Storyboard for a YouTube Video: How to Make One Scene by Scene
A storyboard for a YouTube video is the approved script divided into scenes, with four pieces of information per scene: the text that is narrated, how long it lasts, what is seen and how the image moves. In a narrated video you do not need to draw panels: a table is enough, filled in scene by scene before producing anything.
What a storyboard is and what it is for on YouTube
In film and advertising, a storyboard is a sequence of drawings showing each shot before filming. In a YouTube video narrated over images it serves the same purpose with less drawing and more precision: it links each sentence of the script to what will be seen while it plays.
It is useful for four things that save work later:
- Calculating the length before recording the voice, by adding up that of each scene.
- Planning the assets: how many images are needed, which are yours and which have to be generated or sourced.
- Sharing the work: whoever records the voice, whoever prepares the images and whoever edits all work from the same table.
- Redoing only what fails: if a scene does not work, you change that scene and the rest stays as it was.
The storyboard starts where the script ends. If you do not have it finalised yet, first read how to write a YouTube script and check every fact: splitting into scenes a text that is still going to change is wasted work.
The columns of a storyboard
| Column | What to note | What it is for |
|---|---|---|
| Number | The order of the scene | Talking about “scene 12” without ambiguity |
| Section | The script block it belongs to | Grouping scenes and generating the chapters |
| Narration | The exact text that is spoken, copied from the script | Recording the voice and generating the subtitles |
| Duration | Estimated seconds, and then the real ones | Controlling the pace and the total length |
| Visual | What is seen: an image, photo or clip of your own, or a caption with text | Preparing or generating the asset |
| Movement | How the image moves: zoom in, pan or static | Bringing a still image to life |
| Provenance | Where the asset comes from and under what licence | Being able to answer if someone asks |
The column that gets neglected most is the last one, and it is the one that prevents the most problems. The one most argued over is the visual: describe the image concretely (framing, light, atmosphere, period) rather than with a single word.
How to do it scene by scene
- Start from the approved script and do not rewrite it. The narration of each scene is copied as it is. If, when splitting, you discover a sentence that is not needed, the change is made in the script, which is approved again; otherwise the script and the storyboard end up telling different things.
- Mark the sections with a title. The title is not narrated, but it names a chapter. To add chapters by hand, YouTube requires the first timestamp to start at 00:00, at least three timestamps in ascending order and no chapter shorter than 10 seconds, so avoid very short sections.
- Cut by sentences, with one idea per scene. Join short sentences until they form a unit and leave a long sentence on its own. A scene of about twenty words lasts around eight seconds at 150 words per minute, the reference pace we use for documentary narration: enough for the viewer to look at the image without getting bored.
- Decide the visual for each scene. A generated image, a photo or clip of your own, or a caption with a fact or a quote. Keep the style the channel bible sets across all scenes: the same light, the same palette, the same kind of image.
- Choose the movement. A slow zoom in on a still image (the so-called Ken Burns effect), a sideways pan, a zoom or a static shot. Alternate so the video breathes and keep captions static, since they have to be read.
- Note the provenance of each asset. Who took the photo, what tool the image was generated with, what licence the clip has.
- Read it aloud from start to finish. Picture each image while you read its narration: that is how you find repeated scenes, scenes that say nothing and scenes that drag. Add up the durations and compare them with the target length.
- Adjust with the real voice. Once the voice-over is recorded, each scene lasts as long as its audio. Correct the table with those timings before editing.
A storyboard example
The first scenes of an invented video, for a channel of narrated mystery stories. The section is called “The Key”, which will be the name of its chapter:
| No. | Narration | Duration | Visual | Movement |
|---|---|---|---|---|
| 1 | “The house at the lighthouse had been closed for years when Marta found the key in her grandfather's drawer.” | 7 s | Image: an old key inside an open wooden drawer, warm lamplight | Slow zoom in on the key |
| 2 | “Nobody in the village talked about that lighthouse. Not even her grandfather, who had looked after it for half his life.” | 7 s | Image: a lighthouse on a cliff at dusk, fog, cold tones | Pan from left to right |
| 3 | “She climbed the spiral staircase counting the steps, as he used to do.” | 4 s | Image: a stone spiral staircase seen from below | Vertical pan upwards |
| 4 | “Up there, the lamp was lit. Someone had been there that very night.” | 5 s | Image: the lamp room in dim light, with the light on | Static |
Note three details. The durations come from counting words and will be corrected with the recorded voice. No image has text inside it: if a caption were needed, it would go as its own scene or over the image in the edit. And scene 4 is static on purpose: after three scenes in motion, stillness reinforces the twist.
Common mistakes
- Rewriting the narration in the storyboard. Any text change goes back to the script; otherwise the voice, the subtitles and what was approved stop matching.
- Scenes that are too long. A still image with a whole paragraph over it is tiring. Better two scenes with two images.
- The same asset in many scenes. If the visual repeats, the viewer stops looking at it.
- Text inside generated images. Add it as a caption in the edit: that way you control it and can correct it without regenerating the image.
- Real people, brands or logos in the descriptions. Describe scenes and settings, not celebrities or other people's products.
- Redoing the whole storyboard for one scene. If you already have voice or images, change only the scene that fails.
With the storyboard finalised comes the voice; if you are torn between recording it and using a synthetic voice, you will find the criteria in AI voice-over or your own voice. And for the end of the process, the checklist for publishing a video on YouTube.
The storyboard in the Estudio de Canales
In the Estudio de Canales, from AVANZAIA TECH SL, the storyboard is a stage with its own rules:
- It comes from the approved script. Until you approve the latest version of the script, there is no storyboard.
- It is split without AI. The cutting into scenes is automatic and does not use AI: the narration of the scenes is, word for word, the approved script. Section titles (“## …”) are not narrated and name the chapters, and within each section sentences are grouped up to about 22 words per scene.
- The AI only suggests the visual. For each scene it suggests the type, the description and the movement (Ken Burns, pan, zoom or static), and its instructions ask it to make the description follow the audiovisual style of the bible, with no brands, no recognisable real people or celebrities and no logos. Images are generated in landscape 16:9 format and the model is asked not to put in text; captions are composed in the edit. You can also use your own assets, with their licence and provenance noted.
- Each scene is independent. It is edited on its own and, if what is narrated or what is seen changes, it goes back to pending. Regenerating scene 17 does not touch the others. You can insert, move or delete scenes; a scene whose voice or image has already been generated is changed rather than deleted.
- The whole storyboard can only be redone at the start, while no scene has a voice or image yet; after that, scene by scene, so as not to throw away work or money.
- Real duration rules. The duration estimated from words is replaced by that of the audio when the voice arrives, and each scene shows an estimated budget, which is not an invoice.
- Review before editing. Each scene goes through its checks (narration, current audio, duration, visual, licence and provenance, no failures and with its cost recorded) and a person approves it.
The studio does not generate video clips with AI: it works best with narrated videos built on images. YouTube is a trademark of Google LLC, and this studio is not sponsored or endorsed by YouTube.
Sources consulted
Checked on 9 October 2026. We link to the original source so that you can consult the current text.
- Capítulos en vídeos. YouTube (Google LLC) · Centro de Ayuda.
Frequently asked questions
What is the difference between a script and a storyboard?
The script is the text that is narrated, organised in sections. The storyboard divides that text into scenes and adds to each one its duration, what is seen and how the image moves. You approve the script first and then make the storyboard.
How many scenes does a ten-minute video have?
It depends on the pace. At 150 words per minute, ten minutes is about 1,500 words of narration; with scenes of around twenty words that comes to roughly 75 scenes.
Do you need to know how to draw to make a storyboard?
No. In a video narrated over images, a table with the narration, the duration, a concrete description of the image and the movement of each scene is enough.
What tool do you make a storyboard with?
A spreadsheet with one row per scene will do. In the Estudio de Canales the storyboard comes from the approved script without changing a word, and each scene is redone separately.