OG Paper
An open-source, truly endless canvas — infinite pan, infinite zoom — that runs on every device and saves your work in a format you own.
Write a sentence inside the dot of an "i", then zoom out until a whole notebook is a speck. Pick up the same canvas on your Windows laptop, Android phone, iPad or a browser. And if this project ever stops, your files still open: the format is openly documented, stored in SQLite, and readable with a tiny standard-library Python script.
Status: Phase 1 (the app), early alpha. Drawing, files and the web app work on Windows, macOS, Linux, Android and the browser — see What works today. The Phase 0 spike flies from 10^-0.6 to 10^49 zoom over a 1M-stroke canvas at ~6,000 fps on desktop (docs/PHASE0.md). Design spec: docs/DESIGN.md.
Why
Proprietary infinite-canvas apps such as Endless Paper keep your work in undocumented formats, on one platform, behind a subscription. If the company changes course, years of notes can become unreadable. OG Paper exists so that can't happen.
Goals
- Truly endless — unbounded zoom depth, not just 0.1×–10×. Only what's visible and big enough to see gets drawn.
- Everywhere — Windows, macOS, Linux, Android, iOS/iPadOS and the web, on phones, tablets and desktops.
- Fast — 2024 flagship phones are the performance bar: 120 fps pan/zoom, low-latency ink.
- Files you own — one
.ogp file per canvas, open spec, plus SVG/PDF/PNG/JSON and self-contained HTML exports.
- Sync without a company — multi-device sync through a folder you already sync (iCloud, Google Drive, Dropbox, Syncthing) or a relay you self-host for free.
Try it
- Try mode: https://omegagiven.github.io/OG-paper/try/ — the real app on a demo canvas 10^45 deep, with a guided tour.
- Web: https://omegagiven.github.io/OG-paper/app/ — then “Install app” / “Add to Home screen” for full screen.
The web app autosaves in your browser and has bookmarks (fly to any saved view), a timeline (every stroke is
time-stamped; scrub, replay or restore any moment) and offline copies (download / load a
.ogpt file).
- Windows, macOS, Linux, Android: releases page
(every push to
main also produces builds under the CI run's artifacts).
Run your own server
One small container hosts all your pages on your own hardware, so everyone
on your network (or tailnet) can open and draw on them together, and each
device keeps its own copy that syncs when it reconnects. It needs no
account and no cloud.
The short version (any machine with Docker)
git clone https://github.com/OmegaGiven/OG-paper.git
cd OG-paper
docker compose up -d --build # first build takes a few minutes
docker compose logs og-paper # copy a "Server link" from here
The log shows two server links: one that can make and change pages, and
a view-only one. Treat them like passwords. Your pages live in
./og-paper-data (mounted at /data in the container). The server listens
on port 8991.
Without compose:
docker build -t og-paper .
docker run -d --name og-paper --restart unless-stopped \
-p 8991:8991 -v /path/on/disk/og-paper:/data og-paper
docker logs og-paper
The links name the container's internal address. To get links with the
address people actually use, add --public ws://<your-server-ip>:8991 to
the command (see the commented command: line in
docker-compose.yml), or just swap the address in the
link by hand. Browsers on the public web app (an https:// page) can only
connect to a wss:// address. Put the server behind a TLS proxy (Tailscale
serve, Caddy, Nginx Proxy Manager, Traefik) and use --public wss://your-host. The desktop and Android apps connect to plain ws://
fine.
On a NAS or home server
There is no prebuilt image yet, so the image is built from this repository
once (on the NAS itself, or on a PC and then copied over, see below).
- Synology (DSM 7.2+, Container Manager): copy the repository to a
shared folder (e.g.
/volume1/docker/og-paper). In Container Manager >
Project > Create, choose that folder as the path; it picks up
docker-compose.yml. Build and start it. Open the container's Log
tab for the server links. To use a NAS folder for the pages, change
./og-paper-data to e.g. /volume1/docker/og-paper/data.
- QNAP (Container Station 3): Applications > Create, paste the
contents of
docker-compose.yml (with build: . replaced by a built
image, see below, or create it from the repository folder over SSH with
docker compose up -d --build). Map /data to a folder such as
/share/Container/og-paper.
- Unraid: over SSH,
git clone into /mnt/user/appdata/og-paper-src and
run docker build -t og-paper . there. Then Docker > Add Container:
Repository og-paper, port 8991 to 8991, path /data to
/mnt/user/appdata/og-paper. The server links are in the container's log.
- TrueNAS SCALE: build the image in a shell (
docker build -t og-paper .), then Apps > Discover Apps > Custom App: image og-paper (pull
policy: never), port 8991, host path for /data on a dataset (e.g.
/mnt/tank/apps/og-paper).
- Raspberry Pi, a Linux box, a VPS: the short version above works as is
(64-bit Raspberry Pi OS / any arm64 or x86-64 Linux with Docker).
- Without Docker: download (or
cargo build --release -p og-paper) the
Linux build and run og-paper --serve-dir ~/og-paper-pages. A systemd
service with Restart=always keeps it up.
Building on a PC for a slow NAS: build for the NAS's CPU, then load it
there:
docker buildx build --platform linux/arm64 -t og-paper --load . # or linux/amd64
docker save og-paper | gzip > og-paper.tar.gz
# copy og-paper.tar.gz to the NAS, then on the NAS:
docker load < og-paper.tar.gz
Open port 8991 on the NAS firewall for your LAN. For access from outside, a
VPN or Tailscale is safer than opening the port to the internet.
Open the server's address in a browser (http://<server>:8991/) for a page
that shows people how to connect, with the link to copy.
To let an AI assistant work on your pages, add
http://<server>:8991/mcp?k=<server key> to it as an MCP server; see
docs/MCP.md.
Connect to a page
Every way works from the ⚙ settings menu:
- Your server's pages: ⚙ > Pages > paste a server link under
Servers > Add. Its pages appear: Open one and you're drawing on
it with everyone else; + Page makes a new one. The app keeps a copy
of each page you open; reopening it from On this device reconnects by
itself, and anything you drew offline goes up.
- One page, live, no server: ⚙ > Share live > Host this canvas
(desktop) and send someone the edit or view link; they paste it under
Join (or open the browser link). In a browser: Host in this
browser makes one-time invite links.
- Swap copies: ⚙ > Export > OG Paper copy, send the file, and they use ⚙ >
Merge copy: both sets of changes combine. Sync folder does this
automatically through Syncthing, Dropbox, Drive or a USB stick.
More detail, including the relay for people who are never online at the
same time and how the end-to-end encryption works, is in
docs/SHARING.md.
What works today
- Brush: a simple line (width, pressure, solid / dashed / dotted, opacity), or Advanced: a GIMP-style brush engine with 15 looks (pencil, charcoal, bristle brush, watercolor, airbrush, spray, calligraphy, sketchy, jagged, neon, chain, stitches, felt tip, rainbow, ink) and every setting behind them: 13 tip shapes, hardness, spacing, angle, roundness, pressure and speed dynamics, taper, fade, scatter, jitter, sketchy wobble, flow, paper grain, color jitter, hue cycle and fade-to color. Strokes stay vector and sharp at any zoom.
- Texture: splotches, spatter, leaves, stars, hearts, patterns (dots, hatch, weave, grain, ...), paper grain and sponge, with the same engine settings.
- Ink tools: highlighter, stroke eraser, eyedropper, color dial with custom colors, unlimited undo/redo.
- Bucket fill: tap inside an outline to fill it with real vector ink tucked under the lines (sharp at any zoom), optionally closing small gaps in the outline.
- Shapes: one Shapes tool with rectangle, ellipse, diamond, triangle, star, polygon, line and arrow, and Excalidraw-style options: hachure / cross-hatch / zigzag / solid fill, fill color (or one color for stroke and fill), stroke width and style, sloppiness (clean to cartoon), sharp or round edges, curved and elbow lines, nine arrowheads, opacity.
- Text: 11 bundled open-licensed fonts (hand-drawn, marker, sans, serif, mono, display) plus three single-stroke fonts, or add your own
.ttf / .otf; sizes, alignment, color, opacity. Text is ink (letters are filled outlines), so it stays sharp at any zoom.
- Pictures, PDFs and tables: paste (Ctrl+V or the Paste button, for phones), drag-drop or insert pictures (PNG, JPEG, GIF, WebP, SVG) and crop them in place (non-destructive); import PDFs, each page a picture to write on; paste spreadsheet cells or a Markdown table and get a real table you can restyle and edit; pasted text becomes text.
- Library: save any selection as a sticker and place copies of it on any canvas, at any zoom.
- Diagrams: in diagram mode, lines and arrows snap to shapes, texts and pictures and follow them when they move.
- Saved tools: Minecraft-style toolbars (keys 1–9) of tools with their settings: a red 2 px pen, a dashed arrow, a font at a size... Keep as many named toolbars as you like and switch between them (
[ / ]), plus an inventory that grows as you fill it. Tap an empty slot to save the current tool.
- Select and edit: tap, drag a box or draw a lasso to select; move, resize, rotate, flip, duplicate, delete, bring to front / send to back, copy / cut / paste, and restyle anything from the tool panel; double-tap a text or table to edit it. Each edit is one undo step.
File formats (.ogp 0.3, shapes, text, tables and pictures, the timeline log, bookmarks, .ogpt v3) are documented in docs/DESIGN.md.
Coming next
- Save to / open from the cloud: Save and Open fan out to device, Google Drive, Dropbox or OneDrive, signed in from the page with no project server.
.ogpt copies in the desktop app, and .ogp in the web app, so both share one format; bookmarks and the timeline on desktop.- Vector SVG pictures, textured brushes, layers, frames and presenting — see
docs/PHASE1.md and the to-do list.
Stack
Rust core · wgpu renderer · winit · egui · SQLite. See docs/DESIGN.md for architecture, file format, sync design and roadmap.
Contributing
The repo is public from day one and help is welcome — especially people with stylus hardware (Apple Pencil, S Pen, Wacom) who can test ink feel. See CONTRIBUTING.md. Commits must be signed off (DCO).
License
Created by OmegaGiven. See AUTHORS and TRADEMARKS.md.
