How to make a vertical motion-graphics video for a security service

Last updated 2026-08-24 · 7 min read

What is the short version?

A security service has nothing to film. There is no product to hold and no process to watch, so the whole video is motion graphics explaining an attack and how it gets stopped, which puts every word on screen under scrutiny.

1. Go to Vizard Agent and say what the service protects and against what.
2. Give it your website for the facts and the brand colours.
3. Say the length and who the audience is.

What do you need before you start?

A clear statement of the threat and the defence. Vizard Agent researches your product and pulls the branding itself, so the site does most of the work. What has to come from you is the wording you are actually cleared to use, since in this category the difference between a strong claim and an indefensible one is a single adverb.

• The threat. What actually happens to victims.
• The defence. What your service does about it.
• Your website. Vizard Agent researches the facts and pulls the logo and colours.
• The audience. Executives, founders, high-net-worth individuals.
• The length. Thirty-five to forty-five seconds.

What do you type into Vizard Agent?

Name the audience as precisely as the product. Vizard Agent uses the audience to set the register of the whole piece, and a security video aimed at executives is paced and worded very differently from one aimed at general consumers.

Prompt

Variants worth knowing:

• A named attack chain. Showing the steps is more convincing than naming the risk.
• Data overlays. Stamps, panels and callouts suit this register.
• A brand reveal at the end. Timed to an accent in the music.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real security explainer. The long section in the middle is unusual and worth reading closely: Vizard Agent went back into its own generated scenes, brightened the dim captions to read them, and found wording it decided should not ship.

1. Researches the company's actual product facts and pulls the logo and brand colours from the site.
2. Casts and generates the first narration line, then measures the narrator's pace.
3. Generates the narration, the music and the sound effects, then renders the key scenes.
4. Reviews the rendered frames, retries the scene that failed and reads the error log to find out why.
5. Zooms into the fine print at the bottom of each scene and crops the stamps to read the small text.
6. Brightens the dim captions to read them properly and inspects the edge text of every scene.
7. Finds two absolute-claim captions in its own generated output and removes them, verifying they are gone cleanly.
8. Inpaints the caption areas to match the background, redoing the patch inside the video's own colour space when the first attempt did not blend.
9. Assembles the picture, mixes the narration against the music, re-times the brand reveal to the musical accent, re-encodes at a social-friendly bitrate and checks the dark gradients for banding.

Steps 6 and 7 are the ones worth stealing. A generator will write plausible microcopy into a scene, and in a regulated category that microcopy can promise something the company cannot. Reading your own output at full brightness is how you find it.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 43.87 seconds, AAC audio. Vertical, forty-four seconds, entirely motion graphics with no filmed footage, narrated, in the company's own brand colours, with the closing brand reveal timed to the music.

Forty-four seconds is what an attack chain needs. Vizard Agent gives each step of the attack its own beat, because a security video that compresses the threat into five seconds leaves the viewer with a feeling rather than an understanding.

When does this not work well?

Security marketing is a category where over-claiming is both very easy to do and unusually costly when it happens. Vizard Agent researches the real facts and reads its own generated output looking for stray claims, and none of that replaces a compliance review on your side.

• Absolute claims are a liability. "Unhackable" and "guaranteed" are the words to watch, including ones a generator invents.
• Generated microcopy needs reading. Vizard Agent found two claims in its own scenes here; check the finished file yourself too.
• Research is not verification. Vizard Agent reads your public pages; it does not audit the product.
• Dark gradients band on compression. Vizard Agent checks for it, and social re-encoding can still introduce it.
• No footage means no proof. Motion graphics explain a claim; they do not demonstrate it.

How do you fix a result that came back wrong?

Point at the scene or the wording. Vizard Agent keeps the research, the brand colours, every rendered scene, the inpainted patches and the stem measurements, so replacing a caption or re-timing a beat is a targeted re-render rather than a rebuild.

• "That claim is too strong." Removed and inpainted out, as this run did twice.
• "The stamp caption is unreadable." Re-rendered with locked microcopy at a legible size.
• "The brand reveal lands late." Re-timed against the musical accent.

How does Vizard Agent compare to doing it yourself?

By hand this means commissioning motion graphics, waiting on a studio, and reviewing them at whatever size the review link renders at. The small print in the corner of scene six is exactly what nobody checks, and in a regulated category that is the line most likely to cause a problem.

Common questions

Do I need any footage? None at all. This format is entirely motion graphics, which is why it suits an invisible service.

Where do the facts come from? Vizard Agent researches your public pages rather than inventing product detail.

Will it match our brand colours? Yes, pulled from your site rather than approximated.

How long should a security explainer be? Thirty-five to forty-five seconds. Long enough to show the attack chain step by step.

Why check the video's own fine print? Because a generator writes plausible-looking microcopy into a scene, and plausible is not the same as approved. Vizard Agent brightened and read every caption here and found two absolute claims that had no business shipping.

Can it remove text from a finished scene? Yes, by inpainting the area to match the background in the video's own colour space.

Does it check for compression artefacts? Yes. Dark gradients band easily, so Vizard Agent inspects them after the social re-encode.

Does Vizard Agent check the finished cut? Yes. It reviews the render frame by frame and has the motion and sync watched back.