Start with a text prompt
Describe subject, composition, materials, light, and mood in plain sentences. Specific nouns beat stacked adjectives. If a low angle or soft window shadow matters, say so. Image to Image AI only sees what you write.
Reference-led AI image to image
Image to image inspiration, generated from your text prompt
Find a visual you like, name the traits that matter, and let Image to Image AI create a new picture from your words.
Prompt
A golden paintbrush sweeping a ribbon of rainbow light, with tiny stars against a warm, dark background.
Mode
Text to image
Aspect ratio
1:1
Example artworkGenerator
Where AI image to image starts with words
This Image to Image AI generator gives you working controls. Write a prompt, open one from the Prompt Library, or edit one of three examples, then pick a model, aspect ratio, resolution, and supported quality where available. Unlike classic image to image tools, there is no image input; your text carries the reference.
Loading generator…
Three ways in
Getting started with Image to Image AI
True image to image generation, as in an AI image to image editor, feeds a source picture into the model. Here, you study a reference, name the traits you want, and write them down, so every text-generated result traces back to words you can refine.
Study visual examples
The Prompt Library shows each visual reference next to its prompt. Compare them to learn which words drive crop, color temperature, and texture. That is the observing half of AI image to image work: look first, then translate into text.
Remix a library prompt
Open a library item in the generator and its prompt is already filled in. Swap the subject, keep the lighting, or shift the palette for an image to image feel without uploading anything.
Models
Pick the model that fits the job
Image to Image AI offers three models. Results depend on your prompt and settings, so treat these notes as guidance, not guarantees; running one prompt on two models makes a quick AI image to image comparison.
Nano Banana 2
1K2K4K
7 aspect ratios
A practical pick for drafts and fast iteration across several directions in an image to image AI workflow.
Browse model promptsNano Banana Pro
1K2K4K
7 aspect ratios
Worth trying when a prompt carries dense detail, such as layered scenes or careful lighting notes.
Browse model promptsGPT Image 2
1K2K4KQuality controls
7 aspect ratios
A useful test when an AI image to image prompt includes exact text, labels, or structured layouts.
Browse model promptsOutput
Formats for every AI image to image draft
Aspect ratio, resolution, and supported quality controls shape the final file. Use square for feeds, vertical for stories and posters, and wide for banners or slides. A source-based image to image tool lets the source inform composition; here you choose the frame, so mention framing in the prompt too. Options in Image to Image AI depend on the selected model.
- Aspect ratios
- 7
- Resolution options
- 3
Square · 1:1
Product cards & profile images
Portrait · 9:16
Stories & mobile covers
Landscape · 16:9
Banners & thumbnails
Use cases
Six jobs for AI image to image prompts
Each task pairs prompt advice with a result check. Isolate one dimension per pass, such as lighting or crop, then compare. Image to Image AI returns one image per run, which keeps comparisons clean.
Product and brand visuals
Task: a launch shot for a ceramic mug. Write the glaze, surface, and where the light falls. Name brand colors as hues, since Image to Image AI works from description, not color sampling.
Check: Is the shape believable, do reflections follow one light source, and is there room for a headline?
Photography concepts
Task: pitch a moody portrait concept. Describe lens feel, time of day, and shadow direction, like "late afternoon side light, shallow depth of field." In AI image to image terms, the reference is your mood board and the prompt is your shot list.
Check: Does the light match your notes and the crop suit the pose?
Posters and social graphics
Task: a square post for a weekend market. Put exact copy in quotation marks, keep it short, and say where it sits, like "top third, bold sans serif." Image to Image AI renders lettering as part of the picture, so spelling can drift.
Check: Spell-check every word and confirm it reads at phone size.
Illustration and 3D
Task: a soft 3D mascot for an onboarding screen. Name the medium (clay render, gouache), camera angle, and finish. Keep the backdrop plain; in an AI image to image brief, clutter competes with the subject. For a storyboard reference, describe one clear moment.
Check: Are proportions consistent, and does the style read close to the reference you studied?
Food and drink
Task: a menu image for an iced matcha latte. Specify the glass, ice, foam, garnish, backdrop, and natural or studio light. Name sensory cues like condensation, and ask for one hero drink instead of a crowded table, a simple AI image to image rule for menus.
Check: Does it look appetizing and plausible, with no stray straws or extra glasses?
UI and graphic design
Task: use Image to Image AI to draft a fintech landing page background. Describe shapes, palette, and negative space, and say where interface elements will sit. Skip requests for working UI text; treat mock layouts as image to image mood pieces, not finished screens.
Check: Is the layout area calm, with low contrast behind text?
Three steps
How to use Image to Image AI
The whole AI image to image routine fits in three steps you can repeat as often as needed.
- 01
Find or write a prompt
Browse the Prompt Library for an image to image idea and open it in the generator, write your own, or start from one of three editable examples.
- 02
Set model and format
Choose Nano Banana 2, Nano Banana Pro, or GPT Image 2, then set aspect ratio, resolution, and supported quality. These choices shape AI image to image output and may affect credit use.
- 03
Generate, review, refine
Generate one image and compare it with your reference. Change a single trait, like crop, lighting, or exact copy, and run again; that loop is how reference-led image to image work improves.
Workspace
Three tools, three different jobs
They sit beside the generator, each covering a different part of image to image planning.
Back to the generatorPrompt LibraryPublic visual references shown next to their prompts. Open any item in the generator to learn AI image to image prompting by example.FavoritesA shortlist of items you choose to save, separate from your full record.HistoryYour earlier Image to Image AI generations in one place, so past runs are easy to find.
Credits
How credits work
When credits are configured, each AI image to image run uses an amount based on the model and settings you choose. A verified account is needed to generate when generation is enabled. Small, focused image to image tests help you spend credits deliberately.
Free
$0
Free tier
FAQ
Image to Image AI questions, answered
Can I upload an image for AI image to image editing?
No. Image to Image AI does not currently accept reference uploads, edit existing pixels, or create video. The workflow is text-first: you study a visual, such as one in the Prompt Library, describe the traits you want, and the selected model generates a new picture from that prompt.
How is true image to image different from this workflow?
A true image to image tool feeds your source picture into the model, so its layout and colors can carry over directly. Here, AI image to image is reference-inspired: the visual shapes your thinking, and only your written prompt reaches the model. Results will not copy a reference exactly, but you end up with readable prompts you can reuse and refine.
Which model should I use for image to image prompts?
Start with Nano Banana 2 for quick drafts, then try Nano Banana Pro or GPT Image 2 when a prompt has dense detail or exact text. Outcomes vary, so run the same AI image to image prompt on two models and compare. Image to Image AI creates one image per run, which keeps side-by-side checks simple.
Do I need an account or credits to generate?
When generation is enabled, you need a verified account. If credits are configured, each run uses an amount based on the model and settings you pick, such as resolution. Planning AI image to image drafts one change at a time helps you use credits deliberately in Image to Image AI.
Get started
Ready to write your first prompt?
Pick a reference, describe what you want, and let Image to Image AI generate one image per run.