How to Keep a Logo the Same Size in Every Shot of a Short AI Video
Make this article actionable
Send the article context into Vife Agent and turn it into a plan, checklist, or draft you can keep working on.
Quick answer
If your logo or sticker looks bigger in the close shot than in the wide shot, the problem is almost never the logo itself. It is the object-to-frame ratio: how much of the frame the branded object fills in each shot. When that ratio changes, the sticker changes with it, and the clip reads as inconsistent even though every individual shot looks fine.
The fix is to lock the ratio before you generate, change only one variable per shot, and review by comparing the sticker-to-box height ratio rather than reading any text.
Turn the useful parts into next steps
Vife Agent can convert this guide into a prioritized workflow with tasks, risks, and reusable prompts.
Why a logo "grows" between shots that are each fine on their own
A 15-second vertical product clip usually cuts between three framings: wide, medium, close. Each shot is generated separately, so each one gets its own interpretation of "how big is this box in the frame."
Three things cause the drift:
- Distance drift. The box sits further from the camera in one shot than another. The sticker scales with the box, so it scales with the distance change.
- Framing-note drift. One shot says "wide product shot," another says "close-up detail." Those are different instructions, and the object fills a different share of the frame in each.
- Crop magnification. If the box sits near the frame edge in one shot and gets trimmed, the visible portion of the sticker can appear larger relative to what remains.
None of these are model failures. They are framing decisions you can specify before generating.
The ratio that actually matters
Forget pixel dimensions and forget the sticker's absolute size. What you are managing is:
sticker height ÷ box height — as it appears in the frame
If that ratio holds across all three shots, the clip reads as one continuous product. If it swings, viewers notice immediately, even if they cannot say why.
Step-by-step workflow
Step 1: Fix the object-to-frame ratio before generating each shot
Decide in advance how much of the vertical frame the box should occupy, and keep that number constant across all three shots. A practical starting point for a 15-second vertical clip:
| Shot | Framing intent | Box share of frame height | What changes |
|---|---|---|---|
1 | Wide establishing | ~45% | Nothing — this is your baseline |
2 | Medium | ~45% | Camera height or lighting direction |
3 | Close | ~45% | Background tone or lighting direction |
The counterintuitive part: the box stays roughly the same size in frame across all three shots. You create the feeling of "closer" through lighting, background, and camera height — not by enlarging the object. That is what keeps the sticker stable.
Keep the box at the same distance from the camera and reuse the same framing note in every prompt. Copy-paste it rather than retyping it; small wording changes are how drift creeps in.
Step 2: Change only one variable per shot
If you change framing, lighting, background, and camera height at once, you cannot tell which change caused a scale shift. Change one thing:
- Shot 1 → Shot 2: change lighting direction only.
- Shot 2 → Shot 3: change background tone only.
Camera height is a good third variable if you want a low or high angle, but treat it as its own change, not a bundle.
Step 3: Reserve a consistent margin around the object
Leave breathing room on all sides — roughly 15–20% of the frame width as empty space on each side. This does two things:
- It prevents the box from being trimmed at the edges, which is the main cause of crop magnification.
- It gives you room to reframe in editing without pushing the sticker toward the edge.
If a shot comes back with the box touching the frame boundary, regenerate it rather than cropping it in post. Cropping a tight shot is exactly how a sticker ends up looking oversized.
Step 4: Review by ratio, not by reading
Pause each shot at a moment when the box is fully visible. Compare the sticker's height to the box's height by eye. You are looking for a stable proportion, not a legible label.
This is why the example uses a plain unbranded cardboard box with a simple circular sticker: no text to read, no brand to recognize. If the sticker reads as the same physical size relative to the box in all three shots, the sequence works.
Example input
Here is a copyable prompt structure you can adapt. It is an example, not a tested result — adjust the wording to your own object and check the output yourself.
Vertical 9:16 product clip, single plain unbranded cardboard box
with one simple circular sticker centered on the front face.
Framing note (identical in every shot): box occupies about 45% of
frame height, centered, with clear empty margin on all four sides;
camera at the same distance from the box in every shot.
Shot 1: wide establishing framing, soft even light from the left,
neutral mid-tone background.
Shot 2: same framing and distance, light now from the right,
background unchanged.
Shot 3: same framing and distance, lighting unchanged,
background shifted to a darker tone.
Keep the sticker's size relative to the box constant across all shots.
No text, no logos, no labels.The framing note is the load-bearing part. Repeat it verbatim in each shot description rather than summarizing it.
You can build and assemble these shots in a single workspace on the AI product video generator page, which is where the shot generation and clip assembly live. If you are weighing plan options for a small team, /pricing covers the tiers.
Output review checklist
Run this before you export. It takes about a minute per clip.
- Sticker-to-box ratio is visually stable across all three shots when paused side by side.
- No shot relies on legible text — the sticker reads as a shape, not a word.
- No screenshots or external reference images were needed to judge consistency.
- Box margin is consistent — no shot has the box touching or crossing the frame edge.
- Only one variable changed per cut (lighting, background, or camera height).
- Framing note is identical in every shot's prompt.
- No shot was rescued by cropping in post.
If a shot fails the ratio check, regenerate that shot with the same framing note rather than trying to scale it in editing. Scaling a single shot to match is a fix that tends to introduce its own mismatch.
Common mistakes
Changing framing language between shots. "Wide shot" and "close-up" are different instructions. If you want the box to stay the same size, keep the framing language the same and vary the atmosphere instead.
Letting the box fill the frame. A box that fills 80% of the frame has no margin, and any small framing difference becomes a large visible scale difference.
Reading the sticker instead of measuring it. If your sticker has text, you will unconsciously judge legibility rather than proportion. Use a plain shape for consistency checks.
Bundling changes. Changing lighting and background in the same cut makes it impossible to diagnose a scale shift.
Fixing it in post. Cropping or scaling one shot to match the others usually trades a scale problem for a composition problem.
Assuming a prompt guarantees consistency. A detailed brief improves your odds of comparable framing, but it does not control how any given generation resolves scale. Always review the output.
FAQ
Why does my logo look bigger in the close shot?
Because the object fills more of the frame in that shot. The sticker scales with the object, so a larger object share means a larger sticker. Keep the object-to-frame ratio constant and vary lighting or background instead of framing.
Should I just make the logo smaller in the close shot?
That creates a new mismatch: the sticker would then be too small relative to the box, which is a different inconsistency. Fix the framing ratio at generation time rather than compensating per shot.
How do I check consistency without a reference image?
Pause each shot with the box fully visible and compare the sticker's height to the box's height by eye. You are checking a proportion, not a measurement, and you do not need a screenshot or an external reference to do it.
Does this work if my product has text on it?
The ratio method still applies, but text makes review harder because you start judging legibility. For consistency checks, test with a plain shape first, then apply the same framing discipline to the branded version.
How many variables can I change per shot?
One. Lighting direction, background tone, or camera height — pick one per cut. That keeps scale comparable and makes any drift easy to trace.
What if one shot still drifts after I reuse the framing note?
Regenerate that shot with the identical framing note and check the margin. If the box is touching the frame edge, the crop is magnifying the sticker — that is the most common remaining cause.
Do I need to keep the box at exactly the same distance?
Keep it at the same distance and reuse the same framing note. Distance and framing note are the two inputs that most directly determine the object-to-frame ratio, so treat them as fixed rather than adjustable.
The short version
Lock the object-to-frame ratio, change one variable per shot, keep a consistent margin, and review by proportion instead of by reading. That is the whole method — and it works because you are managing a ratio, not a logo.

