Restyle
Keep the composition but shift the visual style — anime, oil painting, cinematic, comic, and more.
AI Image to Image Generator
Upload an image and a prompt — re-render it into a restyle, variation, or edited version.
How does AI image-to-image work?
Upload a source image to Renoise Canvas, pick a model (Nano Banana 2 for speed, Nano Banana Pro for quality, GPT Image 2 for multi-reference), write a prompt describing the new version, and generate. The model uses your image as a structural reference and re-renders it according to the prompt — same composition, new style or content.
Starting from scratch with no source image? See the text-to-image guide
What you can do with image-to-image
The main reasons to use an existing image as input.
Variations
Generate multiple interpretations of the same image — useful for exploring options before committing.
Edit and transform
Change specific elements: clothes, background, colors, lighting — anything you describe in the prompt.
Run image-to-image in 3 steps
From a source image to a re-rendered result in the same Canvas.
Step 1Upload the source image
Drag your photo, design, or artwork onto the Renoise Canvas upload card. This becomes the structural reference for the model.
Step 2Pick a model and write a prompt
Choose Nano Banana Pro for most jobs. Write a prompt describing the result — "restyle as an anime illustration, same pose, vibrant colors". The model follows the image structure and re-renders it.
Step 3Generate and refine
Generate, compare against your source, then adjust the prompt or try Nano Banana 2 for a faster iteration loop.
Image-to-image in practice
Upload a source, describe the result — each example starts from an existing image.
Anime restyle
A portrait photo re-rendered as an anime illustration with the same face and pose.
Background swap
Product kept, background replaced with a lifestyle studio scene via prompting.
Style variation
The same landscape photo rendered in an oil-painting style — same composition, new medium.
Outfit change
A model photo re-rendered with different clothes while keeping the lighting and pose.
Which model for image-to-image
All three image models support image-to-image in Renoise Canvas. Choose by the job: speed, quality, or multi-reference fusion.
| For img2img | Nano Banana 2 | Nano Banana ProRecommended | GPT Image 2 |
|---|---|---|---|
| Best for | Fast iterations | Quality restyles | Multi-reference fusion |
| Speed | Fastest | Mid | Mid |
| Output quality | Good | High | High |
| Reference images | 1 | 1 | Up to 16 |
| Up to 4K export | ✓ | ✓ | ✓ |
Image-to-image vs text-to-image — and why the source image matters
Text-to-image starts from nothing: you describe an image entirely in words and the model generates it from scratch. Image-to-image starts from something: you provide a source image alongside your prompt, and the model uses that image as a structural, compositional, or stylistic anchor before re-rendering.
The practical difference is control. With a source image you get outputs that preserve the pose, layout, or likeness of the original — helpful when you are iterating on a real photo, a product, or an existing design. Without a source image, the model interprets the prompt freely, which gives more variation but less predictability.
Renoise supports image-to-image on three models. Nano Banana 2 is the fast lane — quick iterations to test prompts before committing to a longer job. Nano Banana Pro is the quality lane — the same image-to-image approach with Studio-level output and text rendering up to 94% accuracy. GPT Image 2 adds the ability to fuse up to 16 reference images into a single generation, useful when you want to blend multiple sources or keep tight control over the output.
From the img2img starting point you can branch into specific transforms: cartoon or anime restyle, background replacement, outfitting change, or outpainting to extend the canvas. Each of those is covered by its own guide, linked below, so this page stays focused on the concept and model comparison.
Specific image transforms to explore
Image-to-image is the foundation — these guides cover specific transforms built on top of it.
Cartoon / anime restyle
Re-render a photo as anime, comic, or illustrated art.
Background generator
Keep the subject, replace the scene behind it with a lifestyle or branded environment.
Clothes changer
Re-render a fashion or product photo with a different outfit.
Nano Banana Pro
Studio-level image-to-image with strong text rendering and 10 aspect ratios.
Single-purpose img2img tool vs Renoise
Single-purpose img2img tool
- One model, one style of re-render
- No path to other transforms
- Separate tool for each style change
- Limited aspect-ratio or resolution control
- Often watermarks output
Renoise
- Nano Banana 2, Nano Banana Pro, and GPT Image 2 — switch per job
- Restyle, then extend, upscale, or animate on the same canvas
- Up to 16 reference images in one generation with GPT Image 2
- 1K, 2K, or 4K output across 10+ aspect ratios
- Watermark-free exports on paid plans
Choose your plan
One plan unlocks all three image-to-image models — Nano Banana 2, Nano Banana Pro, and GPT Image 2.
✓ 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
Re-render your image
Upload a source image, write a prompt, and generate — watermark-free exports on paid plans.
Frequently asked questions
1.What is AI image-to-image (img2img)?
Image-to-image means you provide both a source image and a prompt, and the model re-renders the image according to that prompt. It uses your image as a structural anchor — preserving composition, pose, or likeness — while applying the style or content changes you describe.
2.How is image-to-image different from text-to-image?
Text-to-image generates an image entirely from a written description, with no source image. Image-to-image takes an existing image as input alongside the prompt, so outputs stay closer to the original composition or subject.
3.Which models support image-to-image in Renoise?
Nano Banana 2 (fast, great for iteration), Nano Banana Pro (high quality, Studio-level output), and GPT Image 2 (fuses up to 16 reference images) all support image-to-image in Renoise Canvas.
4.Can I use multiple source images at once?
GPT Image 2 accepts up to 16 reference images in a single generation, making it the right choice when you need to blend or fuse multiple sources. Nano Banana 2 and Pro use a single source image each.
5.What specific things can I change with image-to-image?
Style (cartoon, anime, painting, cinematic), background, clothing, colors, lighting, aspect ratio — anything you describe in the prompt. For specific transforms there are dedicated guides: cartoon restyle, background swap, clothes change, and more.
6.Does image-to-image keep my subject consistent?
The model uses your source image as a reference, so pose and composition are generally preserved. Full identity consistency (especially for faces) is not guaranteed by the model alone — for use cases involving real, authorized faces, FacePass provides the likeness clearance framework.
By Max, RenoiseLast reviewed Models verified: Nano Banana 2, Nano Banana Pro, GPT Image 2