A normal zoom makes the subject larger or smaller. A dolly zoom combines a change of camera position with an opposing lens change. The foreground can remain similarly framed while the background appears to stretch or compress. That difference is the point of the shot, so “dramatic zoom” is usually an incomplete AI video brief.
Use the free lab above to make the relationship visible before writing a prompt. It draws a coral panel and five columns in an idealized scene. These are original calculated shapes, not frames from an AI video generator. Move the distance slider and compare a fixed lens with a compensating zoom. Nothing is uploaded or generated.
What should stay fixed in a dolly zoom
Choose a subject whose size is easy to judge: an upright product, a still character or a sign in an invented scene. Hold its pose, appearance and approximate frame height. Then ask for the camera move. If the subject independently walks toward the camera, it becomes difficult to tell whether the intended optical relationship occurred.
Adobe explains the dolly zoom as opposing camera movement and zoom, with a similarly sized subject and a changing background. Our lab isolates that idea. It omits lens distortion, focusing and real-world camera mechanics; its numbers are illustrative, not calibrated shooting instructions.
Try a move with an unmistakable background
In the lab, the camera starts three metres from the foreground panel with a 35 mm reference focal length. The columns are five metres behind the panel. Move backward: the fixed-lens panel shrinks, while the compensating view keeps it the same height. The columns become more prominent relative to that stable foreground.
For the idealized perspective used here, projected subject size is proportional to focal length divided by distance. Doubling both preserves that ratio. This does not mean that a video model accepts metre or lens controls, or follows a computed path precisely. Review generated framing visually even when you include numerical language in the prompt.
Use this original AI dolly zoom prompt
One continuous dolly-out shot of a still ceramic lamp on a workshop table. As the camera moves away, zoom in to keep the lamp approximately the same height in the frame while the shelves behind it appear to compress. Keep the lamp, table and light direction consistent. No cut, independent lamp movement or simple digital crop. End on a steady composition.
This is an untested creative brief. It defines one subject, one move and a visual acceptance condition. Choose actual duration, resolution and framing in the selected video workflow. The lens values in the lab are not additional API parameters to paste into an unrelated model.
For a push-in version, reverse the relationship: the camera moves closer while the lens widens to preserve the foreground framing. Keep the initial test short enough to review carefully. A full orbit, character action, rapid lighting change and dolly zoom in the same shot introduce several independent failure modes.
Choose image to video or a new scene deliberately
An approved opening image provides a visual baseline for identity, object shape and framing. It does not establish the unseen space behind the camera. Keep the move modest when a specific product or real place must remain recognizable. For a completely invented scene, text-to-video can describe the setting as well as the move.
Clout's Wan video page offers a separate shot-brief planner and the real studio entry. Select the model and operation in the studio, apply its supported controls and review the current credit quote. A camera instruction is a request to evaluate, not proof that a particular model implements a physical lens simulation.
Review the effect at three moments
| Moment | Inspect | What a failure suggests |
|---|---|---|
| Opening | Subject height, pose and background landmarks | No clear baseline for the move |
| Middle | Stable subject size with changing background relationship | A simple zoom, pan or independent subject action |
| Ending | Consistent object geometry and a readable finish | Warping or a move that runs past the useful composition |
Watch the complete exported clip before extracting comparison frames. A convincing first and last image can hide a cut or a warped middle. Compare a stable feature such as the lamp base with its initial size, and use several background landmarks rather than a single blurred edge.
For a fixed-camera animation with deliberate pose changes, use the AI stop motion guide. A camera move and an object animation solve different storytelling needs.
Fix the specific camera failure
If everything grows uniformly, simplify the instruction to a dolly-out with a compensating zoom and a fixed subject. If the subject changes size unevenly, choose a smaller movement and remove competing actions. If the background invents extra objects, use a simpler scene or stop the move sooner. Keep the accepted source and prompt so each revision has a clear comparison.
A dolly zoom changes depth relationships. Bullet time instead emphasizes a moving viewpoint around a held moment. The broader camera movement prompt guide covers ordinary pushes, orbits and tracking shots. Match the technique to the story rather than using all the terms in one prompt.
