From Cinevva to Cocos: the Mini-Game Handoff (2026)
Cinevva and Cocos aren't rivals. They're two stages of the same pipeline. Cinevva is where you find out which idea is worth building. Cocos is where you build the finished, optimized version of the one that earned it. This guide is about the handoff between them: what actually carries across, what's a rebuild, and how to run the workflow without lying to yourself about either.
Let's be upfront about the one thing people hope for and won't get: there is no one-click "export to Cocos project" button. Cinevva doesn't emit a Cocos scene. The handoff is a deliberate rebuild of a design you've already proven, reusing the assets you generated. That sounds like a downside until you see what it saves you. For the head-to-head on when to use each tool, see Cinevva vs Cocos.
Why hand off at all
The expensive parts of a Chinese mini game (the software copyright, the ICP filing, the platform review, and for anything monetized, a publisher relationship) are worth spending on a game you know is fun. The problem is you don't know that at the start. Most game ideas aren't fun, and you find out by playing them, not by planning them.
So the pipeline exists to move the "is this fun?" question as early and as cheap as possible. Prototype in Cinevva in minutes, put it in front of real players, and read the retention. Kill the ones that don't land. Only the survivor gets the Cocos treatment and the regulatory spend. You're not doing double work; you're avoiding building the wrong game in an expensive tool.
What carries over (and what doesn't)
Be precise about this, because it's where teams over- or under-estimate the handoff.
The validated design carries over, and it's the most valuable thing. You're not walking into Cocos with a hunch. You have a game people actually played, retention numbers, and a concrete spec: the core loop, the difficulty curve, the session length, what made testers re-open it. That's the hard part of game design, already de-risked.
The generated assets carry over. The 3D models, audio, and art you generated in Cinevva are standard web formats, not locked in a proprietary blob. Models export as glTF/GLB (the rigger and animation tools download a GLB with the animation baked in), and audio is standard files. Those import into Cocos like any other asset. You regenerate or retouch as needed, but you're not commissioning art from scratch.
The code does not carry over. Cinevva builds a web-native runtime; Cocos is its own engine with its own component model and TypeScript API. The game logic is a rebuild. This is the real cost of the handoff, and it's smaller than it sounds because you're re-implementing a design you fully understand, against a spec that already works, instead of discovering the design as you code.
The workflow, step by step
Prototype in Cinevva with the mini game mode on. Describe the game in a sentence and iterate until it's playable. The WeChat mini game mode keeps you inside the portable subset (canvas UI, touch-first, small assets), so the design you validate is already shaped like a mini game, not a desktop game you'll have to bend later.
Test with real players and read the retention. Share the browser link, ideally in the audience you're targeting. The metric that matters is whether people come back, not whether they say they like it. Make several prototypes and let the data pick the survivor.
Write the spec from the prototype. Before you touch Cocos, capture what worked: the loop, the numbers, the feel. The prototype is your design document; don't lose that knowledge in translation.
Collect the assets. Export the 3D models and pull the audio and art you're keeping. Get them into your asset pipeline in the formats Cocos expects.
Rebuild in Cocos for the 4 MB bar. Now the engine's strengths do the work: the engine runtime is baked into the WeChat client so it doesn't count against your package, and AssetBundle subpackaging is how you fit the rest. See the WeChat mini game engines guide for the packaging detail.
Handle the paperwork in parallel. Software copyright and ICP filing take weeks; start them while you build so they're not the thing holding up your launch.
When you should not hand off
Sometimes the right move is to skip Cocos entirely. If your prototype is small, light, and already runs well, and you're shipping free-with-ads rather than chasing a heavily optimized IAP title, the rebuild may not be worth it. A simple game that already fits the constraints can go out through Cinevva's own export path once it lands, without a second engine in the loop. The handoff is for the winner that needs the last mile of optimization or the deep production polish Cocos is built for, not for every prototype.
The regulatory reality (same either way)
The handoff doesn't change the rules. A free, ad-supported mini game needs a per-title software copyright (软著) or electronic copyright certification plus an ICP filing, all doable by an individual with a mainland ID. Any in-app purchase needs a banhao, which only a Chinese company can hold, so a foreign studio needs a licensed local publisher. Whether the game was born in Cinevva, Cocos, or both, the paperwork is identical. Details are in the WeChat and Douyin publishing guide.
Common Questions
Can Cinevva export a Cocos project?
No. There's no one-click export to a Cocos scene. Cinevva validates the design and produces reusable assets in standard formats; the game itself is rebuilt in Cocos. The value is that you rebuild a game you've already proven is fun, not that you skip the build.
What actually transfers from Cinevva to Cocos?
Three things: the validated design (loop, difficulty, session length, retention data), the generated assets (3D models as glTF/GLB, audio, art in standard formats), and your understanding of why the game works. The code doesn't transfer; it's re-implemented against Cocos's API.
Isn't rebuilding the game wasteful?
It's cheaper than the alternative, which is building the wrong game in an expensive tool. Most prototypes fail. Rebuilding only the proven winner, against a spec that already works, is far less waste than fully producing every idea to find out which one was worth it.
Do I have to use Cocos after Cinevva?
No. If a prototype is simple and already fits the platform constraints, you can ship it through Cinevva's own path without Cocos. The handoff is for games that need heavy optimization or deep production polish.
Can I reuse the 3D models and audio I generated?
Yes. Generated models export as glTF/GLB and audio as standard files, both of which import into Cocos. You may retouch or regenerate for the 4 MB budget, but you're not starting the art from zero.
Validate the idea in Cinevva first. Hand the winner to Cocos only once it's earned it.
Related
- Cinevva vs Cocos for Mini Games — the prototyping-vs-production split in detail
- Cocos Alternatives for Mini Games — the full engine field
- Game Engines for WeChat Mini Games — the 4 MB limit and packaging
- How to Publish on WeChat & Douyin Mini Games — licenses, review, revenue share
- WeChat mini game mode — building inside the portable subset
- Cinevva Engine — our open-source, web-first engine