VirtualDub 1.6.11 is out in the wild, and captures a couple of months of bug fixes. The major change in this version is that, as I noted last time, I've spent a good deal of time expanding the help file. In particular, a lot more of the new 1.6.x capture system is now documented. There are still some holes here and there, but I'm beginning to think that I should do another experimental drive in the 1.6.x series and revisit the capture system, especially since 1.7.x is a ways off anyway (there are some issues with latency that I need to work out).
I did sneak a new feature into this build, though: support for custom video display shaders. This doesn't add any new processing capabilities, but it does allow arbitrary vertex and pixel shaders to be used to control the video display panes. In particular, it permits experimentation with filtering techniques beyond the usual point/bilinear/bicubic set, if you have a powerful enough video card and are sufficiently adept with Direct3D shaders. The file format is the standard Microsoft effect file format (.fx), and the surrounding setup is documented in the help file.
I intend to go into more detail on issues with capture and video shaders, but there's too much to go into here, so those will have to be separate blog entries. In the meantime, click Read More... if you want to see the changelog.
Build 23774 (1.6.11, stable): [October 2, 2005] [features fixed] * Added support for using D3D .fx files for hardware-assisted display. This requires d3dx_25.dll (DirectX 9.0c April 2005) to work. * Added keyboard shortcuts for scene forward/backward. [bugs fixed] * The current frame could not be copied after a "scan for errors" command. * Fixed crash in DV type-1 audio resampling code at end of stream. * Fixed audio read errors on DV type-1 files with drop frames. * Fixed a couple of crash bugs related to using positive audio displacements. * Fixed a rare filter crash triggered by a previously failing filter suddenly becoming able to start while the filter list is being edited. * Fixed bad write length in .wav writer. * If precision was set to "no change" but channel conversion was enabled, the audio was converted to 8-bit. * Error descriptions for failed jobs were often truncated after exiting and restarting the app. * JPEG decoder sometimes reported false parsing errors due to not parsing over comment blocks. * Filter load and AVI append functions could report "operation completed successfully" rather than the proper error. * AVI File Information reported an incorrect preload value for AVIs that started with a video chunk. * Capture: Stop conditions were being saved even if Accept was pressed. * Capture: Emulation test driver crashed on activation if the last used video file couldn't be found. * (AMD64 only) Disassembler was not decoding base registers properly. * (AMD64 only) Cropping was incorrect or outright broken with some video filters. [regressions fixed] * "Split" audio filter was broken. * Frameserver exposed too much audio when range to be served was set through range selection, rather than the tail being deleted.