Before you start
- A free account with ChatGPT (OpenAI), Google or Microsoft
- A browser or the ChatGPT, Gemini or Copilot app
Method 1 Generate an image in ChatGPT
Free accounts have a limited number of images per day. When you reach it, ChatGPT tells you when you can create more.
-
Step 1: Open ChatGPT
Go to chatgpt.com or open the ChatGPT app and sign in.
-
Step 2: Describe the picture
Type “Create an image of…” and your description, or click + and choose the image option if it’s shown. Example: “Create an image of a cosy reading corner by a rainy window, warm lamp light, watercolour style, square.”
-
Step 3: Refine it
Reply with changes: “Add a sleeping cat on the chair”, “Make it wider”, “Use brighter colours”. You can also select part of the image and describe what to change there.
-
Step 4: Download
Click the image to open it and use the download button. All generated images are also collected in the Images section of the sidebar.
Method 2 Generate an image in Google Gemini
Images made with Google’s AI carry an invisible SynthID watermark that identifies them as AI-generated.
-
Step 1: Open Gemini
Go to gemini.google.com or open the Gemini app and sign in with your Google Account.
-
Step 2: Ask for an image
Choose Create image if it’s shown, or type “Create an image of…” with your description.
-
Step 3: Edit with words
Ask for changes in the same chat. Gemini is especially good at editing photos you upload: removing objects, changing backgrounds, restoring old photos or combining two images.
-
Step 4: Save it
Hover over or tap the image and choose Download.
Method 3 Generate an image in Microsoft Copilot
Microsoft Designer (designer.microsoft.com) uses the same technology and adds templates for invitations, social posts and cards.
-
Step 1: Open Copilot
Go to copilot.microsoft.com or open the Copilot app (built into Windows 11), and sign in with a Microsoft account.
-
Step 2: Describe the image
Type “Create an image of…” followed by your description. Copilot generates images right in the chat.
-
Step 3: Download or share
Click an image to view it full size, then download or share it.
How to write a good image prompt
The more specific your description, the closer the result will be to what you imagine. Cover these six things:
- Subject: what’s in the picture. “A golden retriever puppy”, not “a dog”.
- Setting: where it is. “On a beach at sunset”, “in a busy Lagos market”.
- Style: photo, watercolour, pencil sketch, 3D cartoon, flat illustration, oil painting.
- Lighting and colour: soft morning light, neon glow, pastel colours, black and white.
- Composition: close-up, wide shot, from above, centred, lots of empty space for text.
- Shape: square (social posts), wide 16:9 (presentations, YouTube), tall 9:16 (phone wallpapers, stories).
Example: “A close-up photo of a steaming bowl of jollof rice on a wooden table, soft natural window light, shallow depth of field, warm colours, square.”
| Use | Example prompt |
|---|---|
| Birthday card | A cheerful flat illustration of balloons and a cake with the words “Happy 60th, Mum” in gold letters, square |
| Presentation slide | A clean minimal illustration of a team working around a laptop, blue and white colours, lots of empty space on the right, 16:9 |
| Phone wallpaper | A calm misty mountain lake at dawn, soft pastel colours, tall 9:16 |
| Product idea | A photo-realistic mock-up of a reusable bamboo water bottle on a gym bench |
| Story or lesson | A friendly cartoon of the water cycle with labels, suitable for children |
Method 4 Edit your own photos with AI
-
Step 1: Upload the photo
In ChatGPT or Gemini, click + (or the upload icon) and choose the photo.
-
Step 2: Describe one change at a time
For example: “Remove the person in the background”, “Change the sky to a sunset”, “Turn this into a cartoon”, or “Restore and colourise this old photo.”
-
Step 3: Compare with the original
Check faces, hands and text carefully. AI edits can subtly change people’s features, so keep the original file.
Troubleshooting
“I can’t create that image”
Your prompt may break the content rules, for example by naming a real person, a copyrighted character or violent content. Rephrase it with a generic description.
The image has odd hands, faces or text
Ask for a fix (“Fix the hands”, “Correct the spelling of the sign to …”), select and edit just that area, or generate a new version.
The image is the wrong shape
Say the shape explicitly, such as “wide 16:9” or “tall 9:16 for a phone screen”, or ask it to recreate the image in that size.
I ran out of free images
Free plans have daily limits. Wait for the reset, or try another of the three tools.
Frequently asked questions
Can I use AI images commercially?
The main tools generally let you use what you create, but check each service’s terms. In many countries, purely AI-generated images may not be protected by copyright, so others could reuse them too.
Which free AI image generator is best?
All three are good. ChatGPT is strong at following detailed instructions and text, Gemini excels at editing your own photos, and Copilot and Designer are handy for cards and social posts.
Are AI images watermarked?
Many are. Google adds an invisible SynthID watermark, and OpenAI and Microsoft add content credentials (C2PA metadata) that show the image was made with AI.
Summary
Open ChatGPT, Gemini or Copilot, type “Create an image of…”, and describe the subject, setting, style, lighting, composition and shape. Refine with follow-up requests, upload your own photos to edit them, then download and check the result. Use AI images honestly, label them where asked, and never misuse real people’s likenesses.
