Berry AI Studio

The reason I started this tool was straightforward: AI generation produces lots of material, often several images in one run. Keep working with the same prompt and you accumulate many variations. The image managers I was using did not make it easy to review those results together.

I was not just looking for a place to put more files. I wanted to revisit the images made from the same prompt, see how their results differed, and choose which versions to keep working with.

A file browser sees separate images. In this workflow, I see a set of attempts connected by a prompt. Keeping that connection visible is the product idea behind Berry AI Studio.

That changes what matters in the interface. Finding an image is useful; getting back to the related attempts, moving between them, and checking their context is the experience I want to make easier.

The current app supports filtering by prompt content, then browsing the results in a grid or table. Quick Look provides a closer view with arrow-key navigation; the inspector keeps prompts, generation settings, and workflow information available for the selected image.

Favorites and ratings record a choice; tags, albums, and batch actions help organize the results. Prompt filtering currently uses partial content matching. It is a way to narrow a collection, not an automatic exact-prompt grouping or version-history system.

Vue provides the interface, Rust handles services, and SQLite stores the local index inside a Tauri desktop app. Indexed prompt metadata makes the collection searchable beyond its filenames; virtualized browsing avoids rendering every image at once.

This version rewrites the earlier C# application, retained in the repository’s history. It also includes database backup and restore. Local collection management is distinct from online features such as fetching model information from Civitai.

The next question is how little effort it can take to return to a set of same-prompt results, move through the variations, and mark a useful selection. Filtering, previewing, and organizing provide a starting point; the connections between those actions still deserve care.

For this product, a useful outcome is being able to say: these are the attempts I was comparing, this is the version I chose, and this is where I want to continue. That is why the software is worth making.