DEV Community: yesanson

I got tired of uploading sensitive images to random websites, so I built a local-only blur tool

yesanson — Tue, 07 Apr 2026 03:07:00 +0000

A few days ago, I was preparing a technical blog post. I needed to blur a few email addresses and a face in a screenshot before publishing.

I did what most developers would do. I searched for a free online image blur tool, clicked the first result, and uploaded my image.

Then I stopped.

Where was that image going? Was it being stored on a server somewhere? For how long? Who else had access to it? The website had no privacy policy, no mention of data retention, and no reassurance that my file would be deleted after processing.

That feeling of unease stuck with me. So I decided to build something better.

The result is Blur-image.org, a browser-based image blur tool that processes everything locally on your machine. No uploads, no servers, no third parties ever seeing your data.

How it works under the hood

Everything runs directly in the browser using standard web technologies. When you open the tool, all the heavy lifting happens on your device via JavaScript, Canvas API, and local processing pipelines.

There are no backend calls for image manipulation. No API keys sent over the wire. No analytics trackers that log your image metadata. The second you close the tab, every trace of your edit is gone.

The tool supports several distinct operations, each implemented with a focus on performance and privacy.

Selective Blur Tool

This uses canvas-based brush and selection rendering. You can manually paint over areas to blur, adjust blur strength dynamically, and create multiple independent blur regions. All operations are stored locally in memory as coordinate maps, never serialized or sent anywhere.

Full Image Blur Tool

A global convolution effect applied to the entire canvas. The intensity is adjustable, and you can restore specific areas by painting over them after the full blur is applied. This is essentially a local mask that overlays the original image data where you choose.

Face Blur Tool

This one relies on a client-side face detection model. The model runs locally using TensorFlow.js or a similar lightweight library bundled with the tool. It detects multiple faces in a single frame, draws bounding boxes, and applies either Gaussian blur or pixelation over each detected region. You can manually adjust the detected areas afterward because detection is never perfect on the first pass.

The key detail here is that the face detection model loads once into your browser's memory and never sends image data to any external endpoint for inference.

Sensitive Information Blur Tool

This feature runs OCR locally using a browser-compatible text detection engine. It scans the image for patterns that match phone numbers, email addresses, IDs, and physical addresses. The detection happens entirely in JavaScript using regular expressions combined with bounding box recognition. Once text blocks are identified, you can blur them, remove specific detections, or add new ones manually.

Because the OCR runs locally, no text content ever leaves your device. That means you can blur confidential documents, internal screenshots, or personal photos without worrying about a server logging the text.

Text Blur Tool

Similar to sensitive information blur, but designed for general text detection. It identifies text blocks at the word or line level and applies blur or pixelation. This is especially useful for UI mockups, code screenshots, or any image where text needs to be obscured without fully destroying readability.

Pixelate Image Tool

Pixelation is implemented by downscaling a selected region to a lower resolution and then scaling it back up with nearest-neighbor interpolation. You control the pixel block size. This works on the full image or on user-selected areas. From a privacy perspective, pixelation is often stronger than blur because it removes identifying details more aggressively.

Motion Blur Tool

A directional blur effect that samples pixels along a specified angle and distance. This is GPU-accelerated where possible using CSS filters or WebGL shaders. The result is a dynamic, creative effect that feels completely different from standard Gaussian blur.

Why local processing matters for developers

As developers, we think about data flow constantly. We encrypt databases, validate inputs, and audit third-party dependencies. But when we use an online image editing tool, we often throw all those instincts out the window and upload raw, unencrypted images to servers we know nothing about.

That is a strange blind spot.

Local processing eliminates entire categories of risk. No man-in-the-middle attacks on your image data. No server breaches exposing user uploads. No legal gray areas about data retention or ownership.

It also means the tool works offline. Once the page loads, you can disconnect from the internet and still blur, pixelate, and detect faces to your heart's content.

Performance considerations

Because everything runs locally, performance depends on your device and browser. Large images with high-resolution face detection or OCR will take longer and consume more memory. The tool handles this by processing images at reasonable default scales and giving you control over output quality.

The trade-off is worth it. Waiting an extra second for a local process is infinitely better than uploading a sensitive image to an unknown server.

What I built this for

I built Blur-image.org for the exact scenario I found myself in: needing to quickly blur sensitive information in a screenshot without second-guessing where that screenshot was going.

It is also for developers who write tutorials and need to obscure API keys, email addresses, or internal endpoints in their examples. For designers sharing mockups with client data visible. For anyone who has ever felt uneasy hitting the upload button on a free online tool.

Try it
You can use the tool at blur-image.org. No sign-up, no payment, no tracking. Open the site, load an image, and start blurring. Everything stays on your machine.

I am actively working on improvements, and I would genuinely appreciate feedback from other developers. Does the face detection work well on your test images? Is the OCR fast enough for documents? Are there edge cases where the local processing feels too slow?

Drop your thoughts in the comments or reach out directly. I built this to solve my own problem, but I want it to be genuinely useful for other developers who care about privacy as much as I do.

Thanks for reading. Go blur something without worrying about who is watching.

Why I Built Wan AI Platform (And Why Creators Deserve Better Than Juggling Ten Different Tabs)

yesanson — Mon, 30 Mar 2026 04:47:44 +0000

Let me be honest with you: I built Wan AI Platform because I got tired of the chaos.

For years, I watched creators, marketers, and video professionals bounce between half a dozen different tools just to get one project finished. One site for text-to-video. Another for image-to-video. A third for upscaling. A fourth that might handle lip sync if you squinted at it right. By the time you’d exported your final file, you’d burned through three subscriptions, two free trials, and a significant chunk of your patience.

And the worst part? None of it felt connected.

Every tool had its own logic. Its own limitations. Its own weird quirk that you had to learn to work around. What should have been a creative workflow turned into a logistics problem.

I remember sitting there thinking: Why is this still so fragmented?

We’re in an era where AI can generate stunning video from a sentence. Where models like Wan 2.6 can produce 1080p footage with accurate lip sync from audio, text, or a reference clip. The technology is incredible. But accessing it? Actually using it in a real workflow? That still felt like assembling IKEA furniture with instructions in three different languages.

So I stopped waiting for someone else to clean it up.

Wan AI Platform exists because I believe powerful tools shouldn’t require a scavenger hunt to use.

If you want to generate a video from text, it should be right there. If you want to take that video and remix it with a different style, that should happen in the same place—not after exporting, re-uploading, and praying the next tool doesn’t compress it into oblivion. If you want to animate a static image or transform existing footage, it should all live under one roof.

Not scattered. Not cobbled together. Unified.

I built this platform around a very simple conviction: creators shouldn’t have to compromise between power and simplicity.

You shouldn’t have to choose between professional-grade output and a workflow that doesn’t make you want to close your laptop.

So I made sure everything you need is in one place.

The full suite of official Wan AI models—Wan 2.6, Wan 2.5, Wan 2.2, Wan 2.1, and Wan Animate—all accessible without switching services. Text-to-video when you’re starting from scratch. Image-to-video when you’ve got a concept already visualized. Video-to-video when you want to remix or refine existing footage. Text-to-image and image-to-image for everything in between.

No more “I’ll use this one for motion, that one for detail, and cross my fingers they look like they belong in the same project.”

And I’ll be honest: the watermark thing mattered to me more than I expected.

There’s something quietly insulting about generating content you poured creative energy into, only to have someone else’s logo stamped across the corner. If you’re using this for a client project, a marketing campaign, or your own business, the output should be yours—fully, cleanly, without a banner advertising the tool you used.

So everything on Wan AI Platform downloads watermark-free. Commercial-ready, right out of the gate.

The other thing I cared about? Speed.

Not because faster is always better, but because waiting kills momentum. You have an idea. You want to see it. Waiting five minutes—or twenty—while a video renders doesn’t just slow you down. It breaks the creative flow. You stop experimenting. You settle for “good enough” because you don’t have time to iterate.

Our infrastructure is optimized to deliver results in seconds. Not because I’m chasing some technical benchmark, but because when you’re in the middle of creating, seconds matter.

And yes, the quality had to hold up.

It’s easy to make something fast. It’s harder to make something that looks like it belongs in a professional portfolio. Wan 2.6 delivers 1080p HD video with accurate lip sync—whether you’re generating from text, audio, images, or reference clips. Wan 2.5 gives you better motion consistency and visual stability. Even the earlier models, like Wan 2.1, laid groundwork that most tools still haven’t caught up to.

But having great models isn’t enough if the interface makes you fight for every result.

So I built the experience the way I’d want it: intuitive enough that you don’t need a technical background to get started, but powerful enough that professionals don’t feel held back. Choose your model. Pick your input method. Generate. That’s it.

Who is this actually for?

Content creators who need to produce engaging video assets without a production team. Marketing teams running campaigns that demand fresh visuals faster than traditional workflows can deliver. Video producers experimenting with new styles without rebuilding projects from scratch. Digital artists who want to push ideas further, faster. Small businesses and freelancers who need professional-quality content without the agency price tag.

Anyone who’s ever opened ten tabs, signed up for three free trials, and thought there has to be a better way.

I built Wan AI Platform so that moment stops happening.

Not because the technology is magic—it’s not. It’s the result of some incredibly smart engineering from the Wan AI team. But access shouldn’t be the hard part. The hard part should be what you create once you’re there.

So if you’ve been piecing together workflows, juggling tools, or settling for watermarked exports because the clean version was behind a paywall you couldn’t justify—give this a try.

One platform. All the Wan models. No watermarks. No friction.

Just your ideas, turned into content, faster than you thought possible.

And honestly? That’s the way it should have been all along.

How a Simple Task Turned Into Blur-Image.org

yesanson — Sat, 14 Mar 2026 04:15:17 +0000

The idea for Blur-Image.org came from something surprisingly small.

One day I needed to blur a small detail in an image before sharing it. It was a simple task — just hide a piece of information and move on. I assumed I could quickly find an online tool to do it.

So I searched Google for “blur image.”

The results were not what I expected.

Most of the tools I found were very basic. Some could only blur the entire image. Others offered a simple blur brush but with very limited control. None of them felt like a truly capable blur image online tool that could handle real situations — like blurring faces, hiding private text, or precisely masking certain parts of an image.

After trying several of them, I still couldn’t get the result I needed.

Eventually, I did what many people end up doing: I opened Photoshop. For such a small task, installing or launching a large professional program felt excessive. It worked, of course — but it also felt like using a massive machine to solve a tiny problem.

That moment stuck with me.

Modern web applications are incredibly powerful. Today you can edit videos, collaborate on design files, and run complex AI tools directly in your browser. Yet something as common as blurring images properly online still felt incomplete.

It made me wonder:

Why isn’t there a truly professional blur image tool that runs entirely in the browser?

Instead of continuing to search for one, I decided to try building it.

Building the Tool I Wanted to Use

Blur-Image.org started with a simple goal: create a tool that makes blurring images online both powerful and effortless.

Not just a single blur filter, but a set of tools that could handle the real reasons people blur images.

Over time, that vision turned into a complete toolkit.

Selective Blur

Often you only want to hide specific parts of an image. Selective Blur allows you to manually blur exactly where you want.

You can paint blur regions with brush and selection tools, adjust the blur strength, and apply multiple blur areas in one image. This makes it ideal for hiding faces, objects, or sensitive details.

Full Image Blur

Sometimes you want a simple, clean background blur or stylistic effect. With Full Image Blur, the entire image can be blurred instantly with adjustable intensity and smooth, natural output. You can even restore selected areas afterward.

Face Blur

For privacy protection, faces are often the most important element to hide. The Face Blur tool automatically detects faces in an image and applies blur or pixelation. It supports multiple faces and allows manual adjustments when needed.

Sensitive Information Blur

Images frequently contain private information — phone numbers, emails, addresses, or identification numbers. Blur-Image.org can detect this text automatically using OCR that runs locally in your browser. The detected areas can be edited or removed before applying blur.

Most importantly, no image data ever leaves your device.

Text Blur

For screenshots or document images, Text Blur automatically detects text regions so you can blur or pixelate them quickly and precisely.

Pixelate Image

In some situations, pixelation provides stronger anonymization than blur. The Pixelate tool lets you adjust pixel size and apply it to the entire image or selected regions.

Motion Blur

Blur can also be creative. Motion Blur adds directional blur effects with adjustable angles and distances, powered by GPU acceleration for smooth results.

Built for Privacy and Simplicity

From the start, one design principle guided the project: everything should happen in the browser.

Blur-Image.org is:

100% browser-based
No image uploads
No account required
Instant preview and export

Your images never leave your device, which keeps the process both fast and private.

From Frustration to a Useful Tool

Blur-Image.org wasn’t built because I set out to create a big product. It started with a small frustration — a task that should have been simple but wasn’t.

Sometimes the best tools come from exactly that kind of moment.

Today, Blur-Image.org aims to be the tool I originally searched for: a professional, complete, and easy-to-use blur image online tool that works instantly, directly in your browser.

Because blurring an image shouldn’t require installing heavy software — it should just work.

Why I Built RGBtoCMYK.net

yesanson — Tue, 03 Mar 2026 03:56:43 +0000

RGBtoCMYK.net started in a way that many good tools do — with a mistake.

A few years ago, I was working on a print project that looked absolutely beautiful on my screen. The colors were vibrant, balanced, carefully adjusted. I had double-checked everything.

When the printed version arrived, my heart sank.

The colors were technically “correct,” but emotionally wrong. The warmth was gone. The depth was reduced. What felt alive on screen felt restrained on paper.

That was the moment I stopped treating RGB-to-CMYK conversion as a checkbox task and started taking color management seriously.

I began digging deeper into ICC profiles, rendering intents, device color spaces, paper characteristics — the parts most people never talk about but that quietly determine whether a design succeeds in print.

Naturally, I looked for online tools to simplify the process.

But what I found was discouraging.

Most RGB to CMYK websites were clearly built for quick, surface-level conversions. They rarely supported custom ICC profile uploads. If they did, profiles were sometimes misread or applied incorrectly. Rendering intents were often ignored. The results were inconsistent.

They were convenient — but not dependable.

And when you’re sending files to a printer, “almost correct” isn’t good enough.

So I started thinking:
Why is there no online tool that treats color conversion with the same respect as professional desktop software?

Modern browsers are powerful. Web technologies are mature. There was no real reason a browser-based tool couldn’t handle proper color management workflows.

That thought turned into experiments.
Experiments turned into prototypes.
Prototypes turned into RGBtoCMYK.net.

From the beginning, the goal was simple:

Build an online RGB to CMYK converter that professionals could actually trust.

Real ICC profile support.
Accurate profile parsing.
Proper rendering intent handling.
Clean previews.
No shortcuts disguised as “simplification.”

I didn’t build it to compete with heavyweight creative software. I built it because I wanted a reliable, professional-grade solution that was accessible anywhere — without installs, without friction, without sacrificing accuracy.

RGBtoCMYK.net exists because color deserves to be handled properly — even online.

And because no one should open a print package and feel that quiet disappointment again.

Heads or Tails? The Story of Building a Better Coin Flip Simulator

yesanson — Sat, 11 Oct 2025 02:38:34 +0000

It all started with a simple, age-old need: I needed to flip a coin.

It wasn't for anything monumental, just a friendly debate with a colleague over who would pick up the coffee tab. I went online, typed "coin flip," and was met with a sea of functional, but utterly soulless, websites. A pixelated, generic silver coin spun on my screen. Click. It landed on heads. The experience was… underwhelming. It felt like using a basic calculator when you were hoping for something more.

That's when the thought struck me. For centuries, the simple act of a flip a coin has decided fates, settled arguments, and sparked games. It's a tiny moment of suspense and pure, unadulterated chance. Why was its digital counterpart so bland?

Driven by this minor frustration, I dug deeper. What if you needed to decide between more than two options? What if you were a game master needing to simulate multiple events at once? I searched for a site that could let me flip a coin ten, or even twenty times, simultaneously. I found nothing. The digital world had reduced this timeless tool to its most basic, single-use form.

And so, the idea for coin-flip.vip was born. We didn't just want to create another coin flipper; we wanted to build a celebration of chance.

Our first mission was power. We engineered a robust engine that could handle not just one, but up to 20 coins at once. Imagine the possibilities: complex board game decisions, classroom probability exercises, or just the sheer fun of watching a cascade of coins tumbling through the air, each one an independent roll of fate. Suddenly, the simple "coin flip" became a powerful, multi-faceted tool.

But power without beauty is empty. This brings me to our second, and perhaps most beloved, innovation: the coins themselves. We asked, "Why should a digital coin be a boring, flat circle?" A real coin has weight, history, and artistry. We wanted to capture that.

Our team scoured numismatic books and historical archives. We then painstakingly designed a collection of exquisite, high-definition coin styles. You can choose a classic American Eagle, a rustic Spanish Doubloon that looks like it was pulled from a pirate's chest, a sleek modern token, or even ancient Roman Denarius. Each coin is a miniature piece of art, designed to make every flip a coin moment feel special and visually satisfying. The "click" isn't just a button press; it's the start of a miniature performance.

Coin-flip.vip is the result of that journey. It’s the website I wished I had found that first day. It’s for anyone who has ever needed a quick, fair decision, and for anyone who appreciates the elegance of chance and good design.

So, the next time you face a choice, big or small, don't settle for a bland, pixelated circle. Come to coin-flip.vip. Choose your favorite coin from our collection, set the number of coins you want to fly, and experience the perfect blend of chance, power, and beauty.

Let fate decide, in style.