Food businesses often have plenty of photos but still lack one image that feels complete.
A restaurant might have separate photos of a burger, a drink, a coffee, and a dessert. Each photo may look fine on its own, but placing them together in a menu, delivery promotion, or social post usually produces an inconsistent collage. The lighting changes, the backgrounds do not match, and the final composition rarely feels like one real scene.
That problem inspired me to build Unir Imágenes, an AI-powered web tool that combines multiple uploaded photos into one more coherent image.
The problem with traditional collages
A traditional collage places several images next to each other. It preserves the original photos, but it does not integrate them.
For food photography, this often means:
- different lighting across dishes;
- inconsistent shadows and perspective;
- unrelated backgrounds;
- rigid borders between images;
- a result that looks assembled rather than photographed.
The goal of Unir Imágenes is different. Instead of arranging the original photos in a grid, the tool uses them as visual references for generating a new scene.
For example, a user can upload separate photos of a hamburger, smoothie, coffee, and dessert. The generated result attempts to place those elements into one gastronomic composition with more consistent lighting, shadows, background, and visual style.
How the workflow works
I wanted the interface to stay simple, so the process has only a few steps:
- Upload up to four images.
- Select the image category.
- Choose an output format.
- Generate and download the result.
The tool currently supports JPG, PNG, and WebP files. Users can choose between 1:1, 4:5, 16:9, and 9:16 output formats, depending on where the image will be used.
The category selection is an important part of the workflow. At the moment, the available categories include food, products, pets, and other image types.
This provides the image model with additional context. Food scenes need different composition priorities from product photography or pet images. A food result should preserve the appearance of the dishes while making the overall presentation feel appetizing and coherent.
It is not intended to be a collage maker
One of the main product decisions was to clearly separate the tool from conventional collage editors.
A collage editor is useful when users want exact control over borders, grids, spacing, and the original pixels. Unir Imágenes is designed for a different outcome: creating a new, unified scene based on the uploaded subjects.
That distinction also creates an important limitation. Since the final image is generated by AI, it may not reproduce every small detail perfectly. Results depend heavily on the quality, framing, and visual compatibility of the source images.
Clear photos with visible subjects generally provide better references than dark, blurry, or heavily cropped images.
Possible use cases
Although the tool supports different image categories, I initially focused the experience on food photography.
Some practical use cases include:
- combining separately photographed dishes for a restaurant promotion;
- creating one image for a menu or delivery listing;
- presenting a meal, drink, and dessert in the same scene;
- preparing square or vertical food compositions for social media;
- testing visual concepts before organizing a professional photo shoot.
The same workflow can also be used to combine product photos into a cleaner presentation or create a scene using several pet images.
Technical and product lessons
Building the product involved more than connecting an upload form to an image model.
A few details became especially important:
Handling failed generations
AI image generation can occasionally fail. A credit-based product should not charge users for an incomplete result, so failed generations automatically return the credits used for that attempt.
Keeping uploaded images understandable
Users need to know which files they selected, what category is active, and what output ratio they will receive before starting a generation. Keeping this information visible reduces accidental submissions.
Designing for mobile formats
Food businesses and creators often need the same concept in several formats. Supporting square, landscape, portrait, and vertical output makes the tool more useful for menus, advertisements, Instagram posts, and stories.
Setting realistic expectations
The tool aims to preserve the important subjects and integrate them into one scene, but it is still a generative workflow. Clearly explaining that results may vary is better than presenting AI output as perfectly deterministic.
What I am working on next
The current version provides two free generations after signing in, allowing new users to test the complete workflow.
Future improvements may include better control over composition, clearer style options, more example scenes, and additional guidance for preparing source photos.
For now, I am especially interested in learning which use cases are most valuable: restaurant menus, delivery promotions, product presentations, pet compositions, or something else.
You can try the current version at Unir Imágenes.
I would appreciate feedback on the workflow, the generated results, and the types of controls that would make an image-combination tool more useful.
Top comments (0)