Use a clear, specific prompt and a suitable image when the selected flow requires one.
For text prompts:
Describe the main subject and what it is doing.
Add the setting, mood, lighting, camera movement, and visual style when relevant.
Keep the request focused instead of combining too many unrelated actions.
Use the prompt enhancement option when you want help expanding a short idea.
For image inputs:
Use a sharp, well-lit image with the important subject clearly visible.
Avoid very low-resolution, heavily blurred, or heavily obstructed images.
Results can vary even when you reuse the same prompt or image. If the result is not what you expected, try a clearer prompt, a different image, another supported model, or a different template.