Change aspect ratio
Turn a 9:16 portrait into a 16:9 landscape, or a square into a wide banner — the model fills in what was outside the frame.
AI Image Extender
Expand the canvas beyond the original borders — outpaint new content and change aspect ratio with AI.
How do I extend or expand an image with AI?
Upload the image to Renoise Canvas, pick Nano Banana Pro, describe what should fill the new canvas area, and generate. The model outpaints new content around the edges — useful for changing aspect ratio (9:16 to 16:9), widening a tight crop, or adding sky or foreground. This is generative outfill, not pixel-perfect seamless cloning, so results look best in scenes where the extended region is described clearly.
Need to enlarge existing pixels without adding new canvas? See the image upscaler guide
What AI image extension covers
The common reasons to expand an image canvas in Renoise.
Widen a tight crop
A subject that was cut off at the edges? Extend the canvas to give it breathing room.
Add sky or foreground
Prompt the model to continue a sky, ground, or architectural scene beyond the original border.
Extend an image in 3 steps
From a tight crop to a wider composition, all inside one canvas.
Step 1Upload your image
Drag the photo or artwork onto the Renoise Canvas upload card so the model has the original content to work from.
Step 2Set target ratio and describe the fill
Choose Nano Banana Pro, set the output aspect ratio wider than the original, then prompt what the extended region should show — e.g. "continue the beach scene, add open sky above".
Step 3Generate and export
Generate, review the edges for coherence, prompt-refine if needed, then export at up to 4K.
Image extension examples
New canvas filled around the original — the model generates plausible content that continues the scene.
Portrait to landscape
A 9:16 outdoor portrait extended to 16:9 — sky and ground filled in generatively.
Wider product shot
A tightly cropped product given more negative space and context around it.
Landscape panorama
A narrow landscape widened into a panoramic banner by continuing the horizon.
Architectural exterior
A building shot extended to include more street, sky, and surrounding context.
Which model for image extension
Both live in the same Renoise Canvas. Nano Banana Pro for photoreal scenes and lighting continuity; GPT Image 2 when you have detailed instructions or want to fuse multiple reference images.
| For outpainting / extension | Nano Banana ProRecommended | GPT Image 2 |
|---|---|---|
| Best for | Photoreal scene continuation and lighting | Instruction-heavy or multi-reference briefs |
| Scene coherence | Best | Good |
| Reference images | Image-to-image | Up to 16 |
| Up to 4K export | ✓ | ✓ |
| Same canvas | ✓ | ✓ |
What "extending" an image actually means — and what to expect
Extending an image means expanding its canvas: the original photo stays in place and the model fills in new content around the edges. The most common reason is changing aspect ratio — a 9:16 phone portrait needs to become a 16:9 banner, or a square product shot needs to become a wide display ad. Outpainting lets you do that without reshooting.
How Renoise does it matters for managing expectations. There is no dedicated pixel-cloning or frequency-matching outpaint tool. Instead, Nano Banana Pro and GPT Image 2 do generative image-to-image: you describe what the extended region should show, and the model generates new content that continues the scene. For most landscape, architectural, and product extension jobs, the results read as plausible continuations. The limitation is also generative: the model invents the extension rather than cloning pixels, so in very high-contrast or highly textured scenes you may see a seam. Prompting the continuation clearly reduces that — "continue the sandy beach, same lighting, no people" gives the model less room to drift than no description at all.
This is a different job from two things that sound similar. Upscaling enlarges existing pixels — the canvas size and pixel count grow, but no new scene content is added. Background swapping keeps the subject but replaces the entire scene behind it. Extension keeps both the subject and the existing scene, adding new canvas around them.
Renoise capabilities used
Image extension in Renoise draws on these models and features.
Nano Banana Pro
Generative outfill for photoreal scenes — extends the canvas with plausible new content at up to 4K.
GPT Image 2
Instruction-following outpainting; fuses up to 16 reference images to guide the extension.
Up to 4K output
Export the extended image at 1K, 2K, or 4K — watermark-free on paid plans.
Multi-ratio Canvas
Set any target aspect ratio from the same canvas — switch between portrait, landscape, and square.
Standalone outpaint tool vs Renoise
Standalone outpaint app
- Single-purpose: extend only
- No follow-on editing in the same tool
- One model, one style of fill
- Limited aspect-ratio presets
- Often watermarks output
Renoise
- Extend, then restyle, upscale, or animate on the same canvas
- Nano Banana Pro or GPT Image 2 — your choice per job
- Any of 10+ aspect ratios from the same interface
- Continue into a video with another model, no re-upload
- Watermark-free exports on paid plans
Choose your plan
One plan unlocks Nano Banana Pro, GPT Image 2, and every other image model.
✓ Current Plan
Starter
$20/mo
Upgrade Plan1,200©/mo
$1.67 / 100©Generate up to 3,000 images or 150 videos every month.✓Watermark-free exports
✓20 FacePass Assets
✓Image Models
✓Video Models
✓ Current Plan
Standard
$60/mo
Upgrade Plan3,600©/mo
$1.67 / 100©Generate up to 9,000 images or 450 videos every month.✓Watermark-free exports
✓50 FacePass Assets
✓Latest Image Models
GPT Image 2 Nano Banana 2 Nano Banana Pro Midjourney V7
✓Latest Video Models
Seedance 2.0 HappyHorse 1.0
◈ Best Value✓ Current Plan
Advance
$200/mo
Upgrade Plan14,000©/mo
$1.43 / 100©Generate up to 35,000 images or 1,750 videos every month.✓Watermark-free exports
✓Unlimited FacePass Assets
✓Latest SOTA Image Models
GPT Image 2 Nano Banana 2 Nano Banana Pro Midjourney V7
✓Latest SOTA Video Models
Seedance 2.0 HappyHorse 1.0
Extend your image
Expand the canvas and outpaint new content — watermark-free exports on paid plans.
Frequently asked questions
1.What is AI image extension (outpainting)?
Outpainting expands the canvas of an existing image and generates new content to fill the added area. You keep the original image in place and the model fills in what would have been outside the frame — useful for changing aspect ratio or widening a composition.
2.Will the extended edges look seamless?
Results look most natural when the extension is clearly described and the scene has gradual transitions — sky, open landscape, plain backdrop. For high-contrast edges or intricate textures, you may see a seam. Renoise uses generative outfill, not pixel cloning, so the model invents the continuation rather than mirroring pixels.
3.How is extending an image different from upscaling it?
Upscaling makes the same image larger at higher pixel density — the canvas content stays identical, just bigger. Extending adds new canvas area and fills it with generated content. Both can be done in Renoise, but they answer different needs.
4.Can I change the aspect ratio — for example, from 9:16 to 16:9?
Yes. Set the target output ratio wider (or taller) than the source, then describe what the new area should contain. The model generates the extended region to match the rest of the scene.
5.Which model is better for extension — Nano Banana Pro or GPT Image 2?
Nano Banana Pro handles most photoreal scene-continuation jobs. Switch to GPT Image 2 if your brief is very detailed or you want to guide the extension with multiple reference images.
6.How does extending an image differ from changing the background?
Extending keeps the original scene and adds canvas around it — both subject and background stay. Changing the background swaps the entire scene behind the subject. If you want a new environment rather than more of the same one, the background generator guide is the right starting point.
By Aini, RenoiseLast reviewed Models verified: Nano Banana Pro, GPT Image 2