This short tutorial aims at providing some generic information about producing a demonstration video. This tutorial tries to be as much technology neutral as possible.
Producing a video capture of a demonstration is not as complicated as it may sound at first. However it is important to plan as much detail of the video as possible beforehand - the success and speed by which the video is produced depends on that!
Broadly, video production undergoes a number of phases that are captured in this little tutorial. The beginning of all is the planning of the story of the video, the core message it wants to transport. Setting the scope of the video, and the detailed planning of the narrative is done in the story board phase. Based on that, the technical planning of the video taking needs to happen. Which tools to use, which resources/partners to use, at which time of the day, etc. Typically, this includes a number of dry runs to prepare for taking the scenes. Then the scenes are taken (rather, in our case captured form the screen). One by one, scenes get taken using the resources prepared in the previous step, and the raw material is kept for the video editing phase. Finally, the video is produced by compiling all the scenes in to one smooth story-telling video.
It makes a great difference whether the video is supposed to transport the message of technical interoperability of a federated clouds infrastructure (like ours), how a user community accomplishes its tasks in an easy manner, or both!
So, first, set the scope of the video, the message it needs to transport.
Then, write down how you want the video to transport the message, what you want to show. Organise this iteratively into Episodes and Scenes. An rough example is my CloudscapeV demo script - at the same time it is a bad example, since it is only providing an outline of the video in terms of episodes, not scenes - I had the scenes in my mind and didn't feel the need to write them down (probably a bad idea).
My example provides for 5 episodes - eventually I ended up with 10 episodes that mostly contained only one scene. Bad planning on my side.
You should end up with a consistent and coherent list of episodes and scenes that tell the story.
Technical planning tries to remove any technical obstacle that may delay taking the scenes and producing the video off the raw material.
Plan the resources (particularly, your time!) according to the use cases, contact Cloud providers in the task force and ensure that you can use the resources without fiddling during the scene taking, that every thing is configured to need, or configuration values are known beforehand (if, for example, ease of configuration is something you want to show, then that should be already planned in the story board).
Adjust your desktop to the demo
Remember that the audience rarely is as close to the monitor as you are! They are usually at least 2 metres away fro the display you are using. So plan to adjust your desktop to show LARGE THINGS.
Tip: Chrome and Firefox allow to zoom into the displayed website. The higher the final resolution, the more you need to zoom in. 300% - 400% zoom provided me with satisfactory results.
Tip: If you use the console or terminal, pick a font designed for screens, not paper. Verdana is a good example, Times New Roman is not. Set the font size fairly high. My demo uses Menlo regular 20pt!
Do as many dry runs as required to reassure the participants, resource providers, and yourself that the actual scene taking will run smooth.
Select tools according to your requirements (e.g. screen ratio and resolution, etc.) and get to know them. Common requirements are;
Format and resolution
Plan the final screen format and resolution before you take videos. Common screen ratios are 4:3 (common CRT TV sets, beamers, etc.), and, much more common these days, 16:9 (widescreen TVs, plasma panels, LCD monitors). Plan the same for the scene taking phase - the values need not necessarily the same!
Tip: Take scenes in highest possible resolution, and try to stick to the same screen format. Downsampling to final output resolution is easier and yields much better results!
Tip: If part of your demonstration involves command line hacking, be sure to use even higher capturing resolutions - blurred command line scenes can destroy every impression you want to make!
Video capturing tools should be able to capture in full screen, or only windows or areas of the screen. Quicktime seem appropriate for Mac and Windows - for Linux, I have no recommendation.
Video editing capabilities
The video editing software should support a number of capabilities that make producing a smooth video easier. These vary greatly fro tool to tool, so I cannot give a referral to any tool. However, the following capabilities should be supported:]
- Video speed adjusting
- At times the video's realtime is too slow, e.g. when tying on the command line, or waiting for remote responses. In those situations speeding up the video track by 130% helps keep the timing and flow.
- Transitions help with guiding the viewer from scene to scene, or from episode to episode.
- Title screens
- Title screens allow guiding the user into an episode without getting distracted from what happens on screen.
- Text overlay / comments
- To enrich the video with comments, text overlay or comments functions are necessary to be able to produce a video that can stand alone, without a person narrating it.
- Audio track (optional)
- Pretty much any tool provides audio track capabilities for adding your narrative or music to the video.
Taking / capturing scenes
This is usually the bulk of the work. Work through your scenes, and take them one by one. Keep the raw material as single clips in a single repository.
Tip: Take scenes multiple times in slight variations - you'd be surprised that when editing the video you'd end up using an alternative take than the one you planned to use!
Tip: Take scenes with roomy pauses! Add pauses of a couple of seconds that you can use to clip away when editing, or use for transitions. Even within scenes, plan for pauses for narrative, comments, or even for scene splitting when editing the video.
Producing the final video
The details of this step greatly depend on the actual tool you chose. However, some general advice can be given here:
- Start small. Add a title / intro as a still image displayed for some time. Use the slide template (see below), export it as an image and add the image to the movie.
- Add a first clip of the first episode and scene to the movie, and att a transition.
- Proceed with transitions, title images and comments as required.
- If required crop videos with the "bleeding" edges at the start and end to fit the transitions you surround it with.
- Scenes clips should *always* be surrounded by transitions - unless the following scene has exactly the same display layout as the end of the preceding clip! (e.g. when cutting out a long pause of a non-changing screen)
- Be consistent in the transitions you use. If you use different transitions then keep using the same transition as introductions to episodes. Same applies to inter-scene transitions.
- Provide enough time for title scenes display - people need time to read the text and understand!
Iterative video production
Work towards having a full video assembled in your video editor. Depending on the final output resolution and what you are demonstrating, the final video rendering can be very costly in time as it may take several hours!
So to assess the effect of transitions, video flow, etc, produce low resolution versions of your video to see it in an external tool, or share with others to get feedback. These usually take only very little time to produce, allowing for high-frequency iterations where necessary.
The following values and settings have proven useful for the CloudscapeV demo:
- Title image - 6 seconds display
- Episode titles - 6 seconds display
- Fading transition - 0.5 seconds overlaying adjacent scenes clips (Careful - this takes 0.5 seconds out of each clip, and 1 second out of the total video time)
- Terminal scenes - speed up by 125 - 130% depending on typing speed
- Waiting for results (e.g. VM image instantiation) - I sped these up by 500%(!)
- Comments text overlay - 6 seconds
<Back to FedCloud Outreach section.