Gains:
- Ability to understand storyboard, previz and moodboard concepts and prepare scene and frame drafts with artificial intelligence visual generators
- Ability to describe camera angle, plan scale and light intent in accurate terms to artificial intelligence and produce visual reference
- Ability to recognize that the images produced are a planning tool and are subject to the physical and budgetary constraints and copyright rules of the actual shooting.
The script describes "what" the video is; Storyboard shows "how it will look". Going to a shooting day with words is like building a building without stones: everyone dreams of something different. Storyboard and visual preparation bring together the picture in the team's mind into a single image. Artificial intelligence image generators have radically accelerated this phase: you can now produce, test and modify framing sketches in minutes, where it used to take an illustrator days to work. In this unit, we will see how to use AI as a pre-visualization tool, as well as why the visual produced is a "plan" and not a "finished work".
Terms first. Storyboard is a keyframed/visualized outline of the video; each box represents a plan. Previz is the rough visualization of a scene before shooting. Moodboard is a reference board that collects the visual atmosphere (color, light, texture, style) of the project. Frame / shot scale is how close the camera shows the subject: long shot (shows location), long shot, medium shot, close-up (face), detail shot (eye, hand). Camera angle is the direction of view: eye level, from below (frog), from above (bird's eye view). Using these terms towards AI is the key to getting useful visuals.
Speaking "in camera language" to the visual generator
An AI image generator (a tool that produces images from text) is only as good as the detail of the description you give it. A regular user writes "a woman is drinking coffee"; A video professional describes the scene in camera language. The difference is between an amateur shot and a cinematic one. A good visual prompt includes the following layers: subject (who/what), action (what is he doing), framing and angle (close-up, from below), light (soft window light, golden hour), atmosphere/color (warm tones, foggy), style (photographic, cinematic, drawing), aspect ratio (16:9 horizontal, 9:16 vertical).
Produce storyboard frame.Subject: barista in his 30s, behind wooden counter.Action: smiling slightly as he hands cup to customer.Framing: medium-close-up, eye level.Light: soft morning window light from the side, warm tones.Atmosphere: nostalgic, intimate, with slight film grain.Style: cinematic photography, shallow depth of field (background blurred).Aspect ratio: 9:16 vertical.
This recipe also allows you to establish the visual flow of a scene by sequentially producing different framings of the same scene (overall, middle, close up). Thus, the storyboard turns into a box-by-box cinematic plan.
Moodboard and visual consistency
The hardest part of a project is making all the scenes seem to belong to the same "world". AI is great for creating a moodboard: it quickly generates color palette, lighting character and texture references, providing the team with a common visual language. But AI image generators have a known challenge: consistency. It is difficult to reproduce the same character exactly in two separate frames; face, clothing and location may shift from scene to scene. So use the images produced as a "reference to atmosphere and intent" rather than as a "precise visual reference". People provide real consistency in shooting and editing.
The following table summarizes why AI is suitable and risky at the storyboard stage:
Purpose
Is AI suitable?
Why
Framing/composition experiment
Yes
Fast, cheap, lots of variants
Exploration of light and atmosphere
Yes
Tone and color indicate intent
Explaining the intention to the team
Yes
"I want a feeling like this" communication
character consistency
restricted
It is difficult to reproduce the same face
Finished image / final frame
no
Copyright and authenticity issues
Physical proof of drawability
no
Image does not warrant budget/venue
The image produced is a "plan", not a "work"
Just because a storyboard shot looks great doesn't mean you can actually shoot it. AI can produce a scene on a beach in golden hour light, from a drone perspective, with 200 extras — but your budget, permission and equipment may not allow it. The storyboard's job is not to daydream, but to communicate a pullable plan. That's why I always consider the visuals produced with the question "how do I approximate this with the location, actors and equipment I have?" Filter with the question. The visual is the picture of intent; attraction is reality itself.
The second critical point is copyright and style. AI image generators can be trained on large numbers of copyrighted images and mimic an artist or brand's signature styles. Using a storyboard for internal planning is usually okay; However, presenting the produced image to the customer as the "final design", being too similar to a copyrighted style, or producing the image of a recognizable brand/person creates legal risks. Position the storyboard as an internal planning tool; Final images must come from copyright-free sources.
three mini cases
Case 1 — Fast communication. An advertising director would present three different visual directions to the client. Produced 5 storyboard frames with AI for each direction; The color and light showed his intention clearly. The customer quickly made a decision at the meeting, saying "the second direction is exactly our brand". Direction presentation, which used to take 3 days, was reduced to half a day and communication became clear. Visuals became a common language that accelerated conversation.
Case 2 — Unfilmable storyboard. A team created a dazzling storyboard with AI, set on a cliffside at sunset, and had it approved by the client. On the day of shooting, permission was not granted for that location, the budget did not allow for a drone, and the result did not resemble the storyboard at all; The customer was disappointed. Lesson: getting the storyboard approved without aligning it with budget and permitting realities is making a promise that can't be met.
Case 3 — Risk of style imitation. A designer wrote the name of a famous illustrator into the AI prompt, produced a storyboard "in his style" and used it in a commercial presentation. The images, which were too similar to the artist's style, caught the attention of the client's legal team and work was halted. The right way was to describe the style in neutral visual terms (colour, light, texture) rather than by an artist's name.
Weak prompt / Strong prompt
Weak prompt:
Draw a coffee shop scene.
There is no framing, light, angle, style or proportion. The output becomes random and unusable.
Powerful prompt:
Storyboard frame, 16:9 landscape. General shot: small neighborhood coffee shop, morning, looking through glass display case. Camera: slightly from below, near eye level. Lighting: contrast of cold morning exterior light with warm yellow bulbs inside. Style: cinematic, realistic photography, medium grain. Note: no specific artist or brand name; Produce in neutral visual terms.
Common mistakes
- Not using camera language. If framing, angle, light and proportion are not given, the visual will be amateurish and random.
- Mistaking the produced image as the final work. Storyboard is a plan; The final is not visual.
- Not checking drawability. A storyboard that doesn't align with the reality of budget/location/permission is disappointing.
- Style imitation with artist/brand name. It creates copyright risk; Describe the style in neutral terms.
- Expecting character consistency from AI. Reproducing the same face is unreliable; Provides shooting consistency.
Tip: Have a scene reproduced in three frames (wide-medium-close). This trio both establishes the storyboard flow and speeds up your shot scale decisions on shooting day.
Attention: When presenting the storyboard to the client, add a note that "these are drafts of intent/atmosphere, not the final visual." This way, you manage expectations from the beginning.
In summary
Storyboard and previz are the bridge that translates the picture in the head into the common language of the team. AI image generators build this bridge in minutes: testing frames, exploring light, rendering atmosphere. The key is to speak camera language (framing, angle, light, style, proportion) to the AI. But the image produced is a blueprint, not a finished work: it has poor consistency, does not guarantee drawability, and can mimic copyrighted styles. Use storyboard as an internal planning and communication tool; Ensure truth in shooting, consistency with people, and copyright cleanliness in source selection.
Application task
Choose a scene from your script. Have this scene produced by AI in three frames (general, medium, close); Clearly state the framing, angle, lighting, style and aspect ratio in each prompt and do not use artist/brand names. Then I started each shot with the question "how do I shoot this with the location/actor/equipment I have?" Evaluate it with the question and turn it into a drawable note. Collect all three frames and drawability notes on one page.
checklist
- [ ] Have I specified the framing, angle, light, style and aspect ratio in the visual prompts?
- [ ] Have I avoided style imitation of the artist/brand name?
- [ ] Did I position the produced image as "plan" or "final work"?
- [ ] Have I filtered each shot for shootability against the reality of budget/location/permission?
- [ ] Have I conveyed the "these are drafts of intent" expectation note to the client/team?