A good research video can greatly promote your work, elaborate your idea, and complement your paper. It is usually an essential piece of a paper submission in the HCI/VIS research community. Much of the below content is extracted from Vogel's blog.

<aside> 📖 *Research Videos: Planning, Shooting, and Editing by D. Vogel*

</aside>

Planning

Story and structure

One of the main goals of the video is to encourage people to read your paper. So a research video is not a self-standing summary of your paper. The video should present a mini-story. Make sure the length of the video is within 5 minutes. Usually, the following elements can be included in a research video:

• Motivation and background: This part must be very brief. Sometimes just one or two sentences in the voice over without any supporting video, or with a movie trailer like show off of the system.
• System and technique: This is the meat of the video. You can start with an overview of the system. Try to demonstrate the system functionalities in a story or a scenario, and avoid just listing a set of features. This is usually in complement with the scenario or case study sections in your paper (if any). It is fine that not all the system features are shown in the video.
• Experiments and conclusion: It is important to have a brief summary of how you evaluated the system or technique and what the main takeaways are. However, you do not need to explain the detailed study design (e.g., permutation of tasks) or include any charts or equations.

Scripts

Writing scripts are very important for planning the video. Your goal is to:

• Create a structure: What is the general flow? Is it logical and clear? Does it make sense?
• Write the voice over: Mind that spoken language and writing language are different. Do not directly copy and paste sentences from your paper. However, you may rephrase them.
• Describe each shot: What is going into each shot? What will the "user" do? How the system is going to respond?

Doing

Shooting and recording

Shoot the video and record the audio separately. In this way, it is flexible and easy to post-process the video and the audio to create a smooth research video.

You don’t need to shoot in the order on your script. You can shoot multiple shorter sequences and concatenate them later. Shoot in 1080p HD 30 FPS and you can always scale it down later. Interact with the system slowly and sometimes pause a bit, as you can always speed up or cut the video sequences in the editing phase. For screen-capturing, the default Mac Quicktime player or Windows Game Bar is sufficient.

Record the voice over before your edit. Use a good mic and sound excited! Enunciate your words clearly, Don’t read too fast, or too slow. Remember to pause at key times so you can splice the audio and expand it to match the video. AudaCity is a handy free software for recording audio. With the advancement of AI, some auto-generated voice is not bad (e.g., https://elevenlabs.io/), while a human-voice is always preferred.