Open Source & Third-Party Licenses

Last updated: September 20, 2026

PixlReady is built on the work of the open-source community. This page lists the third-party libraries, fonts and character maps that are actually bundled and shipped with this website and run in your browser, together with their versions, purpose and licenses. Tools used only to build the site are not listed because they are never distributed to you.

1. Software bundled with PixlReady

Every component below runs entirely on your own device; none of them uploads your images or PDFs.

Component	Version	What it does	License
PDF.js `pdf.min.js, pdf.worker.min.js`	3.11.174 (legacy)	Renders and previews PDF files inside the browser for the PDF tools.	Apache License 2.0
PDF.js CMap files `vendor/pdfjs/cmaps`	Shipped with PDF.js 3.11.174	Character maps that let PDFs display Chinese, Japanese and Korean text correctly.	BSD 3-Clause (Adobe)
Liberation fonts `LiberationSans-*.ttf`	Shipped with PDF.js 3.11.174	Metric-compatible standard fonts used as fallbacks when a PDF embeds no fonts.	SIL Open Font License 1.1
PDFium / Foxit standard fonts `Foxit*.pfb`	Shipped with PDF.js 3.11.174	Base-14 PostScript fallback fonts for rendering PDFs that rely on standard fonts.	BSD 3-Clause (PDFium Authors)
jsPDF `jspdf.umd.min.js`	2.5.1	Generates and downloads PDF documents (for example, printable ID-photo sheets) on the client.	MIT License
JSZip `jszip.min.js` bundles pako (MIT)	3.10.1	Creates and reads ZIP archives in the browser for batch download of multiple images.	MIT License^*
qrcode-generator `vendor/qrcode/qrcode.min.js`	1.4.4	Generates the share QR code entirely in your browser (no external QR API); the QR image can be downloaded as PNG.	MIT License
heic2any `heic2any.min.js`	0.0.4	Converts Apple HEIC/HEIF photos to JPEG or PNG directly in the browser.	MIT License
Lucide `lucide.min.js`	1.47.0	The open-source icon set used for the tool icons across the interface.	ISC License
psd.js (npm: `psd`) `psd.min.js`	3.x	Parses Adobe Photoshop PSD files so they can be opened and converted in the browser.	MIT License
MediaPipe Tasks Vision `vendor/mediapipe/vision_bundle.mjs, wasm/*`	@mediapipe/tasks-vision 1.0.1	On-device WASM/WebGL inference runtime that powers the on-device AI features (person/background segmentation and face-aware photo framing) used by the background-removal and ID-photo tools. It runs entirely in your own browser tab.	Apache License 2.0
MediaPipe Selfie Segmenter model `models/selfie_segmenter.tflite`	Square 256×256, float16	The segmentation model weights that produce the person/background mask. They are self-hosted on this site, downloaded once and cached; inference and the image stay on your device and no image is uploaded.	Apache License 2.0 (Google)
MediaPipe Selfie Multiclass Segmenter (AI Beta) `models/selfie_multiclass_256x256.<hash>.tflite`	Square 256×256, float32, six classes (background / hair / body-skin / face-skin / clothes / others)	Opt-in AI Beta segmentation model that gives hair its own class for a cleaner edge on flyaway hair. Off by default and loaded only when the user enables AI Beta; self-hosted, cached and run fully on your device, no image is uploaded.	Apache License 2.0 (Google)
MediaPipe BlazeFace short-range face detector `models/blaze_face_short_range.tflite`	Short-range, float16	Locates a face in the ID-photo tool so the picture can be auto-framed (head size and centring). It outputs only a bounding box, a few landmarks and head tilt used for framing; it does not recognise or identify anyone, builds no faceprint, and judges no emotion or sensitive attribute. Self-hosted, cached and run fully on your device; no image is uploaded.	Apache License 2.0 (Google)

^* JSZip is dual-licensed under the MIT License and the GPL version 3; PixlReady uses it under the permissive MIT License. Its embedded compression library pako is also MIT-licensed.

2. License texts shipped on this site

The full license notices for the PDF.js font and character-map data are bundled with the site and remain publicly available here:

The complete Apache License 2.0 notice for the PDF.js JavaScript code is embedded in the header of pdf.min.js and pdf.worker.min.js (the @licstart block). The MIT and ISC licenses above preserve their copyright and permission notices inside the corresponding bundled files; their canonical texts are linked from the license names in the table.

3. On-device AI models

PixlReady ships a few small on-device machine-learning models, all running on the MediaPipe Tasks Vision WASM/WebGL runtime and all self-hosted on this site (under /vendor/mediapipe and /models); they are downloaded once and then cached by your browser:

Google MediaPipe Selfie Multiclass Segmenter (AI Beta) (selfie_multiclass_256x256.tflite, square 256×256, float32) is an opt-in model that treats hair as its own class for a cleaner cutout edge. It is off by default and fetched only when the user enables AI Beta.
Google MediaPipe Selfie Segmenter (selfie_segmenter.tflite, square 256×256, float16) is the lightweight single-class person/background mask used when AI Beta is on but the multiclass model is unavailable.
Google MediaPipe BlazeFace short-range face detector (blaze_face_short_range.tflite, float16) locates a face in the ID-photo tool so the picture can be auto-framed. It returns only a bounding box, a few landmarks and head tilt for framing.

All inference runs locally in your own browser tab. Your photos are never uploaded, stored, logged or used to train any model, and no cloud AI service is called. Segmentation assists image editing only: it does not identify or verify any person and performs no faceprint, emotion or sensitive-attribute detection. The face detector is not face recognition: it outputs framing geometry only, cannot tell who someone is, and stores no face template. Results are AI-assisted and may need a manual touch-up. If any additional on-device model is introduced in the future, its name, version, source and license will be added to this page before it is released.

4. Trademarks

All product names, logos and brands (including HEIC/HEIF, PDF, Adobe, Photoshop and others) are the property of their respective owners. Their mention is for identification and compatibility purposes only and does not imply endorsement or affiliation.

5. Corrections & contact

We aim to keep these notices accurate. If you are an author of a bundled component, or you believe an attribution or license entry is incorrect or incomplete, please contact [email protected] and we will correct it promptly.