Prompt Writing Secrets: Making AI Understand Your Creativity Precisely

Overview

Mastering the art of prompt writing is key to unlocking Nano Banana's full potential. An excellent prompt is like a clear design brief, guiding the AI to precisely output the results you want; vague requirements can only lead to unsatisfactory results. This guide will reveal prompt writing secrets from basic to advanced, helping you comprehensively improve your "AI-oriented thinking" capabilities.

I. Core Principle: Communicate Like Describing a Scene

Nano Banana's core advantage lies in its powerful natural language understanding capabilities. Therefore, the most critical principle is:

Use a narrative, descriptive paragraph to depict the entire scene, rather than simply listing keywords.

For example, instead of using "elderly, potter, close-up, realistic," describe it this way: "A photorealistic close-up portrait of an elderly Japanese potter...". This approach produces more coherent, higher-quality images.

II. Golden Rules & Structure of Prompts

A well-structured prompt can greatly improve communication efficiency. You can refer to the following "golden rules" to construct your instructions:

1. Be Hyper-Specific, Avoid Vagueness

The more detailed your instructions, the more control you have over the results.

  • Counter-example: "Generate futuristic armor."
  • Good example: "Generate an ornate elven plate armor etched with silver leaf patterns, with a high collar and shoulder guards shaped like falcon wings."

2. Combine Chinese and English for Precision and Context

This is a very practical technique, especially suitable for Nano Banana.

  • Use English for technical terms: Such as lighting, materials, styles, etc., to ensure professional terminology accuracy. For example, enhance lighting and skin tone, smooth blemishes naturally, apply light film grain.
  • Use Chinese for contextual guidance: Describe scene atmosphere, character emotions, or story background to help the AI better understand the context.

3. Iterate and Refine, Avoid Stacking

Don't expect perfect results from inputting all instructions at once. AI is better at handling step-by-step tasks.

  • Build the basic scene first: "First, create a peaceful, misty forest background at dawn."
  • Then gradually add elements: "Next, add an ancient stone altar covered with moss in the foreground. Finally, place a glowing sword on the altar."
  • Make fine adjustments: After getting preliminary results, you can make modifications through conversation, such as "Good, but can you make the lighting warmer?"

4. Provide Context and Intent

Explaining the purpose of the image to the AI can help it better understand your needs, thereby influencing the final output.

  • Counter-example: "Create a logo."
  • Good example: "Create a logo for a high-end, minimalist skincare brand."

III. Advanced Professional Prompt Techniques

1. Use Photography and Cinematography Terms to Control Composition

To achieve professional-level visual effects, use professional language.

  • Camera and Lens: Clearly specify photorealistic, close-up portrait, wide-angle shot, macro shot, etc.
  • Lighting and Atmosphere: Precisely describing lighting effects is key to improving work quality. For example, gobo lighting, soft cinematic lighting, dramatic side lighting, subtle catchlight in eyes.
  • Composition and Space: Use terms like minimalist composition, large negative space, etc., to create clean, breathable images.

2. Break Through Limitations with Labels and Collage

This is an advanced technique for Nano Banana that can solve the problem of AI difficulty generating multiple subjects.

  • Labeled Collage Method: First collage all the elements you want to appear in a reference image and label each element with text (such as "white sneaker"). Then guide the AI generation by referencing these labels in your prompt, which can greatly improve multi-element accuracy.

3. Use "Red Box" or "Mask" for Precise Local Editing

This is another powerful advanced feature that allows you to modify images as precisely as performing surgery.

  • How to Use: Circle the area you want to edit with a red box on the image, then describe the specific operation with text, such as "Separate the person in the red box and turn it into a high-definition single portrait" or "Replace the masked area with a Chanel bag". This ensures that other parts of the image remain absolutely unchanged.

4. Use "Semantic Negative Prompts"

Rather than using negative words (such as "no cars"), positively describe the scene you want for better results.

  • Counter-example: "No cars."
  • Good example: "An empty, desolate street with no signs of traffic."

IV. Common Issues & Solutions

  • Poor Chinese text generation: This is a clear weakness of Nano Banana currently. Chinese characters in images often come out wrong or cannot be generated. If you need to add Chinese text, it's recommended to do post-processing or use other tools.
  • Image quality may be compressed: Some generated images may have quality compression issues. You can use third-party high-definition repair tools for optimization.
  • Multi-subject generation still has limitations: Although the collage method can improve results, directly having AI generate multiple independent and precise subjects in one image is still difficult. For example, photoshopping an absent person into a group photo is not very realistic.