Ecommerce background and lighting
Move a product from a plain source setup into a campaign-ready environment while asking the model to preserve the bottle, label, and proportions.
GPT Image 2 powered editing
AI Image Editor
Use this ai image editor to upload one source image, describe what should change, optionally add visual references, choose the aspect ratio and resolution, then generate a new edit.
Ready for a source
Result stage
Ready for a source
Precise edits begin with a clear source
Upload the original, add up to 15 ordered references, then describe only the changes you want.
Source stays first · references follow in order.
Prompt-led editingWhat Is an AI Image Editor?
An AI image editor applies a written transformation to an existing image. The source establishes the visual baseline; your prompt explains the change, the details to preserve, and the intended finish.
Unlike a blank-page image generator, this workflow begins with a source. Optional references can guide color, material, lighting, or style, while the prompt tells GPT Image 2 how those inputs relate.
The result is a newly generated image, not an editable Photoshop document. Inspect identity, geometry, text, hands, edges, and brand details before using an output.
How to Use an AI Image Editor
Use the editor in four steps: upload a source, write the edit and preservation rules, choose output settings, then generate and inspect the result.
- 01
Upload the source
Choose the image that defines the subject, composition, and starting visual state.
- 02
Describe change and preservation
State what should change, then name the identity, geometry, text, or framing that must remain.
- 03
Choose the output
Select an aspect ratio and 1K, 2K, or 4K resolution that fits the publishing surface.
- 04
Generate, inspect, and download
Sign in at generation, review the result against the source, then download or refine the instruction.
AI Image Editing for Real Creative Work
AI image editing is most useful when a real creative asset needs a deliberate variation rather than a completely unrelated replacement.
Fashion color and material direction
Test a wardrobe color change while explicitly preserving the model’s face, pose, garment construction, framing, and surrounding light.
Campaign poster and text update
Direct a new headline and visual hierarchy for a campaign concept, then review every word and layout decision before publishing.
Interior material restyling
Explore a new palette and furniture direction while protecting the room geometry, camera position, windows, and usable floor plan.
What You Can Change—and What to Preserve
You can direct broad visual changes while protecting important details in the same prompt. The current workbench is designed for prompt-led whole-image editing, not pixel-local manual selection.
Background and setting
Replace a studio, room, landscape, or campaign environment while naming the subject details to keep.
Color and material
Change a palette, garment color, surface, finish, or material direction with explicit preservation constraints.
Lighting and finish
Ask for a new time of day, light quality, contrast, reflections, or editorial grade.
Object and composition
Add, remove, or reposition visual elements through a whole-image prompt, then inspect surrounding details.
Typography and layout
Direct short headlines, labels, hierarchy, and negative space; always proofread generated text.
Style and art direction
Shift a visual toward a photographic, editorial, illustrated, architectural, or campaign treatment.
The editor does not provide brush, mask, or layer controls. Avoid promising an exact pixel-local change; state the intended edit and preservation rules, then compare the result with the source.
Write Better AI Image Editor Prompts
A useful AI image editor prompt separates the requested change from the visual facts that must survive it. Add the finish and final use only after those two decisions are clear.
Change01
Name one clear transformation first.
“Replace the background with warm limestone and soft window light.”
Preserve02
List the visual invariants that matter.
“Keep the bottle shape, label spelling, cap, camera angle, and crop unchanged.”
Finish03
Define lighting, material, palette, and visual quality.
“Use restrained reflections, natural shadows, and a premium editorial finish.”
Output04
Explain where the edit will be used.
“Compose for a 4:5 ecommerce campaign image with clear space above the product.”
AI Image Editor vs Adobe Firefly vs Canva Magic Edit
Banana Pro AI, Adobe Firefly, and Canva Magic Edit support different editing decisions. The relevant distinction is the workflow around the change—not an unsupported claim that one tool is universally better.
| Decision | Banana Pro AI | Adobe Firefly | Canva Magic Edit |
|---|---|---|---|
| Editing workflow | Prompt-led whole-image editing in a focused browser workbench | Generative Fill inside the Firefly editing workflow | Magic Edit inside Canva’s broader design and photo editor |
| Selection method | Describe the change and preservation rules; no exposed brush or mask | Brush or select a target region before prompting the fill | Select an area or element, then prompt an add, remove, or change action |
| Reference input | One source plus as many as 15 optional ordered references | A reference image can be added to guide Generative Fill | The selected canvas content provides context inside the design workflow |
| Output controls | Common aspect ratios with 1K, 2K, and 4K choices | Output continues in the Adobe editing workflow | Output continues in the Canva design and publishing workflow |
| Practical fit | Prompt-led whole-image variations when the full source can be evaluated together | Localized edits when brush-based region selection is important | Template, social, and design assembly around the edited image |
Product facts
This page exposes prompt-led editing, one source plus optional ordered references, and 1K/2K/4K controls.
External product facts
Adobe documents brush-led Generative Fill and optional references. Canva documents prompt-based add, remove, and change actions on selected content.
Workflow recommendation
Choose whole-image prompting for prompt-led variations, Firefly for brush-selected local work, or Canva when the edit belongs inside a template and publishing flow.
Formats, References, Resolution, and Credits
The editor accepts JPG, JPEG, PNG, and WebP files, supports one source plus 15 references, and offers 1K, 2K, and 4K output with credit costs shown before generation.
Inputs and formats
- JPG/JPEG, PNG, and WebP uploads
- 20 MB maximum per uploaded image
- One required source plus up to 15 ordered references
- References can guide lighting, material, color, or style
Resolution and credits
- 1K: 20 credits
- 2K: 40 credits
- 4K: 40 credits
- Sign-in and available credits are checked at generation
Aspect-ratio limits
Available ratios are auto, 1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16, 2:1, 1:2, 3:1, 1:3, 21:9, 9:21, 5:4, and 4:5. At 2K or 4K, 5:4, 4:5, 3:1, 1:3, and 9:21 are disabled.
This is a credit-based tool, not an unlimited free service. Generated details can vary, so review text, identity, geometry, edges, and reference adherence before publication.
AI Image Editor FAQ
These answers cover how the editor uses sources and references, which files and resolutions it accepts, how credits work, and where the current prompt-led workflow stops.
What is an AI image editor?
An AI image editor changes an existing image from a written instruction. You provide the source, describe the transformation and what to preserve, and the model produces a new image rather than a traditional editable layer stack.
How do visual references work?
The first uploaded image is the source. You can add up to 15 ordered references for lighting, material, color, or style context. Explain the role of each reference in the prompt instead of assuming the model will infer it.
Which image files can I upload?
The editor accepts JPG, JPEG, PNG, and WebP. Each upload is limited to 20 MB, and the full input stack can contain one source plus 15 references.
Which aspect ratios are unavailable at 2K or 4K?
At 2K and 4K, 5:4, 4:5, 3:1, 1:3, and 9:21 are unavailable. Choose 1K for those ratios or select a compatible ratio before generating a higher-resolution edit.
Is this AI image editor free?
No unlimited-free claim is made. This is a credit-based editor: 1K costs 20 credits, while 2K and 4K each cost 40 credits. The Generate button shows the current credit cost. Sign-in and available balance are checked when you generate.
Does the editor have a mask, brush, or layers?
No. The current editor does not provide brush, mask, or layer controls. It uses prompt-led whole-image editing, so describe the target change and the elements that must remain in precise language.
How can I preserve important details?
Name important invariants explicitly: identity, product geometry, logo placement, exact short text, camera position, crop, pose, and colors. Review the result closely because generated edits can still vary.
Sources and product notes
Product limits on this page come from the implemented editor contract and integration documentation. External comparison statements link to the vendors’ own current feature documentation.
Related AI image tools
Continue with the GPT Image 2 model guide, compare other image workflows in the model directory, or review the current credit plans before generating.
Edit Your Image with AI
Start with the real source, write one clear change and a short preservation list, then choose a resolution that fits the job.
Open the editor