How to stop a title zooming with the shot behind it
Tell Vizard Agent which part of the title should change and which should hold. A title composited into the same layer as the picture inherits every camera move applied to it, so the fix is to separate them and animate one — not to fight the scaling with a bigger font.
What is the short version?
Your opening pushes in on the shot and the title shrinks with it, which is not what a title is supposed to do. The cause is layering, not timing. Vizard Agent pulls the title out of the footage layer so the two can move independently.
- Tell Vizard Agent what should hold and what should move.
- Let it separate the title from the picture first.
- Review the animation frames, not a single still.
What do you need before you start?
Just the video and a clear sentence about the intent. "The title should stay the same size while the shot pushes in" is unambiguous; "the title looks wrong at the start" is where most of these conversations begin and costs a round trip to resolve.
If the title has a backing plate — a coloured box or a cream panel behind the words — decide what it should do separately. Text and plate are genuinely different elements, and the most natural-looking version is often the one where the plate shrinks away while the lettering holds. Vizard Agent can do either, but not by guessing.
What do you type into Vizard Agent?
Name the parts. The reason this job goes around in circles is that "make the title smaller" describes three different results depending on which layer you meant, and all three are reasonable readings of the same four words.
During the opening zoom, keep the lettering the same size and shrink only the cream background behind it.
That is very close to the wording that resolved it in the session behind this article, after an earlier round had shrunk the lettering itself. Being specific about which element moves is what turned a repeated correction into a single change.
What does Vizard Agent actually do?
It looks at how the title was built before it changes anything. A title that is a separate layer is trivial to animate on its own; a title that was rendered into the footage is not, and the difference determines whether this is a small change or a rebuild of the opening.
In that session the sequence ran:
- Checked how the opening title had been built and which layers existed.
- Looked at the opening frames to see how the title currently behaves through the zoom.
- Separated the title from the footage so the two could scale independently.
- Kept the existing stew zoom untouched and changed only the title's size and spacing.
- Rendered it, checked it, and reduced it further when the first pass was not enough.
- Later reversed the choice entirely — holding the lettering and shrinking only the cream plate.
That last step is the useful one. Because the layers were separated, changing which element moves became a small adjustment rather than another rebuild.
What does the result look like?
The push-in reads as a camera move and the title reads as a title. The words stay legible at a constant size while the picture behind them changes, or the plate recedes and leaves the lettering standing, depending on what you asked for.
Vizard Agent checks the title's margins across every frame of the animation rather than on one still, because a title that is correctly placed at the start and the end can still clip against the top of the frame in the middle. That check is specific to animated titles and is easy to skip.
When does this not work well?
When the title is burnt into a delivered file and you no longer have the project. Then separating it means removing it from the picture and rebuilding it, which works but leaves the area behind the words to be reconstructed. Vizard Agent will tell you that is the job rather than pretending it is a layer change.
It is also harder when the title sits over moving detail. A word over a plain wall separates cleanly; a word over boiling liquid does not, and the reconstruction has to invent what was underneath. The result is usually fine at speed and less convincing paused.
And there is a point where the design is the problem. If the title has to hold its size while the shot pushes in a long way, it will end up looking pasted on regardless of the technique — at that point the honest answer is a different opening, not a better composite.
How do you fix a result that came back wrong?
Say which element ended up wrong, and in those terms. "The lettering shrank, I meant the box" is a one-line correction once the layers exist, and Vizard Agent can swap which one animates without touching the footage underneath.
If the size is right but it now collides with something, give the timestamp. In that session the title needed lifting at three seconds because the new framing brought the food up into it — a position change, not a size change, and much cheaper.
If the text has gone soft, it was scaled up rather than re-rendered. Ask Vizard Agent to rebuild the lettering at the final size instead of enlarging the existing layer, which is the difference between crisp type and a blurred version of small type.
How does Vizard Agent compare to doing it yourself?
In an editor with the project still open, this is a two-minute fix — unlink the scale, keyframe one layer, done. The difficulty is entirely about whether you still have the layers, and that is exactly the situation where people end up recreating an opening from scratch.
What Vizard Agent adds is that it checks how the title was built before it starts, so you find out in the first thirty seconds whether this is a layer change or a reconstruction. And it checks the title's position across every animation frame rather than at the ends, which is where the clipping people notice after publishing actually happens.
Common questions
Why did the title shrink in the first place? Because it was in the same layer as the footage, so the zoom applied to both.
Can the title move the opposite way to the shot? Yes. Once Vizard Agent separates the layers, the two animations are independent.
Will separating the layers change the footage? Not if the title was a separate element. If it was burnt in, the area behind it has to be rebuilt.
Can I shrink the plate and keep the text? Yes, and it often looks better. Say it that way and Vizard Agent does exactly that.
Does the type stay sharp? If Vizard Agent re-renders it at the target size, yes. Scaling an existing layer up does not.
What if the title collides with the subject? Give the timestamp and Vizard Agent lifts or moves it at that moment.
Can I change the font at the same time? Yes. Once the title is its own layer, Vizard Agent can rebuild it entirely.
Does this work for lower thirds too? Yes. Any graphic welded to the picture has the same problem and the same fix.
How do I know the margins are safe? Vizard Agent checks the top margin across every animation frame rather than at the ends.
Will the timing change? No. Vizard Agent keeps the existing zoom and changes only the title.
Can I animate the letters individually? Yes, though that is a rebuild of the title rather than a layer separation.
What if there are two titles on screen? Name them. Vizard Agent treats each as its own layer.
Does the shadow follow the text? It should. Vizard Agent keeps the shadow attached to whichever element it belongs to.
Can I see a still before the full render? Yes. Vizard Agent renders the animation frames for review before committing the export.