Vizard Agent

How to set a title card in two languages at different sizes

Last updated 2026-09-10 · 8 min read

Give Vizard Agent both languages, the typeface for each and the size each line should be. It finds a face that can actually render your script before designing anything, sets the two lines at the sizes you gave rather than scaling one to match the other, and checks the strokes still read on a black card at full size.

What is the short version?

A bilingual card is two typographic decisions, not one. The scripts have different natural weights and different comfortable sizes, so setting them at the same point size makes one of them look wrong — usually the one that matters most.

  1. Go to Vizard Agent with both versions of the text.
  2. Say which language leads and at what size.
  3. Give the typeface for each, if you have a preference.

What do you need before you start?

Both texts and a view on the hierarchy. The rest can be described loosely, but the sizes are worth being precise about — "the Korean clearly larger" is a direction, and two point sizes is a specification that comes back right the first time.

What do you type into Vizard Agent?

Write it as a spec. This is one of the few requests where a list of numbers is easier than a sentence, because every item is a value somebody has to choose and you already know what you want.

Prompt

Actor cards: line one in Latin at 58pt, line two in Korean at 96pt so the Korean is clearly larger, with generous line spacing. A classic cinema serif for the Latin, a traditional serif for the Korean. Solid magenta on black — no glow, no blur, no bleed, crisp edges, and reinforce the strokes evenly so they do not look thin.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real bilingual trailer whose cards had to look like cinema typography rather than slides. The font search happens before any design, which is the part that decides whether the rest is possible.

  1. Checks which typefaces on the system can render the script at all.
  2. Looks for a face with the right character for the language.
  3. Locates the capable font file and confirms it covers the text.
  4. Designs the cards with each language at its own size.
  5. Checks the cards as rendered images.
  6. Builds the full set of cards and the end title.
  7. Renders the file and checks the cards in place.
  8. Checks the cards and the ending at full size.

Step one is not a formality. A typeface that looks right in a specification may have no glyphs for your script, and the failure is silent — boxes, or a substituted face that quietly changes the whole card's character.

Step five is worth insisting on. A card judged inside the edit is judged at whatever size the timeline preview happens to be, and typography that looks balanced there can be crowded or spindly on the actual frame, so each card gets looked at as a rendered image on its own first.

Step four is where the two languages stop competing. Setting each line at its own size means the card is designed rather than scaled, and the leading can be opened up so two scripts of different heights do not crowd each other.

What does the result look like?

A card where the leading language reads first because it is larger, the second sits beneath it in a face chosen for it rather than inherited from the first, and both hold their edges cleanly against black at full size.

It looks set rather than typed.

When does this not work well?

Some combinations resist. A very thin face at a large size on black loses its strokes to the display, and a script with tall ascenders under one with deep descenders may need the leading opened past what the card has room for.

How do you fix a result that came back wrong?

Give Vizard Agent the value rather than the impression it left on you. It keeps the definitions behind every card it built, so changing a size or a colour is an edit to a single number and nothing else about that card moves at all.

How does Vizard Agent compare to doing it yourself?

By hand the second language usually inherits the first one's settings, because that is what a text box does. The font check is also skipped until something renders as boxes, at which point the design has already been built around a face that cannot be used.

By hand Vizard Agent
The second language Inherits the first's size Set at its own
Font coverage Discovered when it fails Checked before designing
Weight on black Bolded Strokes reinforced evenly
Colour Picked by eye Set to the value you give
Checking On a preview At full size, as rendered

Common questions

Can it find a font for my script? Yes. Vizard Agent checks what can actually render it first.

Can each language have its own typeface? Yes, and Vizard Agent finds it usually should.

Will it scale one to match the other? No, not if you give both sizes to Vizard Agent.

What if the strokes look weak? Vizard Agent reinforces them evenly rather than switching to bold.

Can I give an exact colour? Yes. Vizard Agent works better from a value than a colour name.

Will it add a glow? Only if you ask. Say no glow and Vizard Agent leaves it flat.

Can it match an existing card? Yes. Send one and Vizard Agent measures it.

What about right-to-left scripts? Yes, though tell Vizard Agent the line order explicitly.

Can it do three languages? Yes, though Vizard Agent finds the hierarchy harder to hold.

Will the cards animate? Only if you ask Vizard Agent to; a still card is often stronger.

Can it change size without touching the font? Yes. Say size-only and Vizard Agent changes nothing else.

What if a name is too long? Vizard Agent will show you the wrap and ask.

Can I get the cards as images? Yes. Ask Vizard Agent to export them.

Why give point sizes rather than describe it? Because "clearly larger" is a range and 96 over 58 is a card, and only one of those comes back the way you pictured it.