Finished video
An expense manual chapter as a 75-second training video
75 seconds that take one chapter of a fictional company's expense manual through the goal, what to prepare, four steps, common mistakes and a recap, with on-screen subtitles
Narration and on-screen text are in Japanese. The sections below describe the video in English.
How it was made: The phone screens, receipt, flow diagram and checklist were all drawn in code. No generative AI was used for the screens or diagrams. There is no narration, so subtitles carry the message; the background music was made with a music generation model (Stable Audio 3) running on our own computer
A video like this$1,000 (excl. tax) · As soon as the same day75 seconds, landscape (from one manual chapter) | Pricing
Overview
We reordered one chapter of a manual into the order in which viewers actually do the task.
- Subject
- Chapter 2, “Photograph a receipt and submit it”, of a fictional company's five-chapter expense manual
- Flow
- Chapter title, the goal of the chapter, what to prepare, four steps (open the app, take the photo, check the details, submit), photos that tend to be sent back, a recap and the next chapter
- Subtitles
- Each scene has one line of subtitles, with the step number (STEP 1–4) and progress on the left, so the content comes across even where sound cannot be played (on-screen text is in Japanese)
- Why one chapter per video
- With one video per chapter, viewers can rewatch only the chapters they need, and when a procedure changes, only that chapter has to be remade
Materials used and the checks on record
- Supplied materials
- None. The company, the app, the shop and the amounts are fictional.
- Where the materials came from
- The phone screens, receipt, flow diagram and checklist were all drawn in code. No photos or outside image materials were used. The background music was made with a music generation model running on our own computer, and the sound effects are free tracks we already had.
- Checks on record
- On the finished video, the length is 75 seconds, the frozen time is 12.8% (longest 1.1 seconds) and the loudness is −16 LUFS. We looked at check stills to confirm that no text is cut off, overlapping or overflowing.
- What we did
- In the “what to prepare” scene the receipt text wrapped one character per line, so we stopped the heading font size from applying inside the receipt. In the photo scene the guide frame was larger than the receipt and did not match “line up the four corners”, so we sized the frame to the receipt. The goal sentence broke in the middle, so we widened the card.
Talk to us about your video
You don’t need a finished plan to get in touch.
Ask about a video like thisSee pricing and the production processVideos by use and industryCompare with other options

