Edge Detect - Sobel, Real-Time
Image Tools
Drop an image or click to browse
JPG, PNG, WebP - runs locally, nothing uploads
What is Edge Detect - Sobel, Real-Time?
How it works
Features & Benefits
- Sobel kernel runs on the GPU via a WGSL fragment shader - interactive even on large images
- Threshold and edge thickness adjust live without re-uploading
- Optional invert mode produces clean black-ink-on-white line art
- Works on any image the browser can decode; nothing is uploaded
Frequently Asked Questions
Why WebGPU and not Canvas?
Canvas-based edge detection works one pixel per JavaScript loop iteration. WebGPU runs all pixels in parallel on your GPU, so sliders are live instead of laggy on large images.
Which browsers work?
Chrome, Edge, and Firefox on Windows/Mac/Linux, plus Safari 26+. If WebGPU isn't available the tool shows a clear fallback message.
Does the image leave my device?
No. Decoding, GPU processing, and PNG encoding all run in your browser tab.
What kind of edge detector is this?
Classic 3×3 Sobel with magnitude thresholding. It is fast and predictable; for noise-robust edges consider a Canny-style pipeline, which we may add later.
Related Tools
Real-time exposure, contrast, saturation, warmth, blur, sharpen, vignette, and grain. GPU-accelerated via WebGPU; nothing uploads.
Per-channel RGB and luminance histograms computed on the GPU. Adjust black point, white point, and gamma with live preview.
Apply an anime painting style to your photos using AnimeGANv2. Works best on portraits with clear subjects. Runs entirely in your browser - nothing is uploaded.
Popular Utilities
Format, validate, and minify JSON instantly in your browser. Your data never leaves your device.
Decode JWT tokens and inspect header and payload instantly in your browser. Your tokens never leave your device.
Count words, characters, sentences, and estimate reading time instantly in your browser. No sign-up required.
Learn More & Guides
Sobel Edge Detection on the GPU: A WebGPU Worked Example
Edge detection is a textbook image-processing problem with a textbook solution. Implementing it on the GPU in WebGPU is also textbook - but the texts haven't been written yet. Here's the worked example.
9 min readReal-Time Photo Finishing in a Browser Tab: A Multi-Pass WebGPU Pipeline
A photo finisher needs nine sliders to do real work: exposure, contrast, saturation, warmth, gamma, blur, sharpen, vignette, grain. Doing all nine in real time used to require a desktop app. Here's how to do it in a tab.
10 min readImage Histograms and Levels: What Photoshop Has Been Doing Since 1990
The histogram is the single most useful diagnostic in image processing. The levels adjustment is the matching tool that lets you do something about what the histogram tells you. Both finally run in real time in the browser.
10 min readPixel Sort: How Glitch Art Became an Algorithm
Pixel sort is one of the few generative-art techniques with a clean algorithmic statement: sort runs of pixels by some key. The visual signature - smeared streaks broken by sharp untouched regions - comes from one tiny detail: the threshold.
9 min read