Creating impressive AI-generated images and videos does not require complicated technical language. It requires clear instructions.
An AI generator cannot see the idea in your head. It only understands the information you include in your prompt. The more clearly you describe the subject, environment, action, camera, lighting, and visual style, the more control you have over the final result.
Quick answer
A strong AI prompt should explain what should appear, what the subject is doing, where the scene takes place, how the camera sees the scene, what lighting and visual style to use, and what should be avoided.
For images, concentrate on the appearance and composition of a single moment. For videos, also describe movement, timing, camera motion, and how the scene should begin and end.
What is an AI prompt?
An AI prompt is a written instruction that tells an artificial intelligence model what to create.
For example, “A woman standing in a city” is technically an image prompt, but it leaves most creative decisions to the AI. The model must decide what the woman looks like, which city she is in, what she is wearing, what time of day it is, and how the image should be composed.
A young woman with short black hair wearing a brown leather jacket, standing beside a modern café on a rainy street in Paris. Cinematic photography, medium shot, eye-level camera, soft morning light, realistic reflections on the wet pavement, shallow depth of field, warm color palette.The basic formula for writing AI image prompts
Subject + appearance + action or pose + environment + composition + lighting + visual style + restrictions
You do not always need every element. Use the parts that matter for your scene.
This guide explains how to write effective AI image prompts and AI video prompts, with practical formulas and examples you can reuse.
1. Describe the main subject
Begin with the most important person, product, animal, object, or location in the image. Avoid descriptions that are too general.
Weak: A beautiful woman.
Better: A young woman with shoulder-length black hair, a light beige suit, small gold earrings, and a confident expression.
Useful details include approximate age, hair color and hairstyle, clothing, facial expression, body position, accessories, distinctive features, and product materials. Do not add details that will not be visible in the final composition.
Every cinematic shot here started as a precise prompt. see the full prompt formula behind it.
2. Explain the action or pose
Tell the generator what the subject is doing. “A man in a coffee shop” is vague. “A man sitting beside the window of a coffee shop, holding a cup with both hands and looking thoughtfully at the rain outside” creates a clearer story and a more natural pose.
Use one clear primary action whenever possible, such as walking toward the camera, holding a product, opening a package, pouring a drink, or looking over one shoulder.
3. Define the environment
The environment provides context and atmosphere. Instead of “in a room,” try “in a minimalist living room with cream-colored walls, natural wood furniture, a large window, and soft morning sunlight.”
A useful environment description may include the location, time of day, weather, architecture, background objects, season, interior design, and number of background people.
4. Choose the composition and camera angle
Composition controls what viewers see and how the subject is framed.
Close-up: useful for emotion, beauty photography, food, jewelry, and product details.
Medium shot: useful for portraits, interviews, lifestyle images, and social content.
Full-body shot: useful for fashion, character design, and poses.
Wide shot: useful for landscapes, architecture, travel scenes, and establishing shots.
Overhead shot: useful for food, desks, product arrangements, and flat-lay photography.
Low-angle shot: makes a subject appear more powerful.
Eye-level shot: creates a familiar and realistic result.
For advertising, you can also reserve negative space: “Medium shot, subject positioned slightly to the right, with open space on the left for text.”
Composition decides what the viewer notices first and how the entire scene feels.
5. Describe the lighting
Lighting influences mood, realism, color, and depth. Useful descriptions include soft natural morning light, golden-hour sunlight, dramatic side lighting, diffused studio lighting, neon light at night, soft window light, strong backlighting, and overcast daylight.
Instead of writing “good lighting,” explain where the light comes from and how it should feel: “Soft golden sunlight enters from the left side, creating gentle shadows across the subject’s face.”
6. Select a visual style
Common options include photorealistic photography, cinematic film still, luxury product photography, fashion editorial, documentary photography, minimalist illustration, 3D animation, children’s book illustration, watercolor painting, retro advertising, and futuristic concept art.
Avoid combining unrelated styles. “Photorealistic luxury fashion photography with cinematic lighting” is clearer than mixing photorealism, watercolor, cartoon, documentary, and anime in one instruction.
7. Add colors and mood
You can specify a palette, such as “warm beige, gold, brown, and soft orange,” or describe the emotional atmosphere: calm and intimate, energetic and youthful, luxurious and elegant, dark and mysterious, bright and optimistic, or nostalgic and emotional.
For branded content, include the main brand colors while keeping the rest of the palette controlled.
Color sets the mood before the subject tells the story.
8. Include useful restrictions
Restrictions explain what should not appear. Examples include no text, no watermark, no additional people, no cluttered background, and no duplicated objects.
For products, ask the model to preserve the exact shape, packaging, label position, colors, and proportions. For portraits, request natural facial proportions and realistic hands.
Do not add a huge generic negative prompt. Mention only the problems relevant to the scene.
A complete AI image prompt example
A young woman with short black hair, wearing a brown leather jacket and dark sunglasses, sitting beside a large window in a modern café in Paris. She holds a white coffee cup and looks calmly at the rainy street outside. Cinematic photorealistic photography, medium shot, eye-level camera, shallow depth of field, soft natural morning light, realistic reflections on the window, warm brown and gold color palette, elegant and intimate atmosphere. Natural skin texture, realistic hands, no text, no watermark, no additional people.How to write AI video prompts
Writing AI video prompts requires everything used in an image prompt, plus motion. A video generator needs to understand what the subject does, how the subject moves, how the camera moves, what moves in the environment, how quickly the action happens, and how the shot begins and ends.
A video prompt should describe a short, filmable shot rather than an entire movie.
The basic formula for AI video prompts
Subject + main action + environment + camera movement + environmental movement + lighting and style + timing + ending + restrictions
1. Focus on one main action per shot
One of the most common mistakes is asking for too many events in a short clip. A prompt that asks a character to leave home, enter a taxi, reach an airport, board a plane, and walk through Paris describes an entire sequence, not one shot.
Break the idea into separate clips. Each generated clip should have one clear purpose.
Keeping a character consistent across scenes starts with writing prompts the right way.
2. Describe the subject’s movement precisely
Avoid vague directions such as “the woman moves naturally.” Explain exactly what happens: “The woman walks slowly toward the camera, adjusts the sleeve of her jacket, briefly looks into the lens, and gives a subtle smile.”
Keep movements physically simple. Too many simultaneous gestures can create unstable hands, faces, or body positions.
3. Define the camera movement
Static camera: useful for interviews, product shots, and controlled scenes.
Slow push-in: moves closer to the subject for a dramatic or premium feeling.
Pull-back: moves away to reveal more of the environment.
Tracking shot: follows a moving subject.
Pan: turns horizontally left or right.
Tilt: moves vertically up or down.
Orbit shot: moves around the subject.
Handheld shot: creates subtle documentary-style motion.
Drone shot: shows the scene from a high aerial perspective.
Use one main camera movement in a short clip. “The camera smoothly tracks backward while the woman walks toward it” is more reliable than stacking several movements.
Behind every shot in this campaign is a prompt built on a clear formula, not guesswork.
4. Add movement to the environment
Small background movements make AI-generated videos feel more believable: hair moving in the wind, steam rising from a cup, curtains moving in a breeze, rain falling, reflections changing on wet pavement, or vehicles passing in the background.
These details make the scene feel alive without adding another major action.
5. Describe the beginning and ending
Defining the first and last moments gives the video a clearer structure. For example: “The shot begins with a close-up of her hand around the coffee cup. The camera slowly pulls back to reveal her face. She turns toward the rainy window, and the shot ends with the camera holding on her thoughtful expression.”
6. Control speed and rhythm
Useful phrases include slow and controlled movement, natural walking speed, smooth cinematic motion, fast and energetic pacing, gentle slow motion, no sudden movements, and subtle natural motion.
For premium or cinematic videos, slower movement often produces more controlled results.
7. Preserve character and product consistency
For characters, specify that the same facial identity, hairstyle, clothing, accessories, and body proportions must remain consistent throughout the shot.
For products, preserve the exact shape, label, logo position, packaging colors, dimensions, and materials in every frame.
When possible, begin with a strong reference image and use image-to-video generation.
A complete AI video prompt example
A young woman with short black hair and a brown leather jacket walks slowly along a modern street in Paris at night. She moves toward the camera, briefly adjusts her sunglasses, looks into the lens, and gives a subtle smile. The camera smoothly tracks backward in front of her at eye level. Her hair and jacket move gently in the wind, blurred cars pass in the background, and neon lights reflect naturally across the wet pavement. Cinematic photorealistic style, shallow depth of field, soft blue and amber lighting, elegant and mysterious atmosphere. The shot begins as a medium full-body view and gradually moves into a medium close-up. All movements are smooth and physically natural. Preserve her facial identity, hairstyle, clothing, and accessories throughout the video. No sudden camera movement, no duplicate people, no changing clothes, and no text.Image-to-video prompt writing
When converting an existing image into a video, you do not need to describe every visible detail again. The image already provides the character, clothing, colors, and composition. Concentrate on the desired movement.
The woman slowly turns her head toward the camera and gives a subtle smile. Her hair moves gently in the wind. The camera makes a slow cinematic push-in. Preserve her exact face, clothing, background, colors, and composition. No sudden motion or visual changes.Common prompt-writing mistakes
Even detailed prompts can fail when instructions are vague, conflicting, overloaded, or overly ambitious. Understanding the most common prompt-writing mistakes helps you create clearer instructions and achieve more consistent results.
Using vague words
Words such as beautiful, cool, professional, and amazing are open to interpretation. Replace them with visible details. Instead of “make it look professional,” try “clean commercial photography, soft studio lighting, balanced composition, neutral background, and realistic product reflections.”
Replace vague praise with details the AI can actually visualize.
Adding conflicting instructions
Avoid combinations such as a static camera with fast camera movement, bright daylight in a completely dark room, a minimal background filled with detailed objects, or several unrelated visual styles.
Writing an unnecessarily long prompt
A longer prompt is not automatically better. Too many low-priority details can weaken the main idea. Put the most important information first: subject, action, environment, camera, lighting, style, and restrictions.
Changing too many things at once
When improving a prompt, change one major element at a time. Fix the subject and composition first, then adjust lighting, environment, camera movement, and restrictions.
Change one element at a time to understand what truly improves the result.
Asking for too much action
AI video models generally produce more stable clips when the action is simple and clearly sequenced. Divide complicated scenes into separate shots.
Reusable AI image prompt template
A [subject] with [appearance and clothing], [action or pose] in [environment]. [Image style], [shot type], [camera angle], [composition], [lighting], [color palette], and [mood]. [Important visual details]. No [unwanted elements].Reusable AI video prompt template
A [subject] with [appearance] in [environment], performing [main action]. The camera [camera movement]. [Environmental movements]. The shot begins with [starting composition] and ends with [ending composition or action]. [Visual style], [lighting], [mood], and [movement speed]. Preserve [important consistency details]. No [unwanted changes or movements].
A strong video prompt directs not only what appears, but also how the moment moves, unfolds, and ends.
Final prompt checklist
What is the main subject?
What does the subject look like?
What is the subject doing?
Where does the scene take place?
What should appear in the background?
What type of shot should be used?
Where is the camera positioned?
How should the camera move?
What lighting should be used?
What visual style should the result follow?
What mood or colors should dominate?
What should remain consistent?
What should not appear?
Create your next AI image or video
Effective prompt writing is not about using as many words as possible. It is about giving the AI a clear creative direction.
Begin with a simple scene, define the subject and action, add camera and lighting instructions, and improve one part at a time.
Once your prompt is ready, bring it to life by creating your next AI image or video with Zoviz Studio.
FAQ
How do you write good AI image prompts?
Start with the main subject and then add the subject’s appearance, action, environment, camera composition, lighting, visual style, and relevant restrictions. Use concrete visual descriptions instead of vague words.
How long should an AI image prompt be?
There is no perfect length. A useful prompt may be one detailed sentence or several short paragraphs. It should include enough information to define the scene without adding unnecessary or conflicting details.
What is a negative prompt?
A negative prompt describes unwanted elements, such as extra people, text, watermarks, incorrect anatomy, clutter, or changes to a product. Negative instructions should be relevant and limited rather than copied from an enormous generic list.
How are AI video prompts different from image prompts?
Image prompts describe a single visual moment. Video prompts must also describe subject movement, camera movement, background motion, timing, speed, and the beginning and ending of the shot.
Should every camera detail be included?
No. Include camera terminology only when it helps control the result. A simple instruction such as “medium shot with a slow push-in” is often more useful than a long list of technical filmmaking terms.
Is English required for AI prompts?
Not always. Many AI generators support several languages. However, prompts should use clear, direct, and visually specific language. Simple wording usually works better than complicated vocabulary.
Why does the AI ignore parts of my prompt?
The prompt may contain too many details, conflicting instructions, or several actions with equal importance. Move the most important instruction closer to the beginning and simplify the rest of the scene.
