How to rebuild a landscape promo as a real vertical instead of cropping it

Last updated 2026-09-05 · 8 min read

What is the short version?

A graphics-led promo does not crop. Its text was set for a wide frame, its cards were laid out for a wide frame, and squeezing that into a vertical one gives you either tiny type or half a diagram.

1. Go to Vizard Agent with the landscape master.
2. Say you want a redesign for vertical, not a crop or a reframe.
3. Say the voiceover timing must be preserved.

What do you need before you start?

The finished landscape video, and nothing else. You do not need the original project files or the fonts, because Vizard Agent takes the typography out of the render itself — which is usually more faithful than tracking down whatever was used two years ago.

• The landscape master. As delivered.
• The target frame. Vertical, usually.
• The audio. Timing is the fixed spine.
• Brand assets, if you have them. Logos, colours.
• What must survive. The claims, the product shots, the end card.

What do you type into Vizard Agent?

Say "not a crop" explicitly and early. Reframing is the reasonable default reading of this request for Vizard Agent, and it is exactly the thing you are trying to avoid — those four words are what separate a genuine rebuild from a squeeze.

Prompt

Variants worth knowing:

• "Not a crop or a reframe." The instruction that changes the job.
• "Keep the voiceover timing." The spine everything hangs on.
• "Re-lay the panels." Says the layout may move.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real software promo rebuilt from scratch for a vertical frame. Step five is the decision the entire piece rests on, and it is emphatically not the obvious move that most people would reach for first.

1. Pulls detailed frames from the opening and studies the sequence.
2. Reviews each section — the logo reveal, the diagram, the cards.
3. Crops text samples to identify the typeface.
4. Lists the available fonts to compare against it.
5. Extracts the words out of the source as high-resolution images instead.
6. Tunes the keying and re-extracts them cleanly.
7. Compares the white text against the coloured backgrounds behind it.
8. Finds a clean frame before a wipe starts to extract from.
9. Re-extracts the pieces that were clipped by the animation.
10. Verifies every extracted element's crop.
11. Crops the interface panels and checks their boxes.
12. Maps where each text block sits in every scene.
13. Measures the cards, the diagram and the inner panels.
14. Finds one card's true right edge where the crop was ambiguous.
15. Renders still previews of every scene in the new layout.
16. Re-renders the previews after adjustments.
17. Adds shadows and checks the key frames closely.
18. Renders the background bed and the dark opening.
19. Reviews the whole cut frame by frame against the source.
20. Verifies the fixed transitions before delivering.

Step five is the one worth stealing. The obvious move is to identify the font and retype the words — and the nearest installed font is never quite it, so every line of text looks subtly wrong. Lifting the actual rendered words out of the source as images keeps the real typeface, the real weight and the real letter-spacing, because they are the original pixels.

Step eight is the practical consequence. A word that is mid-wipe cannot be extracted cleanly, so the extraction happens on a frame before the animation starts — which means watching the sequence closely enough to know where those frames are.

What does the result look like?

A vertical promo that reads as though it had been designed vertical: text at a size that works on a phone, panels stacked rather than squeezed, the diagram legible, and the voiceover landing exactly where it did before.

Nothing is a scaled-down version of the wide one. That is the difference people see without being able to name it — a cropped promo always looks like a cropped promo.

When does this not work well?

The method depends entirely on the source render being clean enough to lift elements out of. Heavy compression, very small text and busy moving backgrounds all make extraction harder for Vizard Agent, and some designs simply have no sensible vertical arrangement at all.

• Heavily compressed sources. Extracted text carries the artifacts.
• Text over moving footage. Nothing clean to key against.
• Very wide diagrams. Some things have no tall arrangement.
• A voiceover written to wide visuals. The words may not fit the new order.
• When you have the project file. Rebuild it there instead; it is easier.

How do you fix a result that came back wrong?

Point at the specific element that is wrong. Vizard Agent keeps the extracted assets, the measurements it took and the scene layouts as three separate things, so moving a card will not disturb the text sitting on it or the audio timing underneath.

• "That word looks soft." Re-extracted from a cleaner frame.
• "The card is cut off." Its true edge re-measured.
• "The text is too small." Re-laid at a larger size for the frame.
• "It drifts from the voiceover." Re-timed to the original audio.

How does Vizard Agent compare to doing it yourself?

With the original project file this is a designer's afternoon and the right way to do it. Without it — which is the usual case, because the promo came from an agency two years ago — it is a rebuild from a render, and the typeface is where that normally falls apart.

Common questions

Why not just crop it? Because text set for a wide frame becomes unreadable, and diagrams lose half their width.

Do I need the project file? No. Vizard Agent rebuilds from the rendered video.

How does it get the font right? It does not match the font. Vizard Agent lifts the rendered words themselves.

Will the voiceover change? No. Vizard Agent treats its timing as fixed and lays the picture to it.

Can it re-order the scenes? Yes, if the audio allows. Tell Vizard Agent whether the order is fixed.

What about the logo? Extracted or supplied. Vizard Agent prefers your real asset if you have it.

Can it change the text? Yes, though new words need a real font. Say which.

Will it look identical? No — it is a redesign. It should look like it belongs to the same brand.

What if the source is low quality? Vizard Agent tells you what it could not extract cleanly.

Can it do square as well? Yes. Vizard Agent works from the same extracted elements.

Does it keep the animations? Vizard Agent rebuilds them for the new layout rather than copying them.

How long does this take? Longer than a reframe, much less than a redesign from scratch.

Can it match a newer brand style? Yes. Give Vizard Agent the current assets and it re-lays to those.

Why extract the words as images? Because a font that is nearly right is visible on every line, and the source already contains the letters set exactly as the designer intended.