How to neatly centre text on a screen inside a scene
Upload the clip and tell Vizard Agent the text on the screen should sit neatly centred. Vizard Agent measures the panel's tilt and tracks its drift across the shot, matches the original typeface, tests the result under every lighting condition in the clip, then paints out the old text and renders the new.
What is the short version?
Text on a screen inside a shot is not an overlay. It sits on a surface that is tilted, that moves with the camera, and that is lit by whatever is happening in the room, so centring it means measuring the surface rather than centring the frame.
- Go to Vizard Agent and upload the clip.
- Say what is wrong with the text — off-centre, crooked, badly spaced.
- Give the wording exactly as it should read.
What do you need before you start?
The clip and the exact wording. Vizard Agent measures the surface, matches the type and handles the lighting itself, so what has to come from you is the text itself and any preference about how it should break across lines.
- The clip. With the screen visible.
- The wording. Exactly, including capitalisation.
- The problem. Off-centre, tilted, uneven line spacing.
- A line-break preference. Four lines or five changes the whole look.
- The original, if you have it. Useful when a previous edit went wrong.
What do you type into Vizard Agent?
Say what neat means to you. Vizard Agent will centre the text on the panel by measurement, and telling it whether you want four lines or five, or a particular emphasis, saves it guessing at a layout decision that is genuinely yours.
Prompt
Variants worth knowing:
- A specific typeface. Or let Vizard Agent match the original.
- A line-break choice. Say four or five lines if you have a preference.
- A colour change. Vizard Agent tests it against the panel's own light.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real screen-text fix. It is almost entirely measurement and testing: the geometry of the panel, the typeface, and then the same text rendered under every lighting condition in the clip.
- Probes the clip and extracts a contact sheet of frames to see the composition.
- Reviews the earlier version and the original upload to understand what changed and how.
- Crops into the text area and compares the current layout against the original closely.
- Measures the coordinates of the panel, detects its boundaries and centre point.
- Analyses the camera movement at several points and tracks how the panel drifts across the shot.
- Checks the installed serif fonts and downloads a matching typeface to sit with the original styling.
- Measures the panel's tilt angle and the centre line of the original text rows.
- Renders several layout variations, compares them in a grid, then tests the type weight and realistic panel glow.
- Extracts frames under four different club lighting conditions, tests the inpainting that removes the old text under each, compares a four-line against a five-line layout across all of them, then optimises the process and renders the full clip.
Step 9 is what separates this from a text overlay. A replacement that looks clean in the dark falls apart when a yellow strobe hits the panel, so Vizard Agent tested the hardest lighting moments before committing.
What does the result look like?
From the run this page is written from, probed on the delivered file: 576x1024, H.264, 30fps, 19.39 seconds, AAC audio. Vertical, twenty seconds, the old text painted out and new text rendered centred on the panel, holding through the camera movement and the changing light.
Twenty seconds sounds trivial and is not. Every frame has the panel in a slightly different place under slightly different light, so the correction has to be recomputed continuously rather than applied once.
When does this not work well?
This is compositing onto a real surface inside a real shot, and the qualities of that shot decide how well it can possibly go. Vizard Agent measures the geometry and tests the lighting, and some material simply does not allow a clean result at all.
- Heavy motion blur resists it. A fast pan smears the panel edges and the replacement with them.
- Extreme lighting is the hard case. A strobe flash across the panel is exactly where this breaks.
- The typeface can only be matched, not recovered. Vizard Agent finds the closest available.
- Something crossing the screen occludes it. A person walking in front needs separate handling.
- Long clips are expensive. Every frame is recomputed, which is why this run was twenty seconds.
How do you fix a result that came back wrong?
Name the moment. Vizard Agent keeps the panel measurements, the tilt angle, the tracked drift, the layout variations and the lighting tests, so a change of wording or line break is a re-render against geometry it already solved.
- "Use five lines instead." Already rendered and compared; a straight swap.
- "It slips at the end of the pan." Re-tracked against the drift measurements.
- "The glow looks fake under the flash." Re-tested against that lighting condition specifically.
How does Vizard Agent compare to doing it yourself?
By hand this is corner-pinning a text layer to a moving panel, tracking it through a camera move, painting out the original text frame by frame, and then discovering the whole thing falls apart when the lights change colour.
| By hand | Vizard Agent | |
|---|---|---|
| The panel geometry | Corner-pin by eye | Boundaries, centre and tilt measured |
| Camera movement | Track and correct | Drift tracked across the whole clip |
| The typeface | Pick something close | Matching font found and downloaded |
| Lighting changes | Find out at render | Tested under four conditions first |
Common questions
Will it match the original font? Vizard Agent looks for the closest installed typeface and downloads a match when it needs one.
Does it work while the camera moves? Yes. Vizard Agent tracks the panel's drift and recomputes the placement per frame.
Can it change the wording entirely? Yes. The old text is painted out before the new is rendered.
What if the lighting changes? That is the hard case, and Vizard Agent tests the replacement under each condition before rendering.
Does it measure the panel's angle? Yes. Vizard Agent measures the tilt of the surface and the centre line of the original rows, so the new text sits on the panel rather than on the frame.
Can I choose the line breaks? Yes, and Vizard Agent will render the alternatives side by side if you are unsure.
How long a clip can it handle? Short clips are the comfortable case. Every frame is recomputed, so the cost scales with the length.
Does it handle the panel's glow? Yes. Vizard Agent renders the new text with realistic weight and panel light rather than pasting flat type onto the surface.
Can it fix a previous edit that went wrong? Yes. Vizard Agent compared the earlier version against the original upload on this run before deciding what to do about it.
How does it remove the old text? By inpainting the panel, tested separately under each lighting condition so the clean background holds when the light changes.