Looks like
you wrote it.

World's most advanced handwriting ecosystem.

writer_id: 0xA1 · 600 DPI match 91%
How it works

From a stack of paper to your handwriting

Step 1

Upload your pages

Scan 10–15 pages of your existing handwriting. No tablet, no stylus — a phone camera is enough.

Step 2

We reconstruct your strokes

Binarize, remove ruling, skeletonize, then recover stroke order and velocity from ink width.

Step 3

Write anything in your hand

Type text, pick a notebook, and PHS renders print-ready pages in your own handwriting.

Why it's different

Built in stroke space, not pixel space

Real pen physics

A stroke-space diffusion transformer models pen-lift, ligatures and micro-tremor. Fitts' Law velocity curves drive ink width and opacity like a real gel pen.

Canvas-aware layout

Page bounds, baselines and margins are known from the notebook template, so the layout controller makes compress-or-yield decisions on actual word bounding boxes.

Per-user LoRA

Your style lives in a 5–11 MB LoRA adapter — slant, letter forms, pressure behaviour. ~45 min to train, milliseconds to load on every generation.

No hardware required

An offline-to-online reconstruction pipeline turns scanned ink into pseudo-stroke sequences, so onboarding is just a stack of paper.

Templates

Write onto the notebooks you actually use

Classmate (A4 ruled)
Single-line ruled, the default Indian school notebook.
Navneet (long ruled)
Wider line spacing, common for fair copies and essays.
Plain A4
Unruled white sheet — baseline managed by the layout engine.
Grid / Math
5 mm grid for equations, diagrams and aligned work.