We made a lot of experiments using several techniques and encoding strategies so that we can end up with an archive of videos, that are a benefit to the Typo3 Users. So the following facts had influence on out decision:
Keep it simple. There are many ways to do a video. This is all a lot of work. So keep it as simple as possible is a good approach, to get the work done. Windows Media Encoder offers the possibility to make the production of these videos very simple.
File size: The final Video Files have to be as small as possible. Modem Users (56 kBit/s) should be able to download the videos, without waiting for hours.
Image Quality: Screen Capture Videos need to have a sharp and lossless quality. The compression algorithm has to be optimized for this.
Audio Quality: Since there is only voice and no music or other effects, the audio encoder has to be optimized for voice encodings.
Multi language: At the time of writing this document, we have English, German and Danish versions. It would be great, if we could mux all these language streams into one video container, so that the you can switch the language immediately, while watching the film, just like you know it from movie DVDs
Price: The software to produce should not be expensive. Luckily there are great open source and free products available, so it's possible to produce these videos, without investing a giant sum of money
Time Compression: The viewer should have the possibility to change the playback speed, if the videos is too fast or to slow for him. Modern Players offer this feature, without disorting the sound quality and producing 'Micky Mouse' or 'Lord of The Rings' styled voices. The pitch shouldn't change.
Compatibility: There are 2 kinds of compatibility. First, the software used, to produce the videos. On which platforms (MS Windows , Mac OS X, Linux, BE-OS) is it available? Second, on which platforms is the final product playable? We will see, that this is not so easy to bring into uni-sono with the points above.
The WMV Screen Codec wins the competition. Its only drawback is point 8, which we compensate, offering additional OGM files. We hope, the open source community will bring out more codecs, which are suiteable for screen capturings. At the time of writing this document, there is no codec available (even commercial ones), that can make are able to compete with the Microsofts product.
There are other containers like Matroska, www.matroska.org – also worth a look is the freely available Helix DNA Producer, found under https://www.helixcommunity.org. The latter is the open source variant of Real Media's Encoder, Player and Server!
Speex (www.speex.org) – This is a new open source voice audio codec. Unfortunaltely its not possible to mux it inot an ogm or mkv container. There is only the ACM version, which only works, using Windows. But I guess this will change..
Streaming: There are possibilities to stream the videos. For WMV the commercial and expensive Media Server is necessary. There are open source projects, like http://www.videolan.org/, the Helix Server https://www.helixcommunity.org and Apples Darwin Streaming Server, www.apple.com.
Finally I'll describe the experiences I have had recording all these videos.
The final conclusion was that it was important to create as good source material as possible so as little editing should be done as possible. This really pays of. So videos should be recorded with video and real time speak with no recording of mouse clicks. That makes the whole process fastest.
I also tried to record videos without speak and intended to do the speak later according to a manuscript I would write. This should produce high quality videos which were not boring and long, but it did the exact opposite (I think). But it turned out that the videos were record all too fast for my script as I wrote it! In other words it became very hard to make the speak match the actual video - at least it would require too much preparation to edit the manuscript to fit! Further creating the manuscript took 6 times longer than to play the video. In other words 10 minutes of video could be texted in 1 hour. That was a LOT of time.
Recording the manuscript took time again. And the result became very bad because having the eyes on both the screen and paper didn't work out - I learned that the best is to comment the video while you capture it! Thus audio and video matches.
So don't make scripts, just capture and comment. If you want shorter and better videos better make a test first so you know what you want to say - that will make it more compressed and effective.
Recording mouse clicks is something that requires a mixer so you can lead the microphone into one channel and the line out from the computer (mouse clicks) into the other channel.
However when audio-dubbings from other languages are to be merged into the video track the most convenient thing is to just use VirtualDub for that - which means the click sounds are not preserved. But it turns out (in my opinion) that the click sounds are not that important! So eventually I just discarded them!
I also had the idea at one point that it would make sense to add the click sounds afterward manually by dragging small audio-clips onto the time line in Premiere. Not very clever...
