BLAYDE MANUAL // HOW THIS WORKS

What is this, actually?

Old manuals for a vehicle, appliance, or hobby tool often have terrible photos: blurry scans that get worse each re-scan. This project replaces them with real photos from real owners, patched straight into your own copy.

Is this legal? Manuals are copyrighted, right?

Yes, and that shaped almost every decision here. Nothing here ever contains the manual's own copyrighted photos or text, only structure (where a photo belongs). Photos are contributed and licensed by real people, then patched into your own copy, on your own machine. Nothing is uploaded or redistributed.

Hasn't someone already done this?

Not quite. Paid services like Cyclepedia shoot their own manual photos professionally, but they're closed, centralized, and scoped to one category. Nobody's built the open, crowd-sourced version that works the same way across a garage, a marina, a hangar, or a hobby bench. That's the gap this fills.

Was AI used in making this?

Yes, openly. This project was built with Claude (an AI assistant) doing the coding and drafting, guided by a human who made every real call: what to build, which tradeoffs to accept, what to reject. Nothing is hidden: see the full public ledger at ledgers/TheBlayde_AILedger.md, including where the human caught real AI mistakes.

Is AI doing something when it patches my manual?

No, on purpose. Building the code used AI; running it doesn't. Patching is plain, predictable software: it reads pixel patterns to find where a photo goes, reads text via OCR to label it, and swaps the image in. No AI model, no server, nothing uploaded. It works the same way today and in five years.

What does each role actually look like?

• Use. Patch your manual. That's it, no account or extra steps needed.
• Contribute. Submit straight from your already-patched manual, then track it in the Contributor Portal.
• Maintain. Its own Maintainer Portal: review incoming photos, invite others to help, or index a manual nobody's started.

Can I hold different roles on different manuals?

Yes, and that's normal. You can use the tool to patch any manual, no strings attached. Contribute a photo to one you care about, or step up further and review submissions on a manual you've asked to help maintain. Each role is scoped to that one manual, not to you as a person.

Where do photos live, and what gets stripped out?

Photos live as ordinary image files stored alongside that manual, nothing fancier than that: no separate photo service, no database. Metadata like GPS coordinates, camera model, and timestamps is all removed on submit. Your photo is stripped to just the pixels before submission, and checked again before it's available for patching into manuals.

Can I keep my photos private and build my own personal manual?

Not yet, though the pieces are already there. You can already save a photo as a draft on your own device. You just can't use it to patch your manual privately. Today, patching only pulls from the shared, publicly-approved pool. A private-manual option, where your own unsubmitted photos enhance just your copy, is planned for a future version.

What if I lose track of what I've already photographed?

That's a real, planned feature. When you're working through several photos in one session, the app shows a live checklist of everything in that section (what's done, what's still needed), so you're never left wondering "wait, did I already do that one?"

Do I need a computer, or can I do this all from my phone?

Our core goal is a lightweight, browser-based tool that works across any device connected to the internet, phone included. No download, no install. That piece is still being designed, but it's a core goal, not an afterthought.

New manuals

One org member spot-checks each one: is the source real, does the indexing look right, was it given an honest pass. That's the whole approval, not unanimous, so nobody becomes a bottleneck. Once approved, whoever requested it becomes the first maintainer, with the same privileges as anyone who joins later.

What if my manual has more than one edition, like a factory manual and a Haynes manual?

Both live under the same manual entry, with the same maintainers. The two scans just don't share photo coordinates. A genuinely different model, different year or variant, becomes its own separate manual entry instead, with its own maintainer team.

What if I go quiet, or can't keep up?

Try to hand things off if you're stepping back. If you can't, that's fine. A manual with no active maintainer still works exactly the same for anyone patching their own copy. A visible "needs a maintainer" label is planned, not built yet.

How many maintainers should a manual have?

Somewhere around 2 to 5 active ones is the healthy range. One is a single point of failure. Past five or so, ownership starts feeling like nobody's job, and a new volunteer is more useful on an understaffed manual than stacking onto one that already has coverage.

Is this free?

Yes, and that's not just a nice idea. It's the reason contributing works at all. Nobody's going to photograph their own brake job for a paywall. Using it needs no account. Contributing needs a free GitHub account. No cost either way.

What stops a big company from just taking this and selling it?

They could legally reuse credited photos (that's what an open license allows, on purpose), but they can't use the Blayde Manual name or take over the actual living community, and a static copy goes stale the moment it's copied, since it stops getting new contributions. The code itself is also licensed (AGPL) specifically so a company can't quietly turn it into a closed paid product without giving their changes back.

Could this ever make money?

If it ever does, it's more likely to look like affiliate links to real parts and tools next to a procedure (the way iFixit works) than a paywall on the manual itself. The free core stays free either way.

Won't storing photos for hundreds of manuals get expensive or slow?

No. Each manual is stored separately and sized independently, not lumped into one giant shared pool. Storage limits apply per manual, not across the whole project, so adding a 500th manual never makes an existing one closer to any kind of limit.

Why not just build an official database?

No official database, and that's a deliberate choice, not a gap. A rigid database would risk excluding exactly the older, rarer manuals this project exists for. Instead, brand, model, and (for a vehicle) year range come straight from the manual itself, checked by a real person before approval, whether or not that manual even has a model year.