AI image and video prompts. Give your idea a direction.
A useful prompt describes a scene you can picture. Start with a subject and setting, then choose the light and framing. For a video, add one action and one camera movement. This guide gives you six original examples to adapt in SceneCraft.
Updated · Examples are writing samples, not tested output comparisons. “Try this prompt” fills the studio; it does not start generation.
Start with the scene, then choose the medium
An image prompt describes a single frame. A video prompt must also explain what changes over time. Write the essential scene first; adding more adjectives rarely resolves conflicting directions.
An image prompt template
[Subject] in [setting], [light direction and quality], [framing], [visual treatment].
“A cup on a table” leaves most choices open. Naming the material, surface, light direction and camera distance gives the picture a more specific brief. Use the studio controls for aspect ratio and resolution.
A video prompt template
[Subject and setting]. [One action]. The camera [one movement or stays still]. [Lighting and mood]. One continuous shot.
Short clips need a focused action. Describe a small moment rather than a full sequence with multiple cuts. Choose duration and frame shape in the studio; a word in the prompt does not replace those settings.
Three AI image prompt examples
Use these as starting briefs for a still life, an environment or a portrait. The buttons also select the example’s frame shape. Learn about model settings in the AI image generator guide.
1:1 · TEXT TO IMAGE
Product still life
A handmade ivory ceramic cup on a warm gray tabletop, soft window light from the left, a gentle shadow to the right, close product photograph, clean background, no lettering.
Names one subject, its surface, the light direction and the composition. Replace the cup with your own product idea without changing every other detail.
An editorial portrait of an adult wearing a dark linen jacket beside a rain-streaked window, soft side light, natural skin texture, shallow depth of field, muted colors, vertical composition.
Clothing, light and framing are concrete choices. Start with a fictional subject; an image reference or a real person’s likeness is a different request.
Each example describes one shot and a restrained motion. Current video examples assume silent output; narration or music in a prompt does not enable audio. Check the video model and credit table before choosing clip settings.
16:9 · TEXT TO VIDEO
A single tracking shot
A lone cyclist follows a quiet coastal road at sunrise. The camera tracks slowly from the side while the sea glows in soft morning light. One continuous shot.
Uses one action and one camera movement. A focused shot is easier to evaluate than a prompt that asks a short clip to cover a whole story.
Close-up of a ceramic cup on a wooden table. A thin ribbon of steam rises slowly in soft morning light. The camera stays still. Vertical framing, one continuous shot.
The steam supplies the motion while the camera stays still. This is a text description of a new scene, not animation of an uploaded product photo.
A misty valley just after sunrise, layers of distant hills in pale gold light. The camera slowly tilts upward from the still grass toward the horizon. One continuous shot.
Gives the camera a starting point and an end direction. If the motion feels busy, simplify the prompt before adding more actions.
Keep the subject and framing while you explore a different lighting direction. For the ceramic still life, try replacing “soft window light from the left” with “hard afternoon sunlight from the left.” Leave the rest unchanged so you can judge what you asked to change. This is a suggested experiment, not a measured comparison of outputs.
If the composition is unclear
Name the main subject and its position. Remove extra objects that compete with it. Pick the frame shape in the studio and use camera distance, such as close-up or wide shot, to explain how much of the scene belongs in the frame.
If the movement is too busy
Try a stationary camera and one moving subject, or a still landscape with one camera movement. Remove conflicting instructions such as a locked-off camera followed by a fast orbit. Add complexity after you have a shot you can evaluate.
Plan a small set of iterations
At the current example settings, three Qwen Image 3.0 2K image submissions use 21 credits; three Wan 2.6 720P 5-second silent video submissions use 300 credits. Each submission is a separate task. Changing settings changes the quote.