When working with AI image or video generation, your prompt is your creative blueprint. The more detailed and intentional you are, the better the results you’ll get. Think of your prompt as giving instructions to a highly skilled Artist. The clearer and more specific you are, the closer the AI will get to your vision. If you only say "draw a cat", you might get any kind of cat in any style. But if you say "a fluffy white Persian cat sitting on a velvet chair in golden sunlight", you guide the AI toward your exact vision.
Getting started with writing prompts
Before you write a prompt, it helps to think of it as a formula. Each detail you include—placed in the right order—makes it easier for the AI to understand and generate the result you want.
Step 1: Describe the subject: What you want to see as the central element of the image or video
Step 2: Describe the style: Specify how you want the final image or video to look. E.g Cinematic, moody lighting, soft shadows, shallow depth of field, high contrast
Step 3: Describe the details: Use vivid adjectives and consistent descriptions to refine your subject. Describe key traits like appearance, behavior, and personality to maintain visual consistency.
Step 4: Set the Context: Explain the setting, purpose, and mood of the image and video for extra clarity.
Step 5: Describe the sound: Complete your creative vision by defining the audio atmosphere. (If applicable)
In short, a vague prompt produces vague results. Without enough guidance, the AI lacks direction and will generate generic or random outputs that may not match your vision.
Take this example:
Instead of: Dog running
Write: A golden retriever trotting through a sunlit meadow, filmed in slow motion, with warm evening light and a soft-focus background, accompanied by gentle acoustic guitar and soft ambient nature sounds (birds and rustling grass).
Check out our blog to learn more.
💡 Pro Tip: Test Before You Spend Credits
Before committing credits, use a GPT to test or refine your prompt.
Simply describe what you’re aiming for: including the model, duration, and desired effect , and ask the GPT how to improve it. It often takes a few iterations to get the perfect result, so thoughtful prompt tweaks upfront can help you get there faster, without using more credits than you need.
Example:
"Create a 12-second Sora 2 prompt for a cinematic scene where a man in a white shirt, tie, and glasses begins walking through a field, then starts hovering and finally flies. The camera should track him as he moves through the air.”
Tips for writing better prompts
1. Think like a director
-
Frame the scene with intention
- For people: Use mid-shots (waist-up) or bust shots (chest-up) to reduce leg and foot errors.
- For animals: Show them in natural positions—lying down, sitting, or standing still.
- Avoid ultra-wide full-body shots unless necessary, as they increase the chance of unnatural anatomy.
- Use lighting to set the mood and define the emotion and realism
- Use words such as ‘soft glow’, ‘backlit silhouette’, ‘dramatic shadows’ to give your scene depth and consistency.
- Define the camera perspective to shape how the view connects with the scene
- For images use words such as ‘bird’s-eye’, ‘first-person’, ‘tracking shot’
- For videos use words such as ‘slow-motion’, ‘dynamic movement’, ‘zooming’.
💡Pro tip: Keep movements simple and subtle: Complex motion often creates visual glitches
- Avoid: backflips, complicated dance moves, or full-body acrobatics.
- For animals: Stick to natural actions like walking, sitting, or turning their head.
- For humans: Use small, realistic gestures like looking over a shoulder, shifting weight, or lightly waving a hand.
Why it works: AI handles smooth, minimal actions more reliably than chaotic or highly detailed movement, so your output looks cleaner and more realistic.
Prompt examples:
Image Prompt: “A close-up of a woman’s face dressed in a flowing, white cloak walking towards the viewer. in the background, sleek, towering structures made of smooth, light-colored materials, reflecting ambient light. The sky is muted colors of orange, pink, and purple, with soft, fluffy clouds. The lighting is ethereal, casting a warm glow on the scene and her face, enhancing the calm atmosphere. The overall style is ultra-modern and minimalist, with an emphasis on clean lines and a serene, almost otherworldly aesthetic. 50mm lens”
Video Prompt: “Slow camera tracking shot back, the subject walking towards the camera, while the background elements subtly sway and a gentle breeze rustles the subject’s fabric"
Video Prompt: "Create a whimsical 3D scene with smooth, polished shapes and vibrant colors. The characters are expressive and full of life, with soft lighting and rich textures adding warmth and depth. The composition feels playful and heartwarming. Rendered in a vibrant, whimsical, and expressive 3D animated style with exaggerated proportions, soft textures, and rich colors, reminiscent of beloved animated films. The vibrant and heartwarming 3D render style, featuring expressive characters, rich colors, and soft, inviting textures, captures the scene’s charm and emotional depth, evoking the warmth of classic animated films"
2. Avoid blocked or covered poses
AI sometimes struggles when body parts overlap in unnatural ways.
- Keep arms away from crossing over the torso.
- Avoid hands covering the face.
- Make sure legs and feet aren’t hidden behind objects when possible.
Think of it like giving the AI a clear view of your subject so it can “read” the anatomy correctly.
Prompt examples:
Image prompt:"A full-body portrait of a ballet dancer in mid-pose, arms gracefully extended outward, face clearly visible, legs fully shown with pointed toes, photographed in a bright studio with a clean white background, cinematic lighting"
Video prompt: "A cinematic slow-motion video of a fashion model walking confidently on a runway, arms relaxed at her sides, face unobstructed and smiling, full body visible from head to toe, captured with a steady tracking shot, warm spotlight on the subject"
3. Keep clips short and sweet
The longer your video, the more likely AI is to introduce inconsistencies—like jittery motion, changes in clothing, or warped limbs.
- Aim for short, clean clips.
You can loop or stitch them together later in post-production for a seamless final product
4. Keep it clear
Stick to one idea or theme. If necessary, split big ideas into separate prompts.
Instead of: A futuristic city, a medieval castle, and a spaceship in a stormy jungle.
Write: ‘A futuristic city at night’ and ‘A medieval castle in the jungle’.
5. Write in present tense with active verbs
The AI interprets actions more effectively when written as if they’re happening now.
Instead of: A person walked through a forest
Write: A person walks through a forest, dappled sunlight streaming through the trees
This approach works especially well for movement and animation prompts.
6. Be cautious of using a prompt to add text to an image or video
AI models don’t actually write text; they learn the visual patterns of letters from training data, but don’t store spelling or language rules. This often results in pseudo-letters or gibberish that may appear similar to letters but aren’t real words. Here are some recommendations when you are using text:
- Keep the word short (1–3 letters work better than full sentences)
- Ask for the text to be on a sign, card, or poster rather than floating
- Specify a common font style (“blocky sans serif” or “typewriter font”)
💡 Pro -Tip: Add any on-screen text using your preferred image or video editor, as the generator is optimized for visual concepts rather than accurately written text.
Artlist’s AI Auto Prompt Enhancer
Artlist’s AI Image Prompt Enhancer takes your ideas and refines them into a richer, more descriptive prompt, perfect for getting more consistent, visually stunning results. All you need to do is write your basic prompt, click Enhance, and let the AI expand it into a more vivid description.
Example:
Original prompt:
A red tomato in half on a wooden chopping board.
AI-enhanced prompt:
Create a photorealistic macro shot of a red tomato sliced in half, displayed prominently on a wooden chopping board. The image should capture the vivid, juicy texture of the tomato, with every seed and droplet of juice emphasized. The background should be softly blurred, creating a strong depth of field that brings attention to the intricate details of the tomato and the rustic charm of the chopping board.
Explore more on AI-models
Writing better prompts for smoother image blends
Need some more help? Head back to our Help Center.