To improve your image generation results, follow these best practices:
Be specific: More details give you more control. For example, instead of "fantasy armor," try "ornate elven plate armor, etched with silver leaf patterns, with a high collar and pauldrons shaped like falcon wings."
Provide context and intent: Explain the purpose of the image to help the model understand the context. For example, "Create a logo for a high-end, minimalist skincare brand" works better than "Create a logo."
Iterate and refine: Don't expect a perfect image on your first attempt. Use follow-up prompts to make small changes, for example, "Make the lighting warmer" or "Change the character's expression to be more serious."
Use step-by-step instructions: For complex scenes, split your request into steps. For example, "First, create a background of a serene, misty forest at dawn. Then, in the foreground, add a moss-covered ancient stone altar. Finally, place a single, glowing sword on top of the altar."
Describe what you want, not what you don't: Instead of saying "no cars", describe the scene positively by saying, "an empty, deserted street with no signs of traffic."
Control the camera: Guide the camera view. Use photographic and cinematic terms to describe the composition, for example, "wide-angle shot", "macro shot", or "low-angle perspective".
Prompt for images: Describe the intent by using phrases such as "create an image of" or "generate an image of". Otherwise, the multimodal model might respond with text instead of the image.
Pass Thought Signatures: When using Gemini 3 Pro Image, we recommend that you pass thought signatures back to the model during multi-turn image creation and editing. This lets you preserve reasoning context across interactions. For code samples related to multi-turn image editing using Gemini 3 Pro Image, see Example of multi-turn image editing using thought signatures.
What's next?
See the following links for more information about Gemini image generation: