A science fair video is not a recording of a presentation. It is a short film that has to do one job: convince a judge, in under three minutes, that you ran a real experiment and understood what happened.
Most entries fail for the same three reasons. The audio is unusable. The script reads like a lab report. And the demonstration never actually shows the result. All three are fixable before you touch a camera.
What judges are actually scoring
Judging rubrics vary, but the communication score almost always breaks into four parts:
- Clarity — can a non-expert follow your question and method?
- Evidence — is the data shown, not just described?
- Ownership — do you understand your own limitations?
- Delivery — can they hear you and see the setup?
Notice that nothing on that list requires expensive equipment. A phone on a tripod with clean audio outscores a cinema camera with wind noise every time.
The 90-second structure that works
Total runtime should be 90 seconds to three minutes. Anything longer gets skimmed. Here is the shape that holds up across age groups:
| Time | Section | What happens on screen |
|---|---|---|
| 0:00–0:10 | Hook | The result, or a question that creates tension |
| 0:10–0:30 | Question and hypothesis | You on camera, one sentence each |
| 0:30–1:15 | Method | Fast cuts of the actual setup being built |
| 1:15–1:50 | Data | Graph or table, held still, narrated |
| 1:50–2:20 | What went wrong | One honest limitation |
| 2:20–2:40 | Conclusion | One sentence, no hedging |
Two structural rules matter more than the timings. Open with a result, not an introduction. “I tested whether music affects plant growth, and the answer surprised me” beats “Hello, my name is…” every time. And show the messy part. A judge who sees a failed trial believes the successful one.
The shot list
Before recording anything, write down the eight shots you need. This prevents the single most common time sink: filming for two hours and discovering you never got a clean shot of the measurement.
- Wide of the setup — the whole table, tripod-height, nothing moving
- Close-up of the measurement — the thermometer, the scale, the ruler
- Your hands doing the step — over-the-shoulder, no face needed
- The subject — the plant, the circuit, the culture dish
- The data on screen — graph held perfectly still for three seconds
- You talking to camera — chest-up, eye level, window in front of you
- The failure — whatever went sideways
- The final result — the cleanest version of the outcome
Shoot each one twice. The second take is almost always better than the first.
Audio is the whole ballgame
A judge will forgive soft focus. They will not forgive audio they cannot understand.
- Record voiceover separately. Do not narrate while filming. Film the visuals, then record the script in a quiet room with the phone held close to your mouth.
- Kill the echo. Closets, cars, and blanket forts all work better than a kitchen or a garage. Soft surfaces absorb reflections.
- Never use camera audio outdoors. Wind destroys it. If you must film outside, record the audio afterward and lay it over.
- Leave half a second of silence at the start and end of every voiceover take. It makes editing much easier.
Writing a script that sounds like a person
Read your script out loud. Any sentence you stumble over is a sentence to rewrite. Three specific fixes:
Cut the passive voice. “The plants were watered every day” becomes “I watered the plants every day.” Ownership reads as competence.
Replace adjectives with numbers. “The plant grew a lot” becomes “the plant grew 4 centimeters in twelve days.” Judges score evidence, not enthusiasm.
Say the limitation out loud. “I only had six plants, so this is a small sample” earns more credit than pretending the sample was sufficient. Judges are trained to spot overclaiming.
Editing: what to cut and what to keep
Rough footage for a two-minute video is usually twenty to forty minutes. The edit is where most of the work happens.
Three things to do in the edit, in this order:
- Cut every pause longer than a beat. Dead air reads as uncertainty.
- Add captions to the numbers. Any figure you speak should also appear on screen. Judges often watch with sound off first.
- Hold the data shot still. Do not pan or zoom across a graph. A static frame that a viewer can actually read is worth more than a dynamic one they cannot.
What not to do: no background music during narration (it competes with speech), no transitions other than cuts (wipes signal amateur), and no title cards longer than two seconds.
Tools that make the edit faster
The editing step is where the time goes, and it is the step most families underestimate. A basic phone editor can do the job, but two things consistently slow people down: captioning numbers accurately, and cleaning up voiceover recorded in a room with echo.
For this specific workflow — trimming a long take down to two minutes, adding burned-in captions, and doing light audio cleanup — a desktop editor with automatic caption generation saves the most time.
Movavi Video Suite is one option in that category. The current Video Suite 2027 bundles three tools that map directly onto the checklist above: an editor with one-click automatic subtitles and voice denoise (the two things that eat the most editing time), a screen recorder for capturing your data slides and graphs, and a converter that outputs whatever format the fair’s submission portal accepts. It is not required to make a good video — a phone and free software will produce the same result with more manual work. The value is time, not quality.
If your fair accepts a poster instead of a video, our science fair project guide covers the experiment-design side, which matters more than the format.
A checklist before you submit
- ☐ Total runtime under three minutes
- ☐ Audio recorded separately, no wind, no echo
- ☐ Every number you say is also on screen
- ☐ Data shot held still for at least three seconds
- ☐ One limitation stated out loud
- ☐ Opening line is the result, not an introduction
- ☐ File format and resolution match the fair’s requirements
- ☐ Export watched start to finish on a different device
That last item catches more problems than any other. A file that plays perfectly on your laptop can fail on a judge’s machine.
Frequently asked questions
Can I use my phone instead of a camera? Yes. A modern phone shoots better video than most point-and-shoot cameras, and the smaller sensor is not a limitation indoors. Spend your effort on audio instead.
How long should the video be? Follow the fair’s limit exactly. If there is no limit, aim for two minutes. Longer videos are not scored higher and are more likely to be skimmed.
Should I appear on camera? If the rules allow it, yes. Judges score ownership, and a face is the fastest way to signal that you did the work. If anonymity is required, use hands and voiceover instead.
What if my experiment failed? Submit it. A well-documented failure with a clear explanation of what you learned scores higher than a vague success. Judges see dozens of successful projects; they remember the one that understood why it went wrong.
Do I need to show my face and my data in the same shot? No. Cut between them. Trying to do both at once usually makes the data unreadable.
<!– disclosure handled by theme –>
