Skip to content

BoothKernelA modern POS for doujin booths

No internet needed · Minimal workflow · Ready out of the box

Booth-Kernel Logo

⭐ Open source: auditable, reusable, hackable

The project is 100% open source (MIT License). Read the code, file issues, send PRs, or fork it and adapt it for your own circle.

📖

Development Timeline

From prototype website to a complete booth app

📖
Aug 2025
Problem: chaotic paper records

Taking over a doujin club during summer, we faced messy inventory and paper-based bookkeeping — inspiring an automated tool.

🧪
Sep–Nov 2025
Prototype: Web service (Flask)🏛️ View early code archive

Built a Flask + Vue prototype and deployed to a cloud server. It validated the workflow but hit a heavy network dependency wall.

💾
Nov 2025
Attempt: Nuitka packaged app

Tried shipping the Python backend as a packaged app. Size bloat and runtime deps made it hard to run on regular member devices; laptops were inconvenient to carry.

🚀
Dec 2025 – May 2026
Refactor: Tauri local application

Rewrote the backend in Rust, built with Tauri v2. Achieved high performance, zero external deps, and fully offline on Windows/Android.

📒
Sep 2026
v1.2: double-entry ledger

Rewrote the ledger with double-entry bookkeeping — every item and every yuan has a source and a destination. Added bundle deals, closing & consignment settlement, and barcode scanning, and redesigned the Admin console and Vendor view around the event lifecycle.

Use Cases
Comic conventions / offline events
Network
LAN only (phone hotspot works)
Data Storage
Local (exportable to Excel)
License
MIT · Free forever

🚫 Why did we build Booth-Kernel?

We tried many approaches running a fan club booth and hit these pitfalls.

📝

Paper-based bookkeeping

  • Error-prone math: Bundle pricing is hard to calculate under pressure; even calculators slow you down.
  • Inventory mistakes: Juggling cash, handover, notes, and stock marking leads to errors quickly.
  • Hard to review: Messy notes make post-event reconciliation and analysis painful.
🌐

WeChat/Alipay mini-program ordering

  • Network dependent: Convention networks are complex; heavy assets often fail to load fast enough.
  • Complicated setup: Certification and configuration are high-friction; every event needs fresh forms.
  • Split ecosystems: WeChat and Alipay payment/data are not interoperable — fragmented experience.
  • Poor visibility: Mini-program QR codes are easy to miss; customers can’t find them.
☁️

Web-based booth system (early attempt)

  • Data risk: Centralized servers are sensitive in multi-club contexts; product and sales data must stay private.
  • Complex deployment: Cloud servers, domains, HTTPS — unfriendly and costly for non-tech users.
  • Offline failure: Server or network issues can halt your booth entirely.
🆕 v1.2.0

📒 From "keeping records" to "keeping the books straight"

v1.2 rewrote the ledger: every item and every yuan has a source and a destination.
Bundle deals, refunds, closing stocktake and consignment settlement are all built on top of it.

🎁

Bundles find the best deal

"Any 3 cards for ¥20", "New book + standee for ¥50" — the cart works out the cheapest combination and shows how much the customer saved.

🧾

Close in four steps, one settlement sheet

Recount orders → Stocktake → Take back → Settle. How much each circle gets and how much you owe whom, all at a glance.

▦

Scan to add to cart

Scan barcodes continuously with the camera, or just plug in a barcode scanner. Works with retail barcodes and your own number labels, fully offline.

🧭

Organized around the event lifecycle

The Event workbench is grouped into Pre-event / On-site / Closing; "Import from previous event" carries over products and bundles in one go.

⭐ New in v1.1.0

📸 AI photo recognition — tuned for anime & doujin goods

Point the camera at an item, take a photo, and it's recognized and added to the cart.
No barcode stickers, no SKUs to memorize.

🖼️

No labels needed

Doujin goods come in small runs and lineups change often; stickering every item is tedious and ugly. The product photo itself becomes the "label".

🤝

Easy when minding a friend's booth

Don't know your friend's products? No problem — one photo gives you the name, price and stock. Zero learning curve.

⚡

Customer self-checkout

Put a tablet at the front of the booth: customers snap, add to cart and pay by QR code themselves. You just collect the money.

🔌

Local inference / hardware acceleration

Once the model is downloaded, inference runs fully offline. The desktop app supports DirectML GPU acceleration; Android runs on CPU.

✅ Our answer: Local app + LAN sync

Booth-Kernel centers on local data + LAN synchronization: enable phone hotspot, connect devices, and collaborate — rock solid even offline.

📶 Offline LAN architecture

No internet needed. Turn on phone hotspot (no data usage) → connect devices → sync instantly.

⚡ Workflow built for vendors

Order → Confirm → Fulfill; the system automatically deducts stock, logs sales, and generates reports.

🎨 Easier for customers to order

Tablet as "digital menu": big images, clear pricing, quick ordering — less friction and crowding.

🔒 Data stays under your control

No centralized cloud. Data stays on your devices, with one-click Excel export for review.

🔄 How does it run on-site?

Two devices are enough for a full workflow; multiple tablets/phones scale easily.

Admin (PC / tablet / phone)
Admin console
Product library · Events & bundles · Order stats · Settlement sheet
Customer (tablet)
Self-service ordering
Large images · scan / photo to add · QR payments
Vendor (phone)
Fulfillment notifier
Order alerts · confirm payment · refunds · closing

🎨 UI Preview

Two landscape shots (customer/statistics) and one portrait (vendor).

Statistics dashboard screenshot
On-site stats: sales trend and per-product sales
Customer ordering screen screenshot
Customer: self-service cart; bundle deals apply automatically and show how much you saved
Vendor picking screen screenshot
Vendor: order alerts, payment confirmation, refunds and closing

Make selling fun again

Let the system handle the tedious parts — keep the passion for fans.

100% open source · MIT · Free forever