Before writing a Runway prompt, decide whether you are asking the model to imagine a scene from text or animate an image you provide. That distinction changes what the prompt needs to explain. A scene description is useful when appearance is open; motion direction is more useful when an image already defines the shot.
This guide follows Runway’s current official text-to-video prompting guide and image-to-video prompting guide. It is not a benchmark or hands-on review. Check the exact model and interface you choose, because controls and recommendations can change.
Define the shot before writing the prompt
Write a single sentence about the purpose of the shot: “The viewer sees a sailboat moving peacefully across the lake.” That sentence helps you decide which details are necessary. It also creates a review criterion for the output.
Now choose the main action, environment, and camera intention. A short clip usually benefits from a manageable idea. If the story needs a boat to depart, cross the lake, reach a dock, and unload its passengers, plan several shots rather than treating the whole journey as one instruction.
Keep a separate note for settings. Aspect ratio, duration, and available model options should be chosen in the tool. They are not guaranteed to change because you write them in descriptive prose.
For text-to-video, describe appearance and motion
Runway’s text-to-video guide organizes prompting around visual and motion information. Use concrete details: what is present, how it looks, and what changes. There is no universal fixed sentence formula that must be used for every shot.
A practical first draft could be:
A small white sailboat glides across a quiet lake at sunrise. Mist drifts slowly above the water. A steady camera follows the boat from the shore, with soft golden highlights and natural colors.
The visual information establishes the boat, lake, time, and light. The motion information describes the boat, mist, and camera. Each piece supports the same calm scene. The example is a writing illustration rather than a documented generated result.
If a precise appearance matters, add that detail directly. “A compact red bicycle with a black basket” is clearer than repeatedly asking for a beautiful bicycle. Do not keep expanding the prompt just to make it sound sophisticated.
For image-to-video, let the image do its job
An input image already communicates much of the subject, composition, light, and style. Runway’s image-to-video guide emphasizes using the text to direct movement. Describe the subject’s motion, changes in the environment, and the camera action you need.
For an image showing the sailboat, a motion-oriented draft might be:
The sailboat moves steadily across the water, leaving a gentle wake. Mist drifts slowly. The camera follows the boat in a smooth, restrained tracking movement.
Repeating a long description of the existing image can distract from the motion goal. If the starting image does not contain the composition you need, consider improving the reference before trying to solve everything through the prompt.
You can also describe an intentional sequence, but keep it understandable. Use clear timing words only when timing matters. Several simultaneous, unrelated changes can make both generation and review harder.
Avoid conflicting camera instructions
Pick one primary camera behavior. “Static camera” and “rapid orbit around the subject” are incompatible directions unless you explicitly plan a transition. Start with the move that best communicates the action.
Watch the output for what actually moved. Sometimes the subject and camera motion can create a different impression from the one you intended. Rewrite the direction in observable language, such as “the camera remains in place while the bicycle passes from left to right.”
Avoid treating a list of photography terms as a reliable command language. Natural, unambiguous descriptions give you a more useful working brief and a clearer basis for revision.
Review the current model’s limits
Do not transfer every rule from an older Runway model to a newer one without checking the documentation. Negative prompt support, output controls, reference behavior, and model availability are version-specific. This article does not add a generic negative list or assume an older rule applies to every current model.
Our prompt generator offers Runway direction for text-to-video and image-to-video. It builds a positive template with separate settings notes. An optional reference preview stays in your browser without analysis or upload. The tool does not call Runway, set provider parameters, or generate a video.
Keep a small revision log
Save the prompt, model name, settings, input reference, and a short observation. Change one main instruction when possible. If the camera is the problem, keep the subject and lighting stable while revising camera direction.
When a shot serves its story purpose, move to the next shot rather than endlessly polishing adjectives. The storyboard guide explains how to connect those individual decisions into a coherent sequence.