How to make a timelapse of a repair job for TikTok

Last updated 2026-08-22 · 6 min read

What is the short version?

A repair recorded end to end is convincing and unwatchable. The timelapse that works keeps every stage visible, labels each one so a stranger can follow it, and runs at a pace that makes the work look skilled rather than slow.

1. Go to Vizard Agent and upload the recording of the job.
2. Say the length and that you want a timelapse of the process.
3. Say the language for the narration and the captions.

What do you need before you start?

The footage and the language. Vizard Agent works out the stages of the repair by looking at the footage, so you do not need to log them, and naming the language up front matters because it determines both the voice and whether the fonts can render the text.

• The recording. The whole job, however long.
• The length. Forty-five seconds is right for a feed.
• The language. For the narration and the on-screen text.
• Stage names. Or let Vizard Agent name them from what it sees.
• Your branding. A logo for the closing badge.

What do you type into Vizard Agent?

Say it is a timelapse of a process. Vizard Agent treats that as an instruction to preserve every stage rather than to pick the best-looking moments, which is the difference between a repair video and a montage of a workshop.

Prompt

Variants worth knowing:

• Stage badges. Numbered labels make an unfamiliar process followable.
• Karaoke captions. Word-by-word highlighting suits this format well.
• No narration. Music, badges and captions alone also work.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real repair timelapse. Two things in it are worth copying: the two-second frame sampling that maps out the whole process, and the font test that happened before a single caption was built.

1. Probes the recording and analyses its scenes.
2. Reviews a grid of frames and checks whether there is any voice on the audio.
3. Searches for dynamic music, downloads candidates and detects their beat grids.
4. Extracts a frame every two seconds and builds contact sheets covering every stage of the repair.
5. Reviews the process in two halves to understand what happens when.
6. Searches for a voice in the target language and generates the narration.
7. Analyses the narration timings and checks the timeline arithmetic against the target length.
8. Tests whether the installed fonts can render the language's characters and checks the result on a frame.
9. Generates animated captions and numbered stage badges, fixes the badge alignment after reviewing them, measures the voice and music in LUFS and renders the final cut.

Step 8 is the one that saves a wasted render. A font that silently drops accented or Cyrillic characters produces captions full of empty boxes, and the only way to know is to render a test frame and look at it.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 45.0 seconds, AAC audio. Vertical, exactly forty-five seconds, the whole repair from start to finish with narration in the requested language, animated captions and numbered stage badges.

Hitting forty-five seconds exactly comes from checking the timeline arithmetic before rendering. Vizard Agent works out how much time each stage gets against the narration length rather than compressing whatever is left at the end.

When does this not work well?

A timelapse compresses time, and that compression is its whole risk as a format. Vizard Agent keeps the stages honest and the labels legible, and the format itself naturally makes skilled work look effortless in a way that can undersell it.

• It makes long jobs look quick. A three-hour repair in forty-five seconds sets an expectation.
• Fast motion blurs detail. The fine work is exactly what gets lost when speeded up.
• A fixed camera limits it. Nothing outside the frame ever appears.
• Fonts fail silently. Vizard Agent tests them, and unusual scripts remain the risk area.
• It is not an instructional video. A timelapse shows that the work happened, not how to do it.

How do you fix a result that came back wrong?

Name the stage. Vizard Agent keeps the two-second frame map, the narration timings, the badges and the loudness measurements, so giving a step more time or renaming a badge is a re-render rather than another pass through the footage.

• "Show the welding for longer." Re-timed; the other stages compress to compensate.
• "That badge is mislabelled." Renamed and re-rendered.
• "The music is over the voice." Re-mixed against the measured levels.

How does Vizard Agent compare to doing it yourself?

By hand this is speeding up an hour of footage at one uniform rate, adding text that lands on the wrong stage of the job, and then discovering that half your captions are rendering as empty rectangles because of the font.

Common questions

Does this work for other trades? Yes. The method is the same for any process filmed end to end — a bodyshop respray, a bike rebuild, a bathroom fit-out. Vizard Agent reads the stages off the footage rather than needing to know the trade in advance.

Can it narrate in my language? Yes. Vizard Agent searches for a voice in the language you name, then checks the narration's length against the target duration before building the timeline around it.

Will the captions render correctly? Vizard Agent tests the fonts against your language before building any captions.

Does it label the stages? Yes, as numbered badges, which is what makes an unfamiliar process followable.

How long should a repair timelapse be? Forty-five to sixty seconds for a feed. Long enough for the stages to read, short enough that nobody scrolls past the middle of the job.

Will it keep every stage? Yes. Asking for a timelapse of the process tells Vizard Agent to preserve the whole sequence rather than pick the best-looking moments out of it.

Can it work without narration? Yes. Music, badges and captions carry it perfectly well.

Does it hit an exact length? Yes. Vizard Agent checks the timeline arithmetic against the target before rendering anything.

Does it pick the music? Yes. Vizard Agent searches candidates and detects each track's beat grid so the stage changes can land on the beat.

Can it add my logo? Yes, as a closing badge on the final card. Vizard Agent reviews the badge alignment on frame before rendering.

How does it know the stages? It samples a frame every two seconds and reads the process off the contact sheets rather than asking you to log it.