🎬 AI Creative Cheat Sheet: Video Prompts, But Way Smarter
A New Dimension with Veo 3.1 and Kling AI
Here’s a natural and engaging English translation of your paragraph, keeping the tone consistent with the image edition intro: In the last installment, we explored AI tools that turn a single line of text into fully generated images. This time, we’re going beyond still visuals — into the world of motion and storytelling. 🔥
In this “Video Edition”, we’ll dive into two of the leading AI video generation tools: Veo 3.1 and Kling AI, and explore how AI is unlocking a whole new realm of creative possibilities in video. 🎥
Want to explore image generation tools first?
👉 🏞️ AI Creative Cheat Sheet — Image Edition (Click!)
📹 Video Generation Services
Ready to go beyond static visuals? Let’s explore the world of motion with Veo 3.1 and Kling AI — two powerful tools that turn text into fully animated videos!
Both models support text-to-video generation, but each is built on a different technical foundation and creative philosophy. That means even with the same prompt, the output can vary wildly in tone, detail, and expression.
So where does each model shine? And which one is right for your next project? Let’s find out together. 🎥✨
1️⃣ Veo 3.1
🎞️ What is Veo 3.1?
Veo 3.1 takes video generation to the next level with improved audio generation and enhanced consistency with reference images. This means it can deliver more natural character motion and reflect prompts with greater accuracy.
It also supports advanced editing capabilities, allowing you to create videos longer than one minute — and it can quickly generate high-quality clips up to 8 seconds in full HD (1080p) resolution.
You can access Veo 3.1 through your Google account using Gemini, Flow, AI Studio, or supported third-party platforms. Some features are free to try, but ongoing video creation will require an API key or a paid subscription plan.
🎞️ What can you do with Veo 3.1?
(1) Reference Image-Based Generation (“Ingredients to Video”)
Maintain consistent character design and visual style, while incorporating specific objects or scene aesthetics into the video.
(2) Scene Extension
Use up to three reference images to expand a video beyond 1 minute. Veo 3.1 connects scenes by using the last second of the previous clip, ensuring both visual and audio continuity.
(3) First/Last Frame-Based Generation
Generate smooth transitions and realistic motion by providing the first and last frames. Veo creates seamless visual flow and accompanying audio between the two.
🎞️ Tips for Writing Prompts in Veo 3.1
✅ Principles
Effective prompts are descriptive and clear. Understanding the core idea, adjusting keywords, and incorporating video-related terminology will help you get the most out of Veo.
Get Riley’s stories in your inbox
Join Medium for free to get updates from this writer.
✅ Prompt Structure
- Subject: Briefly specify the object, person, animal, or landscape to appear in the video. If needed, list multiple subjects in numbered order.
(e.g., “Subject: A city street on a rainy night, a woman holding a yellow umbrella”) - Action: Write a short sentence about each subject’s single action, along with basic direction or speed.
(e.g., “Action: The woman walks slowly along the sidewalk and glances left once”) - Style: Use keywords from films or animations to define the visual tone.
(e.g., “Style: film noir, high contrast, light film grain”) - Camera Position & Motion: Describe the camera’s height, angle, and movement in one line.
(e.g., “Camera: eye-level, slow pan from right to left”) - Framing: Summarize the shot composition in a single term.
(e.g., “Framing: medium shot, single shot”) - Focus & Lens Effects: Define the depth of field and lens characteristics.
(e.g., “Focus/Lens: shallow depth of field, 50mm standard lens”) - Mood: Indicate the tone using color, lighting, and time of day.
(e.g., “Mood: cool blue tones, nighttime, wet road reflections”) - Dialogue: Only include when necessary, using quotation marks.
(e.g., “Dialogue: ‘The rain will stop soon.’”) - Sound Effects (SFX): Clearly specify particular sounds.
(e.g., “SFX: raindrops on umbrella, splashing from a passing car in the distance”) - Ambient Sound: Describe the background noise of the setting in a full sentence.
(e.g., “Ambient Sound: light rainfall mixed with faint city chatter”)
🎞️ Things to Keep in Mind When Using Veo 3.1
- Request Latency: Video generation time can range from a minimum of 11 seconds to up to 6 minutes during peak hours.
- Regional Restrictions for People Generation: In the EU, UK, Switzerland, and MENA, Veo 3.1 cannot generate videos that include children.
- Video Storage: Generated videos are stored on the server for 2 days and automatically deleted afterward. If you want to keep a local copy, download the video within 2 days. Extended clips are treated as newly generated videos.
- Watermarking: All videos generated with Veo include a watermark identifying them as AI-generated content using SynthID, which can also be used to verify the video’s authenticity.
- Safety & Copyright: Generated videos go through safety filters and memory checks designed to mitigate risks related to privacy, copyright, and bias.
- Audio Errors: Veo 3.1 may block video generation due to safety filters or other audio processing issues. No charges will apply if the generation is blocked.
2️⃣ Kling AI
📽️ What is Kling AI?
Kling AI offers dramatically improved physics-based motion and prompt responsiveness, enabling highly natural character movement and consistent video quality that accurately reflects user instructions. It also excels at producing cinematic visuals and nuanced details — all while keeping costs lower than many competitors. With support for high-quality 1080p video, it’s a solid choice for both professional creators and casual users alike.
You can access Kling AI through a Kling AI membership, via API integration, or through third-party apps.
📽️ What Can You Do with Kling AI?
📽️ Kling AI Capabilities
(1) 2.5 Turbo Model
- Audio Generation: Automatically generates voices and sound effects to accompany the video
- Enhanced Prompt Understanding: Interprets user text instructions with greater accuracy
(2) Motion Brush
- Supported in: KLING AI 1.5 model
- Function: Allows users to draw movement paths directly on specific elements within the image to control how they move
(3) Add Elements
- Supported in: KLING AI 1.6 model
- Function: Upload up to four image elements (e.g., characters, objects, backgrounds) to create a consistent video series
(4) Extend with Prompts
- Supported in: Pre–2.5 Turbo KLING AI models
- Function: Extend existing videos using additional prompts to generate longer clips
📽️ Kling AI Prompt Writing Tips
1) Text to Video
Subject (with Description) + Subject’s Action + Scene (with Description) + (Camera Language + Lighting + Atmosphere)
- Subject: The main focus of the video — this can be a person, animal, plant, object, etc.
(e.g., “A young girl walks through a field of sunflowers.”) - Subject Description: Describe the subject’s appearance, outfit, and posture in short, specific phrases.
(e.g., “An athlete with a muscular build, shoulder-length brown hair, white tracksuit, and a determined expression.”) - Movement: Describe the motion or stillness of the subject, designed to fit roughly 5 seconds of action.
(e.g., “The person slowly turns their head and smiles.”) - Scene: Depict the setting where the subject is located, including foreground, background, and supporting elements.
(e.g., “A lone tree stands on a green meadow, gently swaying in the breeze.”) - Scene Description: Convey the mood and details of the environment to form a visually fitting image for a ~5-second video.
(e.g., “A cup of coffee sits on a desk under warm indoor lighting.”) - Camera Language: Use cinematic shot types and lens techniques to shape visual messaging and emotional tone.
(e.g., close-up of the character, ultra-wide shot, low-angle perspective, background bokeh effect.) - Lighting: Use light and shadow to add depth and emotion to the scene.
(e.g., “The setting sun gently wraps around the subject’s face,” or “Tyndall light streams through the dusty air.”) - Atmosphere: Define the emotional tone and sensory impression of the video using color, lighting, and sound cues.
(e.g., “A calm and warm afternoon,” or “A quiet, sentimental mood.”)
✅ Prompt Structuring Guidelines
- Basic Structure: Build sentences around the subject, movement, and setting.
(e.g., “A young man runs along a sunset-lit beach.”) - Detailed Descriptions: Use precise adjectives and adverbs for visual clarity.
(e.g., “Under the warm sunset glow, his hair flows softly in the breeze.”) - Contextual Info: Briefly add the intended use or setting for the video.
(e.g., “For a branding campaign scene.”) - Style References: Include specific artistic styles or moods.
(e.g., “A minimalist composition with Bauhaus sensibility.”) - Model Interpretation: Let your description guide the AI’s visual interpretation.
(e.g., “Compose the scene with a cinematic tone, focusing on the character.”) - Creative Freedom: Don’t be afraid to express bold or surreal ideas.
(e.g., “A dreamlike moment where time stands still.”)
2) Image to Video
Subject + Background + Movement
- Subject: The main focus of the video, which can be a person, animal, plant, or object.
- Movement: Briefly describe whether the subject is still or in motion.
- Background: Describes the environment including the foreground, background, and other elements where the subject is placed.
✅ Sentence Construction Method
- Subject: In detail, consider visual characteristics such as appearance, facial expression, posture, clothing, and props. Also, reflecting the subject’s cultural or situational context and symbolic elements can enhance the meaning and mood of the video.
- Movement: Compose the prompt around the intended movement in the image. In detail, consider the speed, direction, repetition pattern, and the positioning of the subject’s gaze, hands, and feet. Also, design how the movement contributes to the atmosphere and emotional delivery of the scene.
📽️ Things to Keep in Mind When Using Kling AI
- Use simple expressions: Opt for short, straightforward words and sentences instead of complex phrasing.
- Keep scenes simple: Design visually minimal scenes that can be completed within 5–10 seconds.
- For East Asian styling: Using terms like “oriental atmosphere” or “Asian” tends to generate more accurate Chinese-style characters and backgrounds.
- Limitations with numbers: The current model struggles with quantity recognition and may not maintain numerical consistency.
- Split-screen composition: When expressing multiple scenes simultaneously, provide a clearly separated structure.
- Restrictions with complex motion: Physically complicated actions — like a bouncing ball or an object thrown from height — are difficult to render with the current model.
⚡️ A New Future, Powered by AI
A prompt isn’t just a line of text — it’s the key to building new worlds. As our interactions with AI evolve beyond simple Q&A into true creative collaboration, we hope this guide marks the beginning of your own AI journey.
At the forefront of this rapidly advancing technology, Supercent is integrating refined prompt strategies and advanced creative processes into every part of our work — from game development to publishing and marketing.
Join us on this fast-moving creative journey, fueled by AI and driven by imagination. The future starts now. 🚀
Breaking creative boundaries with AI — feels exciting, doesn’t it?
Supercent is looking for teammates who are ready to turn imagination into reality with AI.
[Contact]
- E-mail: help@supercent.io
- Website: supercent.io
