Google was generous enough to provide a free chatbot that can generate images for you. If you've tried that function, then you know that it's not that easy to achieve what you want completely. If you want the AI to generate the image as close as it can to what you're visualizing, here are details you should add to the prompt.
(Photo : Omar Marques/SOPA Images/LightRocket via Getty Images)
1. Start With a General Concept
It's important to be specific with your prompts, but adding broader descriptions could help the AI model narrow down the elements that you want to add to the image, especially when it comes to the background. You can then add more specific details subsequently for a more refined image outcome.
2. Establish a Style and Tone
All the elements in the image you'll generate will be based on the theme you set for the generation of the image. Being clear about the style or tone of the image you want will get you more accurate results, and could automatically take care of the parts that you forgot to specify. With style, for example, you can describe it as "realistic," and for tone, you can add "nostalgic."
3. Secondary Details
If you focus too much on the details of the subject, you might find an image that doesn't acknowledge what you want the overall outcome to look like. You should also describe the secondary details such as how the sky should look, whether the clothing style should be formal or casual, what the items in the background should be, and other elements.
4. Use Positive Descriptors
If you don't want particular details on your image, don't ask the chatbot to exclude it. For example, instead of saying "Don't make the weather sunny," you can instead say "I want the weather to be gloomy and dark." Not only will it narrow what you want down for the chatbot, but you also won't have to correct it if the AI generates everything else other than sunny weather.
5. Use Alternative Words
If you're already distinct about the characteristics you want and it's still not yielding the image you want, you could try using alternative words or synonyms. Instead of saying you want a "busy city," you can choose to input "crowded city" instead. You can switch up both the adjectives and the nouns in this aspect.
Read Also: Google Apologizes, Explains Mishap With Gemini Image Generation
6. Add a Color Palette
The tone of the image is also determined by the color palette that is used. Dark blue and gray tones could make the image look gloomy, and bright vivid colors could make the image look more cheerful and lively. You can specify which individual colors you want by listing them down or even using hex codes to make it as accurate as possible.
7. Reference Artists
Most artists have their own art styles, some of which are more popular than others. With respect for the particular creative's craft in mind, you can try to name one to emulate their art style for your image. You could ask the chatbot to generate an image of a fairytale village in the style of Vincent van Gogh, or a general abstract art in the style of Wassily Kandinsky.
8. Know Its Limitations
Even if the AI image generator has extensive capabilities, there are still limitations to the tool. There are policies in place to avoid misuse such as restricting the use of real people as a reference, as well as brand names and other explicit content. The best you can do is explain the features of the person you want your image to look like.
Related: 10 Google Gemini Functions You Might Not Know About
