How to Create Images with AI: Step-by-Step Guide
Everything you need to go from a blank prompt box to a finished image: the Text to Image flow, the Image to Image flow, and the prompt habits that make results noticeably better.
Generate an Image from a Prompt
The fastest path: describe what you want and let the model build it.
Write your prompt
Open the generator on the Text to Image tab. Describe the subject, style, lighting, and mood. The more specific you are, the closer the result.
Set size and format
Pick an aspect ratio for where the image will be used, choose a model, and set the output format. The defaults work well if you are unsure.
Generate and refine
Click generate and your image arrives in seconds. Download what you like, or adjust the prompt and run it again.
Transform an Image You Already Have
Start from a photo or sketch and reshape it with a prompt.
Upload a reference
Switch to the Image to Image tab and upload the picture you want to work from: a sketch, a product shot, or any photo you want to restyle.
Describe the change
Write a prompt for what should change and what should stay. Be clear, for example: keep the subject, change the background to a sunset beach.
Generate and compare
Generate, then compare the result with your original. Adjust the prompt and repeat until the look is exactly right.
Prompt Tips
Small Prompt Changes, Big Difference
The gap between an okay image and a great one usually lives in the prompt. Compare the weak version with the stronger one.
Be specific about lighting
good lighting
soft golden-hour light from the side, warm tones
Name the camera and lens
a close-up photo
shot on an 85mm lens, shallow depth of field
Describe the environment
a mountain
a misty alpine valley at dawn, low clouds between the ridges
Set the mood
a city at night
a rain-soaked street at 2am, neon reflections on wet pavement
Spell out any text
add a sign
a chalkboard sign reading 'Open Today' in white handwritten chalk
Direct the composition
a person standing
a full-body portrait, subject centered, clear negative space on the left
Choosing Your Settings
Three settings shape every generation. Here is how to think about each one.
Aspect Ratio
Square for social posts, landscape for banners and presentations, portrait for stories and mobile. Pick the shape of the space your image will live in.
Output Format
PNG keeps maximum quality and supports transparency. JPEG gives smaller files for fast web use. Choose based on where the image is headed.
Model Choice
Different models lean toward different looks: clean photorealism or expressive illustration. Switch models to match the style you are after.
Ready to Create?
Put it into practice. Open the generator and make your first image.
Frequently Asked Questions
Common questions about AI image generation — workflow, settings, and getting better results.
How long does a generation take?
Z-Image Turbo finishes in 5–10 seconds. Nano Banana 2, Grok Imagine 2.0 and Seedream 4.5 take roughly 25–60 seconds. GPT Image 2 takes about 40 seconds on Low and 3–4 minutes on High; Seedream 5.0 Pro 1.5–2.5 minutes. The generator shows the expected time for your settings before you click. Keep the tab open until the image appears — the result is delivered to the page you generated from, and a closed tab cannot receive it.
Can I use my own photo as a starting point?
Yes. Switch to the Image to Image tab and upload a photo, then describe how you want it changed: restyle it, swap the background, or adjust the lighting and mood.
What if the result isn't quite what I wanted?
Iterate: refine your prompt and generate again. Small, specific changes (lighting, camera, mood) usually make the biggest difference, and iterating is fast and normal. Lock the seed in Advanced Settings to keep the same base composition while you fine-tune details.
Do I need design skills to get good results?
No. A clear written description is enough. The prompt tips above show how to phrase a request so the model understands exactly what you mean.
Which aspect ratio should I choose?
Match the shape of where the image will be used: 1:1 (square) for social posts, 16:9 (landscape) for banners and slides, 9:16 (portrait) for stories and mobile screens, 4:3 or 3:4 for prints and thumbnails.
Can I use the generated images commercially?
Yes, images you generate are yours to use for both personal and commercial projects. See our terms of service for the full details.