Use a prompt that paints a clear picture
AI-powered image generation tools produce images based on text descriptions. The quality of the images generated depends on the quality of the prompt. The prompt should be specific and include details. Examples of such details are the object or character, the setting or background, the medium, style, or any other adjectives that convey the desired aesthetic.
For instance, if your prompt input is only "tree," the output generated could be a tall oak in a forest, a tiny bonsai on a windowsill, or even a fantastical tree creature. So, instead of a single-word prompt, use a prompt that paints a clearer picture.
Here are a few specifics to include in your prompt:
- Subject. Instead of just using "tree" for your prompt, make it descriptive. For example, you can prompt with "majestic oak tree with spreading branches."
- Setting or background. In your prompt, describe where the tree is, like “a lush green forest" or “a sunny park by a lake."
- Artistic style. Specify a medium for the art or how you want the image to look, such as "watercolor painting," "retro photograph," or "modern digital art."
- Adjectives for the mood. Add words that describe the feeling or ambiance, like "serene," "energetic," "dreamy," "vibrant," or "calming."
- Directive details. Instead of prompting with "ocean sunset," describe the image you want to create. Use a more descriptive prompt like "realistic illustration of an ocean sunset with vibrant hues and silhouettes of palm trees in the foreground."
Let's look at two examples of prompts and their generated outputs.
Example 1
Prompt: A happy female child jogging, cute 3D illustration in the Disney style, octane 3D render, volumetric lighting, sky background.
Microsoft Image Creator outputs:
Example 2
Prompt: Beautiful Renaissance painting of white horses, oil-on-canvas masterpiece with a style reminiscent of Michelangelo, dramatic, mountain background.
Microsoft Image Creator outputs:
When an image is generated, and it doesn't align with your expectations, there's the option to revisit and revise your original prompt. This process involves identifying the necessary modifications to your prompt that better elicits the image you're envisioning. The level of detail in your prompt plays a crucial role in the outcome. The more descriptive your prompt, the more accurate the resulting image is. Therefore, the recommendation is to be as detailed and specific as possible when crafting your prompt.
In addition to revising the prompt, you can also provide feedback directly to the model through a text response. This feedback can specify what aspects of the image need to be adjusted. By clearly communicating these adjustments, you can guide the model to generate an image that more closely matches your vision.
AI-powered image generators can generate realistic and creative images from text descriptions and other inputs. However, there are some potential downfalls to be aware of and avoid when using AI for image generation. Some of these potential downfalls are:
- Complex scenes with multiple subjects. AI tools may need help with generating images with multiple subjects, especially if the subjects are complex or detailed.
- Detailed layout requests. AI tools may not be able to accurately interpret and generate images based on detailed layout requests. For example, a prompt using "a big red table on the left, a friendly dog on the right, and a baby playing with a ball" may not produce the desired result.
- Images with multiple faces. AI tools may not be able to accurately generate images with multiple faces, and the resulting images may be distorted or unrealistic.
- Request for text. AI tools may not be able to accurately generate text within an image. For example, prompting with "a sign saying, 'Happy birthday!'" may not produce the desired result because the generator doesn't know how to spell it.