update📖 3 min read

The Big Rewrite: StillsLab is officially running Semantic Search, and a brand-new UI!

Meet the completely evolved platform featuring AI Semantic Search, a unified design system, and a major speed boost. Here is everything you need to know about the big upgrade.

admin
adminPublished on June 14, 2026
The Big Rewrite: StillsLab is officially running Semantic Search, and a brand-new UI!

Hey everyone,

It’s finally done. Over the last few months, we’ve been quietly pulling off a massive tech surgery on StillsLab.com. We completely threw out the old PHP backend and rebuilt the entire platform from scratch using Next.js.

This isn't just a regular update; it’s a total evolution of how you find visual references. Here is exactly what’s new and why your research workflow is about to get a whole lot smoother:


1. Search that actually gets what you mean (Semantic Search)

The search bar on the filters page is no longer just scanning for exact word matches. It now understands context, mood, and cinematic language.

  • How it works: You can now type like a human. Try something like “cinematic lighting in a cafe” or “dramatic portrait under heavy rain”, and the engine will actually find the right visuals.
  • Exclude the noise (Supporter feature): Need to filter out specific clutter? Just type a minus before the word (e.g., -car). Or, even simpler: just click on any active tag right in the UI to invert it and exclude it from your results.
  • Custom filter logic (Supporter feature): When you select multiple tags — say, a color palette like Warm and Desaturated — you can now toggle between searching for shots that have both styles at the same time, or showing shots that contain at least one of them.

2. "Similar Images" that are actually similar

We completely overhauled our recommendation algorithm. The "Similar Images" feature now does a much deeper analysis of composition, lighting, and overall color contrast, delivering alternatives that actually match the vibe of your selected frame.

3. One-click Google Login

No more managing separate passwords. You can now link your Google account to your current profile in your settings and log in instantly next time.

4. A unified design system & custom layouts

The UI shouldn't feel like a collection of different websites patched together. We’ve rebuilt the platform around a clean, singular design system. Plus, we wanted to give you more control:

  • Custom Homepage: You can now customize how the homepage looks to better fit your display and working style.
  • Flexible Sorting: We’ve added flexible sorting options directly within the filter view.

5. Speed (No more legacy bloat)

By rewriting the entire codebase and wiping out years of unused files and old errors, the site is now incredibly responsive. Snappy page loads, fluid filtering, and zero friction.


Help us crush the bugs

When you rewrite a project this massive, things will break. We’ve done our best to test everything, but we’re realistic: there are definitely some hidden bugs waiting to be found.

If something behaves weirdly, breaks, or just doesn't feel right, please call us out on it! We want to fix it as fast as possible.

Thank you for being part of StillsLab and pushing us to make the platform better. Go explore the new StillsLab.com and let us know what you think!

How did you like this article?

Login to react to this article

Discussion (3)

🔒 Login to join the discussion
J
jflint777Jun 14, 2026, 3:01 PM
Just when I was praising the site to my colleagues for not using AI. If it ain't broke, don't fix it. Disappointing.
J
jflint777Jun 14, 2026, 3:04 PM
Oh, and even more embarrassing is this article was written by AI too. Ironic that you'd take the very human craft of cinematography and turn it into a soulless techbro project.
admin
adminAdminJun 15, 2026, 8:03 PM
Hello, jflint777, thank you for reading and sharing your reaction to this post. Yes, this text was indeed written using AI. I spent a lot of time optimizing the site’s search engine, improving the design, and finally making the site feel cohesive, because previously each page had different visuals and the site looked like a collection of other sites. I made a mistake and had LLM write this text because I was already very exhausted. Regarding the technical side and Artificial Intelligence, I would like to clarify that this only pertains to semantic search and the visual selection of similar images. It runs on transformers.js, which only describes images as vectors. This helps during searches to understand synonyms used in image tags and display images that were previously skipped and not shown when they were needed, which is why the library might have seemed very small. The site is not turning into a typical AI project that generates all the content. My team and I will continue to pour all our energy and love for cinema into it. We will continue to carefully curate all content and images and describe them manually. Simply because we have over 200 requests in “Pending” status, which will result in thousands of images - we simply won’t have the capacity to describe all of them accurately. This will lead to errors, missing words, and, as a result, users will leave the site without finding the frame they need. I promise you and all other users that we will continue to write all the site’s news, produce all the content ourselves, and pour our hearts into this project, because it’s what we’re passionate about. We, too, are against the use of artificial intelligence in creative work and filmmaking.