A Shopify launch page,rebuilt as a Vietnamese stage.

A Shopify studio in Vietnam has a credibility problem that is not about skill: there is very little public, technically demanding Shopify work in Vietnamese to point at. Editions pages are the most ambitious thing Shopify itself publishes each season — WebGL scenes, 3D product renders, Rive animation, hundreds of feature cards — so rebuilding one is a direct way to show we can handle that tier of front-end work.

The second reason is the one we actually care about. Shopify writes Editions from the vendor's side: here is what you now have. Merchants need the other half: which of these can be switched on in an afternoon, which ones change the money flow and need before/after numbers, and which ones are a real project with tests and a rollback plan. Vietfolk ’26 is that other half.

Every one of the 187 feature entries was rewritten rather than translated. Where Shopify says "Sidekick can build custom apps for your business", our version says what we would build first — a replenishment app, a team task board, a B2B CSV importer — how long the setup takes, and what we would need from the merchant to do it.

The result reads as a services page disguised as a launch page, which is the point. Each section ends with the same two facts: what we can set up, and roughly how long it takes. Those are the two questions we get asked in every discovery call.

The page is twelve scroll-driven scenes. The 3D work ships as 47 compressed glTF binaries with 30 KTX2 GPU-compressed textures, so the browser decodes textures on the graphics card instead of the main thread. Twenty-seven Rive files handle the vector animation, and 99 short video clips cover the product demos that would be too expensive to render live.

Scene timing is driven by a Theatre.js project state rather than hand-written scroll math, which is what made it possible to iterate on twelve scenes without breaking the ones already finished.

On a mobile emulation with an unthrottled connection the live demo renders its largest element in 1.2 s with zero layout shift. That number is honest but flattering — see the next chapter.

The page transfers 8.1 MB on first load, and the HTML document alone is 1.36 MB. That is acceptable for a deliberate showcase; it would be indefensible on a storefront. The asset directory is 486 files and 208 MB on disk, and the mirror script that assembled it also saved every Shopify CDN variant — the same image under a dozen query-string filenames — which pushed the raw export past 800 MB before we deduplicated it with a rewrite rule.

That weight broke our deployment. Hostinger's upload API wipes the site directory and then rejects large archives, so a page this size cannot be deployed atomically through it. We moved the whole site to an rsync merge over SSH, which transfers only what changed and resumes after a dropped connection. A demo page changed how the production site ships — that is the kind of consequence a client should hear about before the work starts, not after.

Vietfolk ’26 was originally published as an indexable page with a sitemap priority just below our service pages. That was a mistake we caught in our own SEO audit: the page follows Shopify's Editions structure closely enough that a search engine would reasonably file it as derivative, and an 8 MB page is a poor first impression for a visitor arriving from search.

So this case study is now the indexable face of the work, and the demo itself carries a noindex tag and is linked from here. Same pattern we recommend to clients with heavy interactive pieces: let a fast, well-structured page rank, and let the showpiece be one click behind it.