AVISynth is a head-scratcher at first, especially compared to VirtualDub, which is pretty straightforward. Why bother? Because it is about 1000% more powerful and it is a LOT faster (partly because it works in YUV colorspace, but also because of how it works).

The thing that makes it difficult to understand is that once you install it, there is nothing to double-click on to "open" it. There is no user interface, no real application at all. Instead, you do everything through scripts, which are usually fairly short sequences of commands. I usually write mine in Notepad. You save the script with the extension "AVS" and then you open that AVS file in a program that "understands" the extension that AVISynth added to Windows when you installed AVISynth. For instance, you can simply drag the AVS script file onto Windows Media player, and it will play the results of your script.

However, the usual program to read the AVS files while you are debugging your script is VirtualDub. When I am debugging (and I am working on a very difficult script right now to help me with a totally unique way of transferring 16mm film to video) I keep my script open in Notepad, press Ctrl-S to save that script, switch over to VirtualDub, and press Alt-F then 1 to re-open the last file (which is the script I'm working on). Thus, I can almost instantaneously see the results of any change I've made.

Once I have the script working, I then want to either use the results of that, or I want to encode, either to DV AVI or to MPEG2. For encoding to DV AVI, I use the MainConcept DV encoder and encode directly from VirtualDub to the DV AVI file. If I am being hyper-critical, I'll open the AVS script directly in Vegas, using the Wax plugin (which I have described in other threads) and immediately encode using the world-class encoders built into Vegas. For MPEG2 encoding, I use VirtualDub, in the Direct mode (which bypasses any internal processing) and then frameserve out of VirtualDub into my MainConcept standalong MPEG2 encoder.

There are dozens of other workflows. For instance, one of my favorite ways to use the AVISynth scripts with Vegas is to first perfect the script by reading an AVI file directly, and then viewing the results in VirtualDub, as described above. When I have that perfected, I then change the AVISource statement (which is the one that specifies the input AVI file) to instead point to "VegasFrameserve.AVI". I then fire up Vegas, import all the AVI files, and whatever else I want, and create a project, complete with titles, and the whole schmear. I then click on Render As, and choose Satish's Debugmode Frameserver. I specify VegasFrameserve.AVI as the signpost file. I then open the script in VirtualDub. What I see in VirtualDub are the exact frames that Vegas would render, but without any compression because, by the nature of a frameserver, each frame is sent to the script and then to VirtualDub in a completely uncompressed form. I then use the Direct mode in VirtualDub, use the VirtualDub frameserver, and send the result to the MainConcept MPEG-2 standalone encoder (I'd open the AVS file directly in the MainConcept encoder, but the version I have doesn't read AVS files. If your encoder reads AVS files, you can completely eliminate the VirtualDub "middleman" during the final render stage).