{"schemaVersion":"1.0","type":"Article","types":["Article"],"slug":"my-weekend-toy-got-a-second-weekend-google-still-can-t-find-it-sbjwq","url":"https://zyvop.com/my-weekend-toy-got-a-second-weekend-google-still-can-t-find-it-sbjwq","title":"My Weekend Toy Got a Second Weekend. Google Still Can't Find It.","subtitle":"Many floor plans on one canvas, a terrace result I didn't expect, and a name that belongs to paint booths","tldr":"Crossdraft v2 adds multi-board comparison, GIF export, and guide pages. Learn why the app still doesn't rank for its name due to paint-booth terminology.","keywords":["AI","developer tools","Web Development"],"entities":["Lê Đức Minh","AI Engineer","AI","developer tools","Web Development","ZyVOP"],"keyTakeaways":["💡 TL;DR: Crossdraft v2 expands its browser-based wind simulation tool to support multiple floor plans on a single canvas with independent solvers, revealing that a middle house without side windows achieved higher fresh air percentages than end houses with extra openings.","Despite implementing comprehensive SEO measures including guide pages and structured data, the project remains unranked for its own name due to a pre-existing industry term collision with paint booths.","The updated workspace allows users to compare multiple layouts side-by-side, where only the selected board runs at full speed to prevent performance issues, while the project format migrated to version 3 to maintain backward compatibility with older share links."],"headings":["Why did one board stop being enough?","What did the bigger examples show?","Why add GIF export to a fluid sim?","How do you make a tiny static app findable?","So did it work?","What I'd tell you before your second weekend","How to check this yourself","Frequently Asked Questions","Does an extra side window always improve ventilation?"],"outboundLinks":["https://crossdraft-alpha.vercel.app/","https://www.linkedin.com/in/minhle007/","https://github.com/MinLee0210","https://github.com/MinLee0210/crossdraft"],"contentText":"💡 TL;DR: Crossdraft v2 expands its browser-based wind simulation tool to support multiple floor plans on a single canvas with independent solvers, revealing that a middle house without side windows achieved higher fresh air percentages than end houses with extra openings. Despite implementing comprehensive SEO measures including guide pages and structured data, the project remains unranked for its own name due to a pre-existing industry term collision with paint booths. The updated workspace allows users to compare multiple layouts side-by-side, where only the selected board runs at full speed to prevent performance issues, while the project format migrated to version 3 to maintain backward compatibility with older share links. New larger examples, such as a three-house terrace, demonstrated that adding more openings does not always increase airflow, as side windows facing along the wind direction acted as leaks rather than inlets, reducing the pressure driving air through the main path. The author added GIF export functionality with a \"not CFD\" watermark to clarify the tool's limitations in shared media, yet search results from October 27, 2026, show that queries for the tool's name still return paint-booth content instead of the application. Have you ever shipped a side project, felt good about it for a day, then searched for its name and found... paint booths? That's where I am with Crossdraft, the browser sketchpad that shows how wind moves through a floor plan. The first weekend made it work. The second weekend made it something I'd actually use: many houses on one canvas, bigger examples, GIF export. Then I spent a chunk of that weekend making it findable, did everything right by the checklist, and checked the search results this morning. Crossdraft vputs several floor plans on one canvas, each with its own solver, and adds five bigger examples. In a terrace of three houses, the middle house with no side window flushed best. I also added guide pages, JSON-LD and llms.txt. Nine days later it doesn't rank for its own name, because \"crossdraft\" is a paint-booth term. Boards: each plan has its own solver, wind and undo history. The selected one runs live, the rest tick slowly. Terrace example: middle house 98% and 94% fresh air, end houses with a side window 78% to 85%. Tests went from 71 to 107, plan and section examples from 13 to 18. Search check on 27 October: Crossdraft isn't in the top five for \"crossdraft wind sketchpad\" or \"crossdraft airflow floor plan\". Why did one board stop being enough? Because the whole point of the tool is comparing. The first version had two result slots, A and B: test a layout into A, change something, test into B, squint at the numbers. Fine for one window. Useless when you want your parents' house as it is, with the back door closed, and with a screen at the entrance, side by side. So v2 is a canvas you can zoom and pan, closer to Excalidraw than to a form. You drop boards on it. Each board is its own little world: its own grid, wind, mode and undo history. You can duplicate a board, change one wall, and put the two side by side. Notes and arrows live on a layer above the boards, so you can write \"this is the version with the door closed\" right next to it. Every board carries its own solver. Only the selected board runs at full speed; the others keep ticking slowly so they don't freeze. The whole canvas saves as one project, into the share link or into the browser. The part that matters most is the boring one. Several fluid solvers in one tab is the obvious way to melt a laptop, so only the selected board runs at full speed. The others tick slowly, enough to stay alive, not enough to compete. The other boring part: the project format went to version 3 (boards, positions, notes), and every old share link and saved layout still loads. A link to the first version still opens. That cost a migration path and a pile of tests; across v2 and the guide-page release the suite went from 71 to 107 cases. One bug came out of that work: restoring or importing a project left the active board blank until you touched it. It's the kind of thing you rarely see in your own browser, because you rarely start from a saved project. What did the bigger examples show? I added five larger plans (a street canyon, a three-bedroom corridor house, an L-shaped house, an open loft, a terrace of three houses) and ran them headless for 60 simulated seconds, same setup as last time. The terrace is the one I didn't expect. All three houses are the same, front room and back room, wind from the west. The two end houses also get a side window. The middle house only has a front and a back. The app's note asks: is the extra window worth it? You'd guess yes. More openings, more air. Chart: fresh air per room in the terrace. The middle house, with no side window, flushes best. Wind from the west, 2 m/s, 60 s. House Front room Back room Mean speed, front room North end (side window) 82% 78% 1.12 m/s Middle (front and back only) 98% 94% 2.33 m/s South end (side window) 85% 80% 1.17 m/s My reading, not a measurement: with the wind coming straight down the row, the side windows face along the wind, not into it. They don't add a strong inlet. They add a leak, so the air entering the front room has two ways out instead of one, and less of it pushes through to the back. The middle house has one straight path and the air takes all of it. I'd want a 3D run before I told anyone to brick up a side window, but it's a nice example of \"more openings\" not being the same as \"more flow\". The corridor house told a story I ended up writing a whole post about: the front door and back door on one line, a fast jet down the corridor, and bedrooms beside it at 27% to 39% fresh air. That one's in the feng shui post. The street canyon is the fun one to watch. Wind comes in at 3 m/s and speeds up at the corners and through the cross street. The fastest single cell hits 9.14 m/s, about 3 times the incoming wind. I'd treat that number with care: single corner cells next to walls spike in this solver (that's the \"drama-queen cell\" from my first post), so look at the Speed view's overall pattern, not the maximum. Why add GIF export to a fluid sim? Because people share screenshots, and a screenshot of a flow field without its limits is how a toy becomes a \"simulation proves\" post. Record GIF captures 5 seconds of the canvas with \"not CFD\" drawn into the frames. You can crop it out, but you have to try. The encoder is plain TypeScript, no library, so the app stays a static site. How do you make a tiny static app findable? A single-page canvas app gives a search engine almost nothing to read. So I did everything on the list. Guide pages: what Crossdraft is, cross-ventilation basics, how the simulation works, feng shui and airflow, the examples, and an FAQ, each in English and Vietnamese. With the home page and the two guide indexes, that's 15 URLs in the sitemap. Canonical and hreflang tags so the two languages don't compete, Open Graph and Twitter cards, robots.txt, an llms.txt for AI crawlers, and JSON-LD (WebApplication and WebSite on the home page, TechArticle, FAQPage and BreadcrumbList on the guides). Plus a Guide link in the app menu so crawlers can reach any of it. The funny part: for years I wrote on Medium, which gives authors no way to add JSON-LD at all. On my own static site I control every tag, and that turned out not to be the bottleneck. So did it work? Not yet. This morning, nine days after that commit, I searched: Query Crossdraft in top 5? What ranks instead crossdraft wind sketchpad No Pinterest, Reddit ELI5, paint booth article, a generic sketchpad crossdraft airflow floor plan No Five paint-booth pages free airflow simulator floor plan browser No SimScale, a YouTube video, another web app, Quora, OpenFOAM mô phỏng gió thông gió nhà ống miễn phí No YouTube, Facebook, TikTok, an architecture firm's article The first two rows are the lesson. \"Cross draft\" is an established term in industrial paint booths, so even a query that includes my tool's exact name goes to booth manufacturers. I named the project after the thing it shows (a cross draft through a room) and never searched the name first. Nine days is short, and a new domain on vercel.app starts with no authority, so I'm not reading much into the generic queries yet. The name collision is different. That one won't fix itself with time. What I'd tell you before your second weekend Search your project's name before you commit to it. If an industry already owns the word, you'll spend your SEO budget fighting it. Keep old links working when you change the format. It's tedious, and it's the reason anyone still trusts your share links. Put the limits into the artifact people will share, not only into the page. The \"not CFD\" caption travels with the GIF. Run your own examples and actually read the numbers. The extra side window looked like an easy win. Structured data is cheap to add and easy to get right. Being found is a different problem. If you want the first weekend, the solver and the original nine examples, it's in Fun Project: I spent a weekend teaching my browser how wind works. And if you've solved the \"my project name is someone else's industry\" problem without renaming, tell me how. Share your thoughts in the comments — I'd love to hear how this technology is impacting your industry. Follow me: LinkedIn | GitHub How to check this yourself App: crossdraft-alpha.vercel.app. Source: github.com/MinLee0210/crossdraft at commit f9118e8. The v2 workspace is commit 9396643, the guide pages and structured data f9118e8. Example numbers: a headless Node script runs the solver on the new examples, npx -y tsx v2_examples.mts <path-to-crossdraft> 60 (about 12 seconds on a laptop CPU). Search check: run on 27 October 2026 through Firecrawl search. Results change daily. Test count: 71 test cases at cf5ad64, 107 at f9118e8. TypeScript lines across lib, components and app went from 3,086 to 7,390, though a good share of that is Prettier reflowing long lines. Frequently Asked Questions Does an extra side window always improve ventilation? Not in the terrace example. With wind blowing along the row, the middle house (front and back openings only) reached 98% and 94% fresh air, while the end houses with a side window reached 78% to 85%.","contentHash":"sha256:b1979a4c8c30e86ef412c32f0fb8367b843f3665cb323ccd711373d0f4d05262","authorName":"Lê Đức Minh","authorUrl":"https://zyvop.com/author/l445","authorSameAs":["https://minlee0210.github.io","https://github.com/MinLee0210"],"category":"Web Development","tags":["AI","developer tools"],"audience":"Software engineers and developers building applications with Web Development","tone":"Professional, ai engineer perspective","readingTimeMinutes":7,"wordCount":1850,"faqs":[{"question":"Does an extra side window always improve ventilation?","answer":"Not in the terrace example. With wind blowing along the row, the middle house (front and back openings only) reached 98% and 94% fresh air, while the end houses with a side window reached 78% to 85%."}],"primaryTopic":"Web Development","publishedAt":"2026-11-07T13:41:07.293Z","updatedAt":"2026-11-07T13:41:07.293Z","canonicalUrl":"https://zyvop.com/my-weekend-toy-got-a-second-weekend-google-still-can-t-find-it-sbjwq"}