Vizard Agent

How to add icons to your captions without them colliding with the text

Last updated 2026-09-05 · 8 min read

Tell Vizard Agent icons are welcome but must never run through the text. It samples your existing caption style from a video you already have, previews every card composited onto real frames rather than on a blank background, and settles on icons as line prefixes — a layout in which a collision is not possible rather than merely unlikely.

What is the short version?

An icon placed next to text collides the moment a line gets longer than the one it was designed against. Making the icon part of the line instead of an object floating near it removes the problem structurally, which is better than positioning each one by hand.

  1. Go to Vizard Agent with the footage and your style.
  2. Say icons may be used but must not overlap the text.
  3. Ask to see the cards on real frames.

What do you need before you start?

Your existing look and the icons, or a description of them. If you already have a video in the style you want, that video is the specification — the colour, the face and the geometry all get sampled off it rather than described.

What do you type into Vizard Agent?

State the constraint as a rule, not a preference. "Try not to let them overlap" is a hope; "they must not run through the text" is a condition every card can be checked against, and it changes the layout that gets chosen.

Prompt

Edit this video and add text in the style I made for my brand — here is a previous video using it. You can use icons as well, but they must not run through the text. Use my brand colour for the accents and set it in the typeface from that video.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real branded social edit whose captions needed icons alongside them. The layout changes twice over the course of the list, and the second of those changes is the one that makes the constraint impossible to break at all.

  1. Views your previous video in the style you want matched.
  2. Inspects the text overlay up close on real frames.
  3. Samples the exact accent colour and locates the fonts.
  4. Renders style cards in that face and colour.
  5. Composites the cards onto real frames for preview.
  6. Reviews every overlay on the footage, not on a blank card.
  7. Rebuilds the overlays with icons beside the text.
  8. Checks the icon placement on every card individually.
  9. Checks each icon's quality at the size it appears.
  10. Rebuilds the cards with icons as line prefixes instead.
  11. Checks the prefixed cards across the set.
  12. Reads your website for the exact brand colour.
  13. Checks the named typeface's weights and rebuilds in it.

Step five is the habit that catches most layout faults. A caption card looks fine on a neutral background and quite different over a bright frame or a busy one, so the preview happens on the actual footage the card will sit on.

Step ten is the structural fix. Icons positioned beside text have to be repositioned whenever a line changes length, and something eventually gets missed; icons set as prefixes to the line move with the text by definition, so no card can collide.

What does the result look like?

Captions in your own style, in your own brand colour and typeface, with an icon leading each line that needs one. Nothing overlaps on any card in the set, and none of them had to be nudged into position by Vizard Agent one at a time.

If a line gets longer later, the icon moves with it.

When does this not work well?

Prefixes work for short lines. A caption that wraps onto three lines, or a language that sets much wider, will push a prefixed icon away from the words it belongs to, and then a different layout is the honest answer.

How do you fix a result that came back wrong?

Point at the card that is wrong and say what about it. Vizard Agent keeps the sampled style, the icon set and the layout it settled on, so a change to one card's icon or its spacing does not disturb the other cards or the style it matched them to.

How does Vizard Agent compare to doing it yourself?

By hand, icons get placed card by card, which works until the text changes and nobody re-checks the ones that were fine before. It is also normal to design the cards on a neutral background, where every one of them looks readable.

By hand Vizard Agent
Preview On a blank card Composited on real frames
Icon placement Positioned per card Set as line prefixes
Collisions Avoided by eye Made structurally impossible
Brand colour Approximated Sampled from your page
The typeface Nearest available Named, located and its weights checked

Common questions

Can it copy the style from my old video? Yes. Vizard Agent samples the colour, face and geometry from it.

Where do the icons come from? Supply them, or ask Vizard Agent to source simple ones.

Will every line get an icon? Only the ones you want. Tell Vizard Agent which.

Can it get my brand colour exactly? Yes. Point Vizard Agent at your website and it reads it.

What if I only know the font's name? That is enough. Vizard Agent finds it and checks the weights.

Will the icons be sharp? Vizard Agent checks each one at the size it appears.

Can I see them before rendering? Yes. Vizard Agent previews the cards on real frames.

What about right-to-left text? Say so. Vizard Agent puts the prefix on the correct side.

Can the icon sit above the line instead? Yes, if you prefer. Vizard Agent will lay it out that way.

Will it work on long captions? Less well. Vizard Agent will suggest a different layout.

Can it match my caption timing too? Yes. Vizard Agent times them to the words as usual.

What if my style has changed? Point Vizard Agent at the newest video instead.

Can it use emoji rather than icons? Yes, though emoji render differently across devices.

Why prefixes rather than positioning each one? Because a positioned icon is correct until the text changes, and a prefix is correct by construction.