Combine two images into one prompt
Two images to prompt writes a single prompt that combines elements of two pictures, for example a person from the first and the scene from the second. You decide how they are combined in your own words, and the tool formats the result for Nano Banana, Midjourney, ChatGPT, Flux, Stable Diffusion or Veo 3.
Dominant colors
Your optimized prompt
🔒 Runs in your browser. Your text is not stored on our servers.
How to use it
Upload image 1
This is normally the subject: a person, a product, a pet or an object. JPG, PNG or WebP, max 4 MB.
Upload image 2
This is the scene, style or mood you want to apply to the first image.
Say how to combine them
Write the merge instruction, such as “put the product from image 1 on the kitchen counter of image 2”.
Pick a target and generate
Choose the model, generate and copy the prompt. For Nano Banana, attach both images in the same order.
How the merge works in each mode
When AI mode is available on this site, both images are resized in your browser (longest side 1024 px) and sent in order to a vision model along with your instruction. It describes the combined result precisely — the identity details of the subject, the setting, lighting, colors and style — in the format of your target model, without naming real people. If you leave the instruction empty, it combines the subject of image 1 with the setting and style of image 2.
Without AI mode, the tool analyzes both images locally: the palette, lighting and framing of the first, and the palette and lighting key of the second. Because local analysis cannot see objects, most targets leave the subject in brackets for you to fill in. For Nano Banana, your instruction does that job, so the local result already reads like a usable edit instruction:
Using the two provided images, put the dog from image 1 on the sofa in the living room of image 2. Keep the identity and key details of the subject from the first image, and match the low-key lighting and deep brown, charcoal, pale yellow palette of the second image. Aspect ratio 4:5.
The aspect ratio is taken from the first image.
Which target to choose
| Target | What you do with the prompt |
|---|---|
| Nano Banana | Attach both photos in Gemini and paste the prompt; best for real merges and edits. |
| ChatGPT / DALL·E | Attach both images in ChatGPT, or use the text alone to generate a new image. |
| Midjourney | Use the text with your own image references added in Midjourney. |
| Flux, Stable Diffusion | Generate a new image that describes the combination; there is no real merge unless your tool supports reference images. |
| Veo 3 | Use it as a starting description for an image-to-video clip. |
In local mode, the Midjourney, ChatGPT and Flux results include your merge instruction at the end. The Stable Diffusion and Veo 3 local results describe only the first image, so add the combination yourself before using them.
Writing a merge instruction that works
The instruction is the most important input on this page. Name which image provides what and be concrete:
- Weak: “mix them”.
- Better: “the woman from image 1 wearing the jacket from image 2”.
- Best: “place the ceramic vase from image 1 on the marble shelf of image 2, same size as the books, matching the warm window light”.
Typical uses include product placement in lifestyle scenes, trying an outfit or hairstyle on yourself, putting a pet into a holiday card background, and applying the color mood of a film still to your own photo. Use only images you own or are allowed to use, and never merge real people into situations that could mislead or embarrass them.
Pro tips
- Upload the subject first and the scene second; the tool and the AI assume that order.
- Use images with similar lighting direction for the most believable merge.
- Mention scale (“same height as the chair”) so the subject does not float or look giant.
- When pasting into Gemini or ChatGPT, attach the images in the same order you uploaded them here.
- If identity drifts, add “keep the face identical” to your merge instruction.
Frequently asked questions
Can I merge two images into one with this tool?
The tool writes the prompt; the merge itself happens in a model that accepts multiple images, such as Nano Banana or ChatGPT. Attach both pictures there and paste the prompt.
What if I leave the instruction empty?
The default is to combine the subject of image 1 with the setting and style of image 2.
Does two images to prompt work without AI mode?
Yes, with local analysis. You get both palettes and lighting keys and a template prompt, but you need to write the subject yourself in the bracket.
Are my images uploaded?
In local mode, no. In AI mode, both resized images are sent to the site's server and AI provider only to write the prompt.
Related tools
Updated: