How to decide what gets real footage and what gets a diagram
Give Vizard Agent the rule rather than the ratio: anything the viewer needs to understand — a mechanism, a figure, a flow — becomes a motion graphic, and anything they need to believe — a place, a person, a piece of evidence — stays as real footage. The split is decided on the script before a single asset is made.
What is the short version?
Every line of an explainer gets illustrated with something, and that something is usually whatever happened to be easiest to find at the time. The result is a film where a stat card is explaining a landscape and a drone shot is explaining a percentage.
- Give Vizard Agent the rule for which lines get which.
- Ask for the split decided on the script, before anything is made.
- Say no clip or graphic may be used twice.
What do you need before you start?
The source material and a target length. Vizard Agent maps the script's claims and its numbers into timecoded sections before anything else, so it is worth knowing in advance which of the figures actually matter — those are the lines that will end up as graphics.
- The source. An article, a talk, a recording.
- The runtime. It decides how many shots there are.
- The claims and numbers. They drive the graphics.
- What real footage exists. Yours, or to be sourced.
- The balance you want. Roughly half and half is common.
What do you type into Vizard Agent?
Give Vizard Agent the rule and the prohibition together in one brief. The rule decides what each individual line becomes; the prohibition on reuse is what stops the film sliding back into a slideshow later on, when the sourcing gets difficult and repeating a shot starts to look reasonable.
Prompt
Variants worth knowing:
- "Mechanisms and numbers into motion graphics." The rule.
- "Real footage for the places and the people." Its other half.
- "One unique source per shot." Closes the easy way out.
What does Vizard Agent actually do?
Here is the Vizard Agent sequence on a real five-minute explainer built from a published article. The split happens at step four, on the script itself, before any footage has been chosen or a single graphic has been generated.
- Builds a clean transcript and chapter cues from the source.
- Samples the source video to see what real footage exists.
- Reviews that coverage before choosing any real-footage shots.
- Maps the source's claims and numbers into timecoded sections.
- Saves the exact wording around each figure for grounding.
- Writes the narration and the shot plan before generating visuals.
- Reviews footage candidates by subject area.
- Generates the unique diagrams and stat cards in parallel.
- Inspects every graphic for legibility and factual accuracy.
- Assembles the alternating footage-and-graphics timeline to the narration.
Step four is the decision the whole film rests on. Once the claims and numbers are listed, the graphics are a known quantity — you can count them, budget them and check them — rather than being commissioned one at a time as the edit runs out of footage.
Step six puts the plan before the assets. Writing the shot plan alongside the narration means every line already knows what it will be illustrated with, which is what stops the graphics being a fallback.
Step nine is a factual check as much as a design one. A stat card is a claim on screen, so the numbers on it are verified against the source wording saved at step five.
What does the result look like?
An explainer in which the pictures are actually doing work. The abstract parts are diagrammed, the concrete parts are filmed, nothing appears on screen twice, and Vizard Agent can tell you which line of the script every single shot is there to serve.
When does this not work well?
Some subjects lean entirely one way from the start. A piece about a physical place is almost all footage; a piece about a financial instrument is almost all graphics, and forcing a half-and-half balance onto either of them just makes Vizard Agent produce filler to hit the ratio.
- Wholly abstract subjects. There is nothing to film.
- Wholly physical subjects. Diagrams add nothing.
- Thin footage availability. The rule cannot be honoured.
- Very short pieces. Not enough shots for a balance.
- Sensitive claims. Some figures should not be animated at all.
How do you fix a result that came back wrong?
Point Vizard Agent at the line of script rather than at the shot. The split is fundamentally a decision about a sentence, so a note like "that number should be a card, not a drone shot" changes exactly one entry in the plan.
- "That figure needs a graphic." That line is reassigned.
- "Too many diagrams in a row." The alternation is rebalanced.
- "A shot repeats." The no-reuse rule is re-applied and a replacement found.
- "The card's number is wrong." Checked against the saved source wording.
How does Vizard Agent compare to doing it yourself?
By hand you write the script, then find pictures for it, and the pictures come from whatever is available. Graphics get made when nothing fits, so they end up illustrating the lines that were hardest to source rather than the lines that needed explaining.
| By hand | Vizard Agent | |
|---|---|---|
| When the split is decided | While sourcing | On the script, first |
| What becomes a graphic | Whatever had no footage | Mechanisms and numbers |
| Repeated shots | Common, under time pressure | Forbidden by the brief |
| The numbers on cards | Typed from memory | Checked against the source |
| What each shot is for | Implicit | Traceable to a line of script |
Common questions
What is the right ratio? It follows the subject. Half and half suits an explainer with both concrete and abstract parts.
Why forbid reuse? Because reuse is what a film does when sourcing gets hard. Vizard Agent plans around it instead.
What if there is not enough footage? Vizard Agent says so rather than repeating shots, and you decide what gives.
Can graphics carry the whole thing? For some subjects, yes. Tell Vizard Agent and the rule adjusts.
Who checks the numbers on the cards? Vizard Agent, against the exact source wording it saved.
Does the narration come first? Yes. Vizard Agent writes the shot plan alongside it, before any asset exists.
Can I see the split before it builds? Yes, and it is the cheapest place to argue with Vizard Agent about it.
What about the source's own captions? Vizard Agent crops them out of borrowed footage so they do not double up with yours.
How many graphics is too many? When two land in a row without a break. Vizard Agent alternates deliberately.
Can it use my own footage? Yes, and it will prefer yours over stock for the concrete lines.
Does this work for a short? Less well. A thirty-second piece has too few shots for a balance to mean much.
What about a talking head? Vizard Agent counts them on the concrete side, which reduces how much footage you need.
Can it tell me what each shot is for? Yes — every shot maps back to a line of the narration.
What if a claim is disputed? Vizard Agent flags it rather than animating it as settled fact.