Wan 2.1 Prompting: Best Practices

Here’s a detailed breakdown of best practices for Text-to-Video (T2V) and Image-to-Video (I2V) prompting in Wan 2.1.

🔹 Text-to-Video (T2V) Prompting

When generating video from text alone, the AI needs clear guidance on scene setup, motion, composition, and style.

1️⃣ Use a Clear and Structured Format

A well-structured prompt ensures predictable results. Use the following format:

[Main Subject] + [Action] + [Scene Details] + [Lighting] + [Camera Angles] + [Style]

✅ Example:

"A futuristic samurai dashes forward in slow motion, his glowing katana slicing through the rain. Neon cityscape in the background, reflecting vibrant lights. The scene is cinematic, captured in a dynamic wide-angle shot with a slight camera tilt."

2️⃣ Always Describe Motion Clearly

Since T2V AI generates movement from scratch, specifying actions ensures smooth, intentional motion.

✅ Good Motion Descriptions:

• "A golden retriever runs joyfully through a sunlit field, ears flapping in the wind."

• "A motorcyclist drifts around a sharp curve, tires kicking up dust."

• "A spaceship glides silently through a nebula, glowing particles trailing behind."

🚫 Avoid Static Descriptions:

• "A man in a field." (No action or details)

• "A city at night." (Too generic, lacks movement)

3️⃣ Specify Camera Movements for Cinematic Shots

To create more dynamic results, use filmmaking techniques.

🎥 Examples of Camera Movement Keywords:

• Tracking Shot: "The camera follows a runner sprinting across the beach, sand flying with every step."

• Slow Zoom: "A close-up of a detective lighting a cigarette, slowly zooming into his intense gaze."

• Handheld Effect: "A soldier navigates a war zone, the shaky cam mimicking intense action."

• Drone View: "A vast desert landscape seen from a high drone perspective, slowly panning across the dunes."

✅ Example Prompt with Camera Motion:

"A cyberpunk hacker typing rapidly on a holographic keyboard. The camera slowly circles around her, revealing neon screens filled with encrypted code."

4️⃣ Enhance Visual Style & Lighting

Wan 2.1 adapts style based on your descriptions. If you don’t specify, it may default to a generic look.

🖌 Style Keywords:

• Cinematic (Hollywood-style, film grain, deep contrast)

• Anime (vibrant colors, exaggerated expressions, cel-shading)

• Realistic Documentary (handheld, natural lighting, raw textures)

• Dreamlike (soft focus, ethereal glow, surreal movement)

🔦 Lighting Keywords:

• Golden Hour: "A warrior walks through a misty battlefield at dawn, the golden light reflecting on his armor."

• Neon Glow: "A detective walks under flickering streetlights in a noir-inspired cyberpunk city."

• Low Light with Flashlight Effect: "A first-person view of someone exploring an abandoned house, only a flickering flashlight illuminating the darkness."

✅ Example Prompt Combining Everything:

"A masked vigilante stands on a rooftop at midnight, overlooking a futuristic Tokyo. The rain-soaked city glows with neon reflections. The camera slowly zooms in, emphasizing his glowing cybernetic eye. Cinematic, film noir style."

🔹 Image-to-Video (I2V) Prompting

For I2V, the AI animates an existing image, meaning you focus more on movement and transformation rather than scene setup.

1️⃣ Focus on Motion of Existing Elements

Instead of describing a scene, focus on what should move.

✅ Good I2V Motion Prompts:

• "The character's cape flutters in the wind while sparks fly from their sword."

• "The neon signs in the background flicker rhythmically as cars pass by."

• "The dragon's wings slowly expand, revealing intricate glowing patterns."

🚫 Avoid Generic Prompts:

• "A futuristic warrior." (No movement specified)

• "A car in a city." (What is the car doing?)

2️⃣ Specify the Type of Animation

• Looping Animation → "The candle flame flickers gently in a seamless loop."

• Subtle Motion → "The character blinks slowly while the background remains still."

• Extreme Action → "The mech suit suddenly activates, its arms extending outward with hydraulic force."

• Environmental Effects → "Leaves drift through the air as the wind sways the grass."

✅ Example Prompt for a Portrait Image:

"The astronaut in the image waves slightly while distant stars twinkle in the background. A soft parallax effect adds depth to the scene."

3️⃣ Use Camera Movement to Add Depth

Since the image is static, camera effects create the illusion of movement.

🎥 Examples of I2V Camera Effects:

• Parallax Effect → "Foreground objects move slightly while the background stays in place."

• Subtle Camera Shake → "A warrior stands still, but the camera shakes slightly as if it's a handheld shot."

• Slow Zoom → "A gentle push-in effect towards the subject's face, creating dramatic tension."

✅ Example I2V Prompt for a Landscape Image:

"The clouds in the sky drift slowly while the camera pans gently to the right, creating a parallax effect between the mountains and foreground trees."

4️⃣ Animate Objects & Textures, Not Just the Subject

Instead of only moving the character, add motion to background and environment for realism.

✅ Examples of Background Motion:

• "The water ripples softly, reflecting the sunset."

• "The subway lights flicker as the train moves in the background."

• "Fireflies glow and drift around the scene, adding a dreamlike effect."

🚫 Avoid One-Dimensional Prompts:

• "The knight moves." (How? What else moves?)

• "The scene changes." (Changes how? Be specific!)

✅ Example I2V Prompt Enhancing Background Motion:

"The wind gently moves the character's scarf while distant city lights shimmer. The background rain falls continuously, giving a moody cyberpunk vibe."