Case study: szurubooru

When a couple of friends want to share funny images or breathtaking photos with each other, and keep these shared memories to be easily findable in the future, a solution is to use an image board.

However, the popular existing image boards were clunky, required a lot of steps to perform simple tasks, and had questionable privacy.

• Who owns my data?
• Is anything I do shared with advertisers?
• Do I really need to click so many times to tag a few landscape posts with "river"?

With a general idea in mind, I've collaborated on a list of features with the team. Using Adobe Photoshop, I've created low-fidelity designs for post view—the place users spend most time in—and file upload that cuts off all the needless fat.

It turned out that the initial mockups nailed the idea of emphasizing content over flashy controls; the first design decisions shaped the direction of szurubooru's final interface. Minimalist, blocky approach rippled throughout all other pages, making user content the star of the show.

Having learned the shortcomings of competing image boards, my primary goal was designing a user interface that keeps the most important navigation elements and editing options in the same place, regardless of the size or type of viewed content.

While other engines keep tags at the top and serve the user a wall of options that's difficult to parse visually, szurubooru organizes it in readable chunks that don't shift across posts.

To make the experience similarly delightful for board operators, I've been continuously reporting usability issues and designed solutions to convert tedious management into a breeze. Tagging multiple posts became a matter of a few clicks instead of having to visit each post individually to enter its edit mode.

Image boards reach their full potential with tags—each post should be easy to find, so users can e.g. find photos from a specific mountain hike, tagged #Panorama_Ridge. If they want a broader search, perhaps they'd just input ridge and expect the board to only show these images.

For that, the application requires users to feed it tag data. It was a logical step to only accept new media entries if the uploader specified some tags for it, rejecting untagged files.

As it turned out, this caused friction with users. Sharing new images became a chore, and the daily number of uploads halved.

The mandatory tags feature was scrapped, giving way to a simplified upload form. After relaxing the requirements, submitting new images and videos became easier than before, so upload numbers recovered and even increased further.

With each new feature I did my best to keep the GUI minimal, yet still intuitive. If various functions are related, users can be sure they can find them close to each other instead of blindly searching for that one button.

If the user wants to maximize the application's potential, they can head to the built-in exhaustive documentation for search syntax, supported comment formatting and keyboard shortcuts for even faster navigation. I've written parts of the Help pages, ensuring there are real-world examples for sunny and rainy day scenarios—a detail I always long for in technical docs.

The project received over 275 stars and was forked 60+ times on GitHub.

Consistent UI and design aimed at both regular and power users makes using the page a breeze, especially for quick uploads, rapid tagging and managing the database.

One of the instances has well over 50 000 media posts and 20+ users, most of which joined before 2015 and still visit the image board.

I've learned that pushing for quality must not be performed at the expense of users, who might turn away from the service. Incorporating feedback from user behavior helped recover from the rushed change.