.png?fit=max&auto=format&n=J3TfNmZhqEoKcaaO&q=85&s=468b5adb026aa33181cc81ab54ab68db)
Be specific and direct Describe the action, setting, style, and motion in clear terms. Avoid vague instructions that don’t guide the output. ✅ “Woman walking through modern office with soft warm lighting, smooth tracking camera, professional mood” ❌ “Make a video of a person in an office”
Describe motion clearly Camera movement and action pacing are critical. Be explicit about how things move. ✅ “Camera slowly zooms in on the product, showing all angles with smooth 360° rotation” ✅ “Person walks confidently down hallway, camera tracks alongside smoothly” ❌ “Show the product moving”
Refine with focused instructions When improving results, change one element at a time (motion, lighting, pacing, or style). This helps the AI apply updates more accurately. ✅ “Increase the camera movement speed” ✅ “Add warmer golden hour lighting” ✅ “Slow down the motion for a more cinematic feel” ❌ “Make it better, faster, and more dramatic”
Choose the right tool for the job
- Video Generation — Create new videos from text or image reference
- Video Remix — Adjust lighting, pacing, colors, or mood on existing videos
- Image-to-Video — Start with a reference image for consistent character/object
- Canvas — Crop or reposition video in the timeline
Only include actionable details Focus on elements the AI can control: action, setting, lighting, camera movement, pacing, mood, color grading, and audio. Skip commentary that doesn’t change the output. ✅ “Soft natural window lighting from the left, warm color tone” ❌ “Make it look cinematic” (vague — not actionable)
Keep the structure simple [Action] in [Setting] · [Lighting] · [Camera Movement] · [Mood] · [Duration] [Aspect] [Resolution] Examples:
Optimize for forward progress Use prompts that move toward your goal rather than asking for review or general improvement. ✅ “Slow down the walking pace for a more deliberate movement” ✅ “Add more dramatic shadows in the lighting” ✅ “Increase camera zoom speed for more impact” ✅ “Make transitions smoother and more gradual” ❌ “Does this look good?” ❌ “Can you make it better?” ❌ “I don’t like it”
Match duration to content type Different video lengths serve different purposes. Choose intentionally.
- 4-5s — Quick hooks, ads, clips
- 6-8s — Social media, product demos, stories
- 10-12s — Narrative content, tutorials, storytelling
Set resolution strategically Start with appropriate resolution, not always maximum. ✅ 720p for iterations — Fast, affordable (120-600 credits depending on model) ✅ 1080p for final export — Premium quality, share-ready (600-3,200 credits) ❌ Don’t upgrade resolution until composition is final
Consider model choice before prompting Different models excel at different styles. Match your prompt to the model’s strengths.
- Wan 2.5 — Fast iterations (use simple prompts)
- Seedance Pro — Flexible durations & aspect ratios
- Veo 3.1 Fast — Quality with speed balance
- Kling 2.6 Pro — Cinematic visuals (detailed prompts work better)
- Veo 3.1 — Premium results (reward detailed prompts)
Be specific about camera movement Camera work is what makes videos feel professional. Don’t leave it implied. ✅ “Camera slowly orbits around the product from left to right” ✅ “Static camera showing the full scene, subject moves toward camera” ✅ “Tracking camera smoothly following the person as they walk” ❌ “Dynamic camera” (too vague)
Specify lighting for mood Lighting sets the entire tone. Be explicit. ✅ “Soft morning light from the left window, warm golden tones, subtle shadows” ✅ “Studio lighting with key light and fill, professional, crisp” ✅ “Dramatic backlighting with shadows, moody and cinematic” ❌ “Good lighting”
Use Video Remix for these common refinements Instead of regenerating, use Video Remix when:
- Lighting feels too bright/dark → Adjust exposure and color grading
- Motion feels too fast/slow → Adjust pacing
- Colors don’t match brand → Change color palette
- Background needs adjustment → Modify or replace
Common Mistakes to Avoid ❌ Too much detail in one prompt — Focus on the essential elements ❌ Conflicting motion descriptions — “Fast action” + “slow smooth motion” confuses the AI ❌ Ignoring technical specs — Always include duration, aspect ratio, and resolution ❌ Vague lighting descriptions — “Good lighting” doesn’t guide the output ❌ Regenerating instead of remixing — Wastes credits on major reworks ❌ Not matching aspect ratio to platform — 16:9 for web, 9:16 for mobile
Prompt Refinement Workflow
- Start simple — Basic action, setting, mood
- Generate with budget model — Wan or Seedance (test prompt)
- Review — Does it match your intent?
- Use Remix for tweaks — Adjust lighting, pacing, mood
- Upgrade model if needed — Better quality for final version
- Export at 1080p — Only when satisfied with composition
Example: Iterating a Product Video Iteration 1 (Budget test - Wan 2.5, 720p, 50 credits): “Wireless headphones rotating on white background · studio lighting · 360° orbit · 5s · 1:1 · 720p” Review: Rotation is good, but lighting feels flat. Iteration 2 (Video Remix - 40 credits): “Add more dramatic lighting with soft shadows, increase reflection on the product” Iteration 3 (Final - Veo 3.1 Fast, 1080p, 1,200 credits): Use the refined concept with upgraded model for premium quality Total cost: 50 + 40 + 1,200 = 1,290 credits vs. 3,200+ for multiple full regenerations
Pro Tips
- Test prompts with budget models first — Refine concept before upgrading quality
- Use Video Remix extensively — 5-10 remix iterations vs. 1-2 full regenerations
- Finalize composition at 720p — Upgrade to 1080p only for final export
- Be consistent with terminology — Use same words across iterations for clarity
- Build prompts incrementally — Start basic, add detail as you refine
- Match prompt detail to model quality — Budget models: simple prompts; Premium models: detailed prompts