About BlurResolve
The story behind the privacy-first image deblurring tool
About the Project: Necessity is the Mother of Invention
Hi, I'm Karol Ostalski, a software developer from Poland. I created this tool, because of my repeating personal frustration.
Like many other people, I often "take notes" by taking picktures of meeting boards, paper notes, business cards, leaflets, or utility meters. Sometimes in haste or poor lighting, these photos often turn out blurry. While there are many online "unblur" tools, they all share a major flaw: they upload my images to a remote server.
Shortly speaking I don't like such situations. This was a dealbreaker. Whether it was a business document or personal data, I didn't want my images sitting on someone else's server. I asked myself: “In 2026, why can’t my phone or PC fix this locally?”
That’s how BlurResolve.com was born—as a side project dedicated to fixing images directly on your device, ensuring 100% privacy. It doesn't have to be installed. It works directly as web app in your browser - like many more online apps (email, games, chats, etc.).
How It Works: All you need is math
BlurResolve is based on Image Deconvolution. Unlike popular AI tools that might "guess" or "hallucinate" what text looks like, deconvolution is a mathematical technique used to reverse the distortion in a digital image.
When a photo is out of focus, every pixel is distorted in a predictable way (the "Point Spread Function"). By applying the Lucy-Richardson Deconvolution algorithm, we can mathematically reconstruct a sharper version of the original image.
It has some limitations and side effects (for example "ripples" near image boundary), but because of simplicity it can quite easly work in web browser and is quite light for load. When page is loaded then you can use this tool without access to network.
Learn More
Why do I use the GPU?
Nowadays smartphones take high resolution photos. For example if you got 12MP camera in your device, then single image consist of 12 millions of pixes. For image deconvolution each pixel has to be processed hundreds times. A standard CPU might take minutes to process the multiple iterations required. Processing steps are very repeative for each pixel, so GPU (Graphics Card) is the perfect tool for the job. By utilising your device GPU operation can take much shorter time - from minutes to seconds.
About the Author
I’m a software developer with nearly 20 years of experience. I’ve always been a tinkerer; as a kid, I was constantly disassembling clocks and toys to see how they worked. I reassembled them, but I often ended up with extra parts left over ;)
I’ve lived in Warsaw, but I eventually moved to Łomża for a calmer, quieter life for my family. My work always revolves around web applications. My background spans from ancient PHP4 to modern web (backend-frontend). Regardless of the technology (JavaScript, Python, C, Java, etc.), when something isn't working as it should, I find the cause and fix it at the source. It brings me a lot of joy ;)
Connect with me: Find me on LinkedIn