Create an illustration for a fictional night market in ChatGPT, change one color and save both versions. Compare the edited image with the original to see whether it retained the subject, composition and space for the event text.
Before you start
Sign in to ChatGPT with image creation available to your account. Check your account’s allowance before starting two requests. Free accounts can have separate usage limits for image generation and file uploads. ChatGPT Free Tier FAQ An uploaded image adds another input to review, so use the generated image for this exercise. If you use your own reference later, check permission to upload it and your workspace’s settings. ChatGPT data controls
How to use ChatGPT Image Generation: your first result
Step 1. Start with a new conversation
Open a new ChatGPT conversation. Ask for an image directly in the chat, or open the Images experience and enter the prompt there. Use one conversation for the baseline and its edit so you can refer to the original. Images in ChatGPT
Step 2. Generate the poster artwork
Enter the first prompt below and wait for the image to finish. Open it and check that there is a single lantern, a clear area at the top and room around the illustration. Save the original with the image’s Save control. Keep it open separately for comparison.
Step 3. Ask for one color change
Send the edit prompt below as a follow-up to the image. You can also open the generated image and use the editor’s selection tool to highlight the lantern before describing your change. Review the entire canvas either way. Edits can affect details outside a selected area.
Step 4. Check what the edit retained
Place the two versions side by side. Check that the lantern changed to amber, the hanging loop still connects to the housing and the background and framing stayed usable. Look for an orange halo at the edges, an unexpected second lantern or stray lettering. A useful edit keeps the title area empty.
Step 5. Save the version for layout
Use Save on the edited image if it passes inspection. Open your downloaded file and check its dimensions and crop. Keep the two files with distinct names. Add the event name, date and address as editable text in a design editor so they can be corrected or resized later.
Example input and expected result
First image
Create a vertical editorial illustration for a neighborhood night market. Put one blue paper lantern with a visible hanging loop in the lower half, against a matte dark navy background. Leave the top third empty for an event headline. Flat illustration with clean edges, no lettering or logos.One edit
Change only the blue lantern panels to warm amber. Keep the same lantern shape and hanging loop, the dark navy background, the vertical crop, and the empty top third. Leave the artwork free of text.The target is a pair of usable images with the same subject and framing. Choose the amber version only if it passes the same layout and artifact checks as the blue version.
Three tips for better results
- Name what to preserve as well as what to change.
- Keep the original file so you can check the revision against it.
- Test readability after adding editable text at the final poster size.
Common beginner problems
The edit changes too much
Return to the saved original and ask for the change again with the original shape, placement and blank area named. For a precise recolor, use a design editor.
The chat reports a usage limit
Check the notice in your account and wait for access to return before repeating the request. API billing is separate from ChatGPT.
What to try next
- Add the event details in a layout tool and review the crop.
- Use the targeted image-edit review on another defect.
- Read the ChatGPT Image Generation overview to compare conversational editing with developer workflows and other tools.
Frequently asked questions
Can I edit a photo I already have?
Upload an image you have permission to use, then describe the change. Check its full output against the original.
Does this use the same account and controls as Gemini or the OpenAI API?
This exercise runs in ChatGPT. Gemini has its own image workspace, and the OpenAI API uses the developer platform with separate billing.
Who has rights to the output?
As between you and OpenAI, the applicable terms assign output rights to you to the extent allowed by law. Review third-party rights before using an asset publicly. OpenAI Terms of Use