Image to image AI lets you start from an existing picture instead of a blank canvas. You upload a photo, describe the change you want in plain language, and the model generates a new image that can preserve the structure of your original while applying the transformation. It is a practical way to restyle artwork, create product photo variants, restore old photos, or turn a rough sketch into a finished illustration.
This guide explains how image to image AI works, walks through five practical use cases, and shows you exactly how to run your first transformation on MakeImg — including which model to pick and how much each generation costs.
What Is Image to Image AI?
Image to image (often shortened to i2i or img2img) is a mode of AI image generation where the model receives two inputs instead of one:
- A reference image — the photo, artwork, or sketch you want to transform.
- A text prompt — a description of what the output should look like.
The model uses the reference image as a structural anchor. Composition, subject placement, and overall layout can carry over from the original, while the prompt guides style, lighting, materials, and any elements you want to add or change.
This is different from text-to-image generation, where the model builds a scene from a written description alone. With image to image, your reference provides a starting composition, which can make the result easier to guide. Exact preservation still varies by model, prompt, and source image.
How It Works Under the Hood
The exact process varies by model, but the practical workflow is the same: the system analyzes your reference image and combines it with the instructions in your prompt. A clear source image and a specific prompt give the model better information to work with; blurry inputs and vague instructions make faithful edits harder.
5 Practical Use Cases for Image to Image AI
1. Style Transfer
Turn a photo into a watercolor painting, an anime frame, an oil portrait, or a Studio Ghibli-style scene. Upload the photo, describe the target style, and the model repaints the same scene with the new aesthetic. This is the most popular i2i use case because it requires zero artistic skill — the composition is your photo, the style is one sentence.
Example prompt: "Convert this photo into a soft watercolor painting with visible brush strokes and pastel colors."
2. Product Photo Variants
E-commerce sellers use image to image to multiply one product shot into an entire catalog. Photograph your product once, then generate variants: different backgrounds, seasonal themes, lifestyle settings, or studio lighting styles. Instead of booking a new photoshoot for every campaign, you generate on-brand variations in minutes.
Example prompt: "Place this product on a marble countertop with soft morning light and blurred kitchen background, professional product photography."
3. Photo Restoration and Enhancement
Old family photos with scratches, faded colors, or low resolution can be restored. Upload the damaged scan and ask the model to remove noise, repair visible damage, or add color. Always compare faces and small details with the original because generated restoration can invent details.
Example prompt: "Restore this old photograph: remove scratches and dust, fix faded colors, sharpen facial details, keep the original composition."
4. Character Consistency
Illustrators and storytellers face a classic problem: getting AI to draw the same character twice. A reference image can improve consistency. Use an existing character image, then prompt for a new pose, outfit, or scene. The model may carry facial features and identifying details into the new result, but consistency is not guaranteed and usually takes iteration.
Example prompt: "Same character, now sitting in a cafe reading a book, warm afternoon lighting, same art style."
5. Sketch to Finished Artwork
A rough pencil sketch or even a crude digital doodle contains the one thing AI struggles to guess: your intended composition. Upload the sketch and describe the finished piece. The model treats your lines as a blueprint and renders a polished illustration on top of them. Architects, concept artists, and product designers use this to jump from idea to presentation-ready visual in one step.
Example prompt: "Turn this sketch into a detailed digital illustration, vibrant colors, clean line art, fantasy concept art style."
Step-by-Step: How to Use Image to Image on MakeImg
New accounts receive welcome credits, so you can try a Fast image-to-image model without entering a payment method.
Step 1: Open the Image to Image Tool
Go to the Image to Image generator and sign in. If you are new, registration takes seconds with email or Google, and your free welcome credits are added automatically.
Step 2: Upload Your Reference Image
Click the upload area and select your source picture. For best results use a clear, well-lit image — the model can only preserve details it can actually see. Photos, scans, screenshots, and sketches all work.
Step 3: Write Your Prompt
Describe the output you want. Two rules make prompts dramatically better:
- Name the transformation explicitly. "Convert to watercolor style" beats "make it nice".
- Say what should stay the same. Adding "keep the original composition and pose" tells the model what not to touch.
Step 4: Pick an Aspect Ratio
Choose 1:1, 16:9, or 9:16 — or select "original" to match your uploaded image's dimensions, which is usually the right choice for restorations and style transfers.
Step 5: Choose a Model and Generate
Select a model based on your quality and budget needs (see the comparison below), then click Generate. Completion time varies by model and provider demand. When the task finishes, download the result or adjust the prompt and run again.
Which Model Should You Use?
MakeImg gives you access to multiple image to image models through one interface, priced in credits per image. Here is how they compare:
| Image-to-image model | Credits per image | Best for |
|---|---|---|
| Nano Banana 2 Lite | 5 | Default Fast model for everyday edits and iteration |
| Seedream 5.0 Lite | 5 | Budget-friendly edits and style exploration |
| Nano Banana | 5 | Quick everyday editing |
| Nano Banana 2 | 8 | Higher-resolution general-purpose work |
| GPT Image 1.5 | 8 | General-purpose editing at Standard pricing |
| GPT Image 2 | 8 | Detailed instructions and editing at Standard pricing |
| Nano Banana Pro | 18 | Pro-tier fidelity and fine detail |
A simple decision rule:
- Testing an idea or iterating on prompts? Start with the default Nano Banana 2 Lite or another 5-credit Fast model.
- Comparing Standard models? Nano Banana 2, GPT Image 1.5, and GPT Image 2 each cost 8 credits and offer different editing profiles.
- Need a Pro-tier option? Nano Banana Pro costs 18 credits and is the image-to-image Pro model currently available in the generator.
You can read detailed capability notes for every model on the models page.
Tips for Better Image to Image Results
- Start from the best input you have. Upscale or crop your reference before uploading rather than asking the model to fix everything at once.
- Change one thing at a time. "Watercolor style" plus "different background" plus "new outfit" in one prompt gives the model too many jobs. Chain generations instead: transform the style first, then feed the result back in for the next edit.
- Use style presets as a starting point. The MakeImg generator includes one-click presets (watercolor, cinematic, anime, product photography, and more) that append proven prompt suffixes for you.
- Iterate cheaply, finish expensively. Do exploration on Fast models, then rerun your winning prompt on a Pro model for the final export.
Frequently Asked Questions
What is the difference between image to image and text to image?
Text to image creates a picture entirely from a written description. Image to image starts from an existing picture and transforms it according to your prompt, preserving the original composition. Use text to image when you are starting from nothing; use image to image when you already have a visual you want to build on.
Can I use image to image AI for free?
Yes. New MakeImg accounts receive free welcome credits — enough to generate images with Fast models at no cost, with no credit card required. See our guide to free image to image AI for details on exactly what the free tier covers.
What image formats and types work as input?
Standard formats like JPG and PNG work. In terms of content, photos, digital art, scans of physical artwork, screenshots, and hand-drawn sketches are all valid inputs. The main requirement is clarity: the sharper the source, the more faithful the transformation.
How long does a generation take?
Generation time varies by model, provider demand, and the requested edit. Some tasks finish in tens of seconds, while others can take a few minutes. The generator tracks the task and shows its final status.
Do I own the images I create?
MakeImg's Terms of Service state that you retain ownership of images you generate. Commercial usage terms depend on your selected pack or plan, so review the license listed on the pricing page.
Start Transforming Your Images
The fastest way to understand image to image AI is to run one generation yourself. Upload a photo, type one sentence, and watch it become a watercolor, a product shot, or a restored memory.
Try the free Image to Image AI generator — your welcome credits are waiting, no credit card required.