Prompt Guides
Master the art of AI generation with our comprehensive guides
Image Generation Guide
Creating perfect AI images is about understanding how to structure your prompts. Follow this guide to craft effective prompts that bring your vision to life.
Important: The examples below are just guidelines - you can describe anything using simple, natural words!
Step-by-Step Prompt Structure
Define the Subject
Start by specifying the main focus and character count. This helps the model understand composition.
Examples (you can use any number combination):
Note: When requesting multiple characters (2girls, 2boys, etc.), AI may sometimes duplicate the same character instead of creating distinct individuals.
Describe Details
Add appearance, actions, and attire details using clear, simple tags.
Examples (use simple descriptive words):
Tip: You can describe any detail you imagine using simple words!
Set the Background
Specify the setting and environment for your scene.
Examples (describe any location you want):
You can specify any environment - indoor, outdoor, real, or fantasy!
Add Quality & Style Tags
Enhance output quality with these recommended tags.
Quality tags (these improve overall output):
For realism (add these for photorealistic style):
Camera Angles & Lighting
Camera Angles
Add after quality tags for camera perspective:
Lighting Options
Add after angle tags for lighting mood:
Understanding Negative Prompts
Negative prompts tell the AI what you don't want in your image. They can help reduce unwanted features, but they're not a guarantee.
Important Limitations
- •Not guaranteed: Negative prompts sometimes help, but they don't always work perfectly
- •Positive is stronger: Positive tags have more influence than negative ones. If something is in your positive prompt, it's hard to remove with negative prompts
- •Use as refinement: Think of negatives as fine-tuning, not as absolute rules
How to Use Negative Prompts
If you notice unwanted elements appearing in your generations, add them to negative prompts to reduce their occurrence.
Problem: Asking for 1 girl but getting 2 girls
Problem: Getting glasses when you don't want them
Problem: Getting blue hair instead of other colors
Problem: Character sitting when you want standing
Problem: Low quality or distorted results
Best Practice
Tip: Instead of adding many negative tags, focus on getting your positive prompt right first. Use negative prompts only for persistent unwanted elements that keep appearing.
Example Prompts
Example 1Simple Character
Example 2Action Scene
Example 3Multiple Characters
Note: AI models sometimes duplicate the same character instead of creating distinct individuals. This is a current limitation of the technology.
Example 4Fantasy Warrior
Example 5Magical Library Scene
Example 6Underwater Adventure
Best Practices
- Avoid conflicting composition tags like close-up and wide shot together
- Use recommended composition tags: upper body, portrait, full body
- Keep lighting consistent between character and background
- Start with minimal parameters and add details gradually
Video Generation Guide
Video generation requires a different approach than images. Focus on describing movement and camera motion rather than static details. The system automatically handles scene details based on your character.
Important: Ask for Actions Present in the Image
For best results, describe actions or movements that match what's already visible in your starting image.
Example:
If you want a character to ride a horse, first generate an image with the character already on a horse. Then for video, use a prompt like: "Girl is riding a horse."
Three Golden Rules
Subject Movement is Essential
Always describe what the subject is doing. Camera motion and details are optional, but movement is crucial.
Keep Prompts Short & Precise
Don't describe scene details - they're handled automatically. Instead of:
long-hair blonde woman with green eyes moving from side to side
Simply write:
moving from side to side
Use Slower Movements
Fast actions like dancing or parkour may cause artifacts. Gentle, slow movements work best.
Video Prompt Structure
💡 Use natural language and simple words to describe any movement or camera action you imagine!
Camera Movements
Works Well
These camera movements generally produce good results:
- ✓Camera pushes in / pulls out
Use "camera pushes in" instead of "zooms in" - push in works great, pull out requires careful structure
- ✓Camera pulls back
Perfect for revealing scenes progressively, works very well
- ✓Camera follows / tracking shot
Following subject motion works well when explicitly described
- ✓Camera orbits / circles around
Circular motion around subject creates dynamic shots
- ✓Camera tilts up / tilts down
Vertical camera movement - may need prompt refinement
- ✓Camera slowly zooms out
Slow zooms work better than fast ones
Limited Support
These movements may have issues or inconsistent results:
- ✗Whip pan / fast pan
Too fast - rapid horizontal movement not supported
- ✗Crash zoom / rapid zoom
Fast zooms cause artifacts and poor results
- ⚠Camera pans left / pans right
Works sometimes but doesn't always respect the specified direction
- ⚠Camera roll / 360 rotation
Possible but needs multiple attempts and refinement
💡 These are just common examples - feel free to try any camera movement you can describe with simple words!
Example Video Prompts
Prompt:
Moving elegantly from side to side, head tilts to the side, camera pushes in on her slightly
Prompt:
Smiling and tilting head, camera pushes in on her face
Prompt:
Camera pulls out, background turns into fireworks
Pro Tips for Better Videos
- Focus on gentle, deliberate movements rather than fast action
- Describe camera movements explicitly using simple words like "camera pushes in" or "camera follows"
- Combine subject movement with camera motion for dynamic results
- Experiment with different prompts - results vary by character style
- Avoid describing appearance - let the character model handle that
- Remember: These examples are guidelines - feel free to describe any movement using simple, natural words!
About Video Generation Quality
We use the latest AI video models available, but video generation is still evolving technology. Videos may sometimes be corrupted, not match your description perfectly, or have unexpected artifacts. This is part of learning how to prompt for current video model.
Feel free to experiment! Don't be discouraged if results aren't perfect on the first try. Video AI is improving rapidly, and experimentation helps you learn what works best.
💡 Testing & Energy: If you're experimenting and run low on energy credits, contact us and we'll add more so you can continue testing what works and what doesn't!
Quick Navigation
Quick Tip
Experiment and iterate! AI generation is an art. Don't be afraid to try different combinations and see what works best for your vision.