Finding similar images: search by similarity

Similarity search in Polimake lets you find images that are visually or conceptually similar to a specific image. It's perfect for discovering variations, finding alternatives, or maintaining visual consistency across your projects.

This guide will show you how to use similarity search, when it's useful, and how to leverage it to find exactly what you need.

What is similarity search?

Similarity search:

• Uses an image as a reference to find others that are similar
• Analyzes visual characteristics such as composition, colors, objects, and style
• Finds variations of the same image or theme
• Maintains visual consistency across your projects "

" It's like having an assistant that finds images that "go well together" or are "in the same style."

How to access similarity search

From the detail view

To search for similar images:

1. Open an image in detail view (click any image)
2. Look for the "Similar Images" or "Similar" section
3. Click to see results

Location: The similar images section appears in the detail view of each image, usually at the bottom or in a side panel.

From the context menu

Alternatively:

1. Right-click an image
2. Select "Find similar" or "Related images"
3. Review the results that appear

How it works

Analysis of the reference image

When you search for similar images:

1. Polimake analyzes the reference image
2. It extracts visual and conceptual characteristics
3. It compares them with all the images in your project
4. It ranks them by similarity
5. It shows the most similar ones first

Similarity factors

The search considers:

• Visual content: Similar objects, people, scenes
• Composition: Similar structure and layout
• Colors and style: Similar color palette and aesthetic
• Concept: Related themes and concepts
• Metadata: Similar tags and descriptions

Common use cases

Finding variations

When you have an image you like:

• Search for similar ones to find variations
• Discover alternatives on the same theme
• Find images from the same style or shoot

Example: You have a product photo that works well. You search for similar ones to find other shots of the same product or related products with a similar style.

Maintaining visual consistency

For projects that require coherence:

• Use a reference image in the desired style
• Find other images that match visually
• Maintain consistency in colors, style, and composition

Example: You're creating a catalog and want all the images to have the same style. You use a reference image and find others that match.

Discovering related content

To explore your library:

• Start from an image you're interested in
• Discover related images you didn't know you had
• Find connections between different images

Example: You have a beach photo and discover you have other related beach photos you'd forgotten about.

Finding alternatives

When you need options:

• Search for ones similar to a specific image
• Find alternatives that could work just as well
• Compare options to choose the best one

Example: You need an image for a banner. You have one option but want to see similar alternatives before deciding.

Interpreting results

Ranking by similarity

The results are shown:

• Ordered by similarity: The most similar first
• With a visual indicator: It may show a similarity score
• Grouped: Easy to compare visually

Types of similarity

You may find:

• Exact visual similarity: Same object, different angle or moment
• Conceptual similarity: Same theme or concept, different execution
• Style similarity: Same visual style, different content
• Composition similarity: Same structure, different theme

Tip: Review the first results - they're usually the most similar. If you're looking for something specific, you can refine your search.

Best practices

Use clear reference images

For best results:

• Choose images with clear, recognizable content
• Use images that represent well what you're looking for
• Avoid images that are very abstract or ambiguous

Combine it with other methods

For maximum efficiency:

1. Use similarity search to find initial candidates
2. Refine with semantic search if you need more options
3. Use filters to limit to specific folders

Explore variations

Don't limit yourself to the first results:

• Review several pages of results if available
• Explore different types of similarity
• Discover unexpected connections between images

Limitations and considerations

Requires analyzed images

For it to work:

• The images must be analyzed (this is done automatically on upload)
• It may take time if you've just uploaded images
• Wait for the analysis to finish before searching

Subjective similarity

Similarity can be:

• Subjective: What's "similar" can vary
• Contextual: It depends on which aspects you consider important
• Variable: Different reference images can give different results

Project-specific

Similarity search:

• Searches within the current project typically
• Doesn't cross projects automatically
• Is limited to your current library

FAQ about similar images

Can I search for images similar to ones in other projects?
Currently, similarity search looks within the current project. To search in other projects, navigate to that project first.

How similar are the results?
Very similar visually or conceptually. The first results are usually the closest matches, and the similarity decreases in later results.

Can I adjust how strict the similarity search is?
Currently the search uses optimized default parameters. You can refine results using additional filters if available.

Does it work better with certain types of images?
It works best with images that have clear, recognizable content. Very abstract images or those with ambiguous content may give less accurate results.

Can I use similarity search from the gallery?
Yes, you can access it from the context menu of any image in the gallery, or by opening the image in detail view.

Do the results include the reference image?
No, the reference image is excluded from the results to show only different similar images.

Conclusion

Similarity search is a powerful tool for discovering connections in your library and maintaining visual consistency. Use it to find variations, alternatives, and related content you didn't know you had.

Combine similarity search with semantic search and organizing with folders for a complete, efficient workflow.

Next steps

1. Open an image in detail view
2. Explore similar images for that image
3. Try it with different images to see variations
4. Use it to maintain consistency across projects
5. Combine it with other search methods

Discover hidden connections in your library!