlier this week. Today he published a bugfixed version 1.2.1 which I could finally compile. I had to setup compilers, DirectShow- and Platform-SDKs and more, so it took me a while. I hope you don't mind. This build replaces all the older stuff - 1.2.-127 is obsolete, and 1.1.3 is outdated with that, too.

I added Dark Shikari's variable adaptive quantization patch again. Please test it by using luminance masking. First tests show that the build works as it should. That's the build with -VAQ, the other build is from the plain sources.

Plugh pointed me to another bugfix that made it into CVS but wasn't backported to Xvid-1.1.3. There was an issue with motion vectors (MVs), an array was too small so a wrong MV could have been chosen in some cases. Since those MVs are costly bitrate-wise, it should have happened very seldom. Anyhow, I updated Xvid-1.1.3 and Xvid-1.1.3-VAQ.

Plugh pointed out an error in Xvid-1.1.3 that occurs when calling the plugin-system when there is a b-frame. The call may have handed the wrong frame, for example a p-frame to the plugins. This is a kind of a back-port of the patch in CVS.
Both 1.1.3-builds are corrected, the "vanilla CVS" and the one with Dark Shikari's VAQ-patch.

Some users requested a build with VAQ based on the old SMP-experimental code from sysKin (Xvid-1.2.-127-VAQ).

I found some time to download and apply Dark Shikari's port of h264 variable adaptive quantization. Enable it by activating Adaptive Quantization in the GUI. Please be aware that I didn't check if AQ still works, I just applied the patch and created a new installer. You can find the patch from Dark Shikari here. A discussion about the code is taking place over at Doom9's.
On another note, I started a blog at Google's Blogger this morning. Every now and then I'll post something there which isn't necessarily fitting into this page directly. It is called Koepi's Blog, not too surprising.

Good news again: matrics pointed the www.koepi.info-Domain to my website! So now my site is reachable via a proper domain again. :-)
Unfortunately I didn't find the time to compile an experimental build with Dark Shikari's VAQ-patch. It's supposed to provide a more aggressive lumimasking (err, Adaptive Quantization, that is), which lowers and raises quantizers of macroblocks within a frame so that blocks with sharp edges get higher quantizers and thus use less bits. Sounds interesting for a test.

Finally! My site is back up. Some domain grabber managed to fetch the koepi.org domain so for now I moved to koepi.leffe.dnsalias.com. Many thanks for the hosting, Leffe!

Also on the bright side, the domain koepi.info has been registered by Doom9-Forum member matrics and it points to the freecodecs.com-mirror of my xvid binary. So things will be running smoothly shortly!

A possible security issue has been reported in xvid. Isibaar and skal fixed the issue with a patch which has only very little performance impact. Please download and install the new XviD binary!