Trace Id is missing

Modify and refine AI image prompts

So you’ve written a prompt and created an image. Now what? Maybe you want to improve image quality, change the color of the image, or remove an object from the picture. Whatever it is, you know you want to change or enhance your result. Here’s how you can do that by using Designer in Copilot.
Try Designer Download Mobile App
Alien plants with luminous particles in dreamlike setting

Add modifiers for context

You can experiment, get opinions, or revise your prompts to generate better images. Want to change the background color or remove people from the picture? Adjust your prompts accordingly. While you’re at it, try different forms or alternatives of your prompts to see how the results change. 

For example, you could write "a cat wearing a hat" and then switch it to "a cat wearing a sombrero" or "a cat wearing a hat that is cheese." Add even more details to get a desired result. For example, you can say, “Create an image of a cat wearing a top hat and monocle sitting on a red velvet cushion with a background of a library with books and a fireplace.”

A sophisticated cat wearing a top hat and monocle sitting on a red velvet cushion
A view of the Eiffel Tower at sunset through a large circular window

Adjectives are your friends

Play with qualifiers and perspective to refine further. Change colors or adjust the size, shape, position, orientation, perspective, lighting, mood, etc., of your image. For example, you can say, “a large circular window with a view of the Eiffel Tower at sunset” or “a small smiling emoji with heart eyes in the bottom right corner.”

Experiment with infinite variations

From photography and finely detailed sketches to 3D images and elaborate paintings, there are virtually endless ways to experiment with Designer. For example, how many ways can a black and white cat sit on a windowsill?
An astrophotography image of a small black and white cat sitting on a windowsill

 

Create an astrophotography photo of a small black and white cat sitting on a windowsill.
An editorial illustration of a small black and white cat sitting on a windowsill

 

Create an editorial illustration of a small black and white cat sitting on a windowsill.
A stop motion animation of a small black and white cat sitting on a windowsill

 

Create a stop motion animation of a small black and white cat sitting on a windowsill.
So go ahead! Stretch your imagination and see what kinds of fun results and different styles you can get with slight or dramatic modifications to your prompts. Let’s explore some more options with our favorite feline.
A pop art painting of a small black and white cat sitting on a windowsill

 

Create a pop art painting of a small black and white cat sitting on a windowsill.
A children’s book illustration of a small black and white cat sitting on a windowsill

 

Make a children’s book illustration of a small black and white cat sitting on a windowsill.
A 3D environmental art image of a small black and white cat sitting on a windowsill

 

Create an environmental art 3D image of a small black and white cat sitting on a windowsill.

Avoiding common mistakes

It’s easy to overcome and avoid some common pitfalls when using text-to-image models. From low quality to inconsistent results, or just inaccuracy, here are some tips and hints for fixing them.

  • Not specifying what you want clearly or accurately. This can result in images that don’t meet your expectations. While you can prompt with a word or two, a minimum of six descriptive words (not including words like a, for, or from) is better and more likely to give you the desired output. Aim for a sentence or two for even better results.

  • Not providing enough information or context, which can give you generic or low-quality outputs. Add modifiers and more descriptive words and phrases like higher resolution, softer focus, brighter, glowing, dark, dynamic, swirling, etc. to achieve the specific look you want.

  • Add clarifying words to avoid ambiguity or confusion. Remember, you can say “dog,” or you can say “golden retriever.” You can just say “portrait of a woman’, but if you add words like lifelike, abstract, photo-realistic, or “in the style of (insert favorite artist’s name here),” along with other defining characteristics, you'll get better results.

  • If all else fails, try a different prompt to improve your image quality, a synonym, or a different word altogether. There really are no wrong answers. The more you experiment and play, the quicker and more intuitive it becomes.

Conclusion

Now that you know a little more about text-to-image models, you’re ready to start using this powerful, creative tool to generate amazing images and beautiful art with Designer in Copilot. Refer to the tips and best practices in this guide any time to write more effective prompts. Explore, experiment, and discover new ways to express your vision with different types of art and modifiers. Most of all, have fun and enjoy the process.
Make AI art

More from the Designer in Copilot prompting guide:

Images created using Designer in Copilot.
Back to top