How to illustrate a figure of speech without taking it literally
Tell Vizard Agent which lines are figures of speech. Footage is chosen from what a sentence says, and a metaphor says something it does not mean, so it looks for abstract material that carries the feeling — smoke rather than a burning house — and checks that the shot is not quietly asserting something the script never claimed.
What is the short version?
B-roll is matched to meaning, and a metaphor's literal meaning is not its meaning. Illustrate "drowning in paperwork" literally and you have made a comedy; illustrate it with a slow drift into a pile of documents and you have made the point.
- Go to Vizard Agent with the script.
- Say which lines are figurative rather than descriptive.
- Ask for abstract footage for those.
What do you need before you start?
The script and a pass over your own writing. Vizard Agent can spot obvious figures of speech, but the borderline ones — a phrase that is half idiom and half description — are decided by what you meant, and only you know that.
- The script. With the figurative lines marked.
- What each one means. The feeling, in plain words.
- The register. Serious pieces cannot risk the joke.
- Whether abstract footage suits the piece. Some are literal throughout.
- Any line that IS literal. Say so; it may look figurative.
What do you type into Vizard Agent?
Mark the lines for Vizard Agent and say what each one actually means. "Make it less literal" is not enough on its own; the useful instruction names the specific line and gives the feeling behind it, because the feeling is what a shot has to carry.
Prompt
Variants worth knowing:
- "Metaphors, not descriptions." Names the category.
- "Abstract footage that carries the feeling." The brief.
- "Tell me what you picked." Cheap to reject early.
What does Vizard Agent actually do?
Here is the pattern Vizard Agent follows across real explainers and short films whose scripts leaned on imagery. Notice that the search terms change completely once a line has been understood as figurative, and that change is where the whole difference comes from.
- Reads the narration line by line.
- Checks the context around an ambiguous phrase.
- Establishes whether the line is literal.
- Searches for abstract material where it is not.
- Chooses footage that carries the feeling rather than the noun.
- Times it to the line rather than to the word.
- Checks the shot does not assert anything the script did not.
- Shows you what it picked before assembling.
- Replaces anything that reads as literal.
Step two is worth paying for. Some phrases genuinely could be either, and reading what surrounds them is how you tell a line about a real fire from a line about a bad quarter.
Step five is where the judgement actually sits. Carrying a feeling is not the same as finding a related noun, so Vizard Agent looks for material whose movement and light match the weight of the line rather than for footage that happens to contain the thing the words mention.
Step six is a quieter version of the same problem. Cutting to the shot exactly on the metaphor's key word points at it and makes the literalism obvious, whereas letting the image run under the whole line lets it stay atmospheric.
Step seven is the one that matters most in a factual piece. A literal shot makes a claim — a burning house on screen says a house burned — and a metaphor illustrated literally can turn a figure of speech into an assertion nobody intended to make.
What does the result look like?
A film whose imagery supports the writing instead of arguing with it. Nobody watching notices the decision, which is the sign it was made correctly — a literal shot under a metaphor is the one people remember, for the wrong reason.
Vizard Agent tells you which lines it treated as figurative, so you can disagree with any of them.
When does this not work well?
Abstraction has a cost. Too much of it and the video stops being about anything, and some registers — instructional, technical, news — expect the picture to show the thing being described and read abstract footage as evasive.
- Instructional video. The picture should show the thing.
- News and reporting. Abstraction reads as evasion.
- Too many metaphors. The film becomes a mood reel.
- Very literal scripts. Nothing to interpret.
- Idioms in translation. They may not survive at all.
How do you fix a result that came back wrong?
Say which direction it went wrong in, in those terms. "Too literal" and "too vague" are opposite failures with opposite fixes, and Vizard Agent needs to know which of the two you are looking at before it can move in the right direction.
- "That is a literal house fire." Abstract material instead.
- "Now it means nothing." Closer to the subject.
- "That line was literal." Marked, and illustrated directly.
- "It cuts on the word." The shot runs under the whole line.
How does Vizard Agent compare to doing it yourself?
By hand the b-roll search is done from the script's nouns, because those are what a stock library understands. The result is technically on-topic and occasionally absurd, and it is usually defended on the grounds that it matches the words.
| By hand | Vizard Agent | |
|---|---|---|
| The search term | The noun in the line | What the line means |
| Figures of speech | Illustrated literally | Identified first |
| Ambiguous phrases | Guessed | Read in context |
| Unintended claims | Not considered | Checked for |
| Review | On the finished cut | Picks shown before cutting |
Common questions
Can it tell a metaphor by itself? Often, and Vizard Agent reads the surrounding context when it is unsure.
Should I mark them in the script? It helps Vizard Agent a great deal, especially on the borderline lines.
What does abstract mean here? Texture, light, movement — Vizard Agent will show you options.
Will the video feel vague? Not if it is a few lines. Vizard Agent will warn if it is most of them.
What about idioms? Same treatment, and Vizard Agent will flag ones that translate badly.
Can it generate the shot instead? Yes, and Vizard Agent finds generated abstraction easier to aim than stock.
Does timing matter? Yes. Vizard Agent runs the shot under the line rather than on the word.
What if a literal shot is funny on purpose? Then say so, and Vizard Agent will lean into it.
Does this apply to captions? No. Vizard Agent keeps captions carrying the words exactly as spoken.
Will it change my script? No. Vizard Agent changes what is on screen, not what is said.
How do I know what it chose? Ask Vizard Agent to list the picks before it assembles.
Is abstract footage harder to find? Sometimes, and Vizard Agent will suggest generating it when searching stalls.
What about a mixed line? Vizard Agent reads the context and will ask if it is genuinely ambiguous.
Why does literal illustration fail? Because a picture asserts, and a metaphor was never making that claim.