Immich MediaKit FAQ

What is Immich MediaKit?

Immich MediaKit is a photo library management tool designed for Immich users, specifically created to help you organize large photo libraries more efficiently.

Key Features:

• 🔍 Finding similar or duplicate images using AI-based detection (via deep learning and vector search)
• ⚡ Batch operations to clean up and manage assets more easily
• 🐍 All built in Python, with a focus on stability and simplicity

Why choose MediaKit over other tools?

A lot of older tools in this space are broken, abandoned, or tricky to set up — so I created this as a modern, actively maintained alternative that just works.

Frequently Asked Questions

Q: Does this tool enhance the duplicate utility already present in Immich with preselection logic?

A: Yes! MediaKit provides Auto Selection functionality. You can set your desired weights, for example:

• Name Longer +3
• FileSize +3

After running Find Similar, the tool will automatically select the photos you want within the identified groups. Then, you simply choose whether to delete the selected photos or the unselected ones.

Q: What advantages does this have over Immich's built-in functionality?

A: MediaKit provides advanced features that Immich doesn't offer:

• Multi-group simultaneous processing: While Immich only lets you process one group of photos at a time, our tool allows you to handle many groups simultaneously by enabling Multi Mode and setting Max Size
• Batch operations: More efficient handling of large quantities of photos
• Automated selection: Automatically choose which photos to keep or delete based on your preferences

Q: Would this work on external libraries as well?

A: Yes! We also support external libraries. Essentially, we just read all the photos; as long as they display correctly in Immich, we can access and use them. However, whether or not they are ultimately deleted is still decided within Immich, so you must enable the recycle bin feature. Of course, if you, like me, run Immich using Docker Compose, I think you need to ensure that the external library's path is not set to read-only.

Q: Does the file location of the photos matter? Will it compare photos across the system or just one folder at a time?

A: This concept is a bit different. The tool I designed approaches this from the Immich user's perspective:

1. When you select a user (e.g., User A who has 10,000 photos) and execute a fetch command to import them into MediaKit, subsequent similarity searches will be performed solely on those 10,000 photos
2. MediaKit will not know your original file locations. We rely solely on Immich's thumbnails (preview is recommended) to find similar photos
3. During the vectorization phase, depending on the number and size of your files, you might need to wait for some time as we need to vectorize the photos
4. We still need to know Immich's actual path (in my case, it's the path configured in my Immich Docker Compose setup) because MediaKit needs to read the thumbnails to obtain photo features

Once the import is complete, you can then search for similar photos based on your chosen criteria.

Q: How does this handle specific duplicate file situations?

A: For example, if you have about 8000+ photos you're having to manually go through because the preselection logic isn't granular enough:

If you have two photos that are exact duplicates, you want Immich to choose the largest photo by default, but if the filename is an actual timestamp instead of something generic like "image_0283.jpg", you want more intelligent selection.

MediaKit's Auto Selection feature can solve this problem. You can set weights based on filename length, file size, and other multiple criteria, allowing the tool to automatically make more suitable choices.