背景ぼかし
被写体はくっきり残しつつ背景だけをぼかします。AI 自動検出または手動の枠で範囲を指定し、ぼかし強度とエッジのぼかし量を調整できます。
AK
appkiro.comUpload Image
Add your image
Adjust Blur
Customize blur settings
Download
Download your image
Drag & drop your image here
or click to browse
Supports: JPG, PNG, WebP, HEIC
Max file size: 20MB
AI segmentation finds the subject. The first run downloads the model (~80 MB).
AI Powered
Advanced AI for accurate subject detection
Privacy First
Images stay on your device
High Quality
Native blur, native filters
Fast Processing
Realtime preview while you adjust
Blur Background isolates the subject of an image and softens everything behind it. The preview updates in real time as you tweak intensity, feathering, and the foreground brush, so you can dial in the shot before exporting.
Auto Detect uses an in-browser segmentation model (RMBG-1.4 from Hugging Face) to find the foreground. Manual Select lets you drag a rectangle to define the keep-sharp zone — handy when the scene confuses the AI or you want creative framing.
Blur Intensity sets the strength in pixels. Brush Size controls how far the foreground mask extends so subjects with hair or fluffy edges keep a soft halo. Edge Smoothing feathers the mask boundary so the transition feels natural instead of cut-out.
Files stay on your device. The preview, mask, and export all run on your GPU/CPU. For other privacy-respecting tools see Background Remover and Pixelate Image.
What does Blur Background do?
It keeps your subject sharp while softening everything else, mimicking the shallow depth-of-field of a portrait lens. The blur is applied locally in your browser, so the original photo never leaves your device.
Who benefits from this tool?
Photographers stylising portraits, marketers prepping product cards, content creators isolating talking-head subjects, and anyone wanting a cleaner Zoom or LinkedIn shot.
When should I pick Auto Detect over Manual Select?
Auto Detect uses an AI segmentation model to find the subject — perfect for people, pets, and most objects. Switch to Manual Select when the AI mis-reads a busy scene or when you want a precise rectangular focus area.
Where does the AI model run?
Inside your browser via @huggingface/transformers. It downloads the RMBG-1.4 weights once (~80 MB) and caches them in the browser HTTP cache so the next runs are instant.
Why does the result look different in JPG vs PNG?
JPG re-encodes with lossy compression and discards transparency; PNG and WebP preserve every pixel. For maximum quality and edge fidelity, export PNG or WebP.
How do I keep the file small?
Pick WebP at 90% quality — it produces noticeably smaller files than PNG with near-identical visual fidelity.