html5 wrapper,(webGL (three.js/babylon.js) + wrapper) vs Unity

Hello there,

I am working on a browser game for a while and I intend game to include a tilemap in addition to spreadsheet part. At first I was intending to have a simple heightmap with canvas and isogenic engine , something like

but after then thinking that this looks very 90s, decided to switch a 3D terrain solution to avoid "few millions slope sprite" problem. So ended up at a webGL based heightmap solution something like

  in fashion of

as inline element and/or fullscreen

As webGL is apparently low level, I considered three.js over briefly checked babylon.js which seems lacking canvas fallback or software rendering if I am not mistaken.

So my current plan is using three.js with canvas or maybe even DOM fallback (by server side rendering) for viewing issues (map etc) and using webGL 3D only for situations involving "action".

But I have concerns about availability of webGL and even more about performance especially on mobile platforms. I read that webGL performance will no longer be an urgent issue with IOS 8 and Android Lollipop for mobile platforms (hopefully with Windows 10 mobile as well) , which is near future (in terms of market share)

But even decentish desktops/laptops I tested struggle at webGL demos, easily dropping to 5-10 fps, I am not being that much optimistic about mobile promises. And even worse there may be webGL wide availability but it is even worse if can't deliver acceptable performance.

Due to this concerns and even before that due to MBs of asset size problem, I had considered having wrappers for mobile platforms basically consisted of an asset manager, local storage and webview. But I am not sure about increase in performance when it comes to webGL and if I can easily tailor wrappers to my needs. I don't want to miss player base having low end devices in a browser run game.

So when I look at webGL with/without wrappers,

pros are moderate learning curve, theoretically highly cross platform (in case an universal wrapper used or no wrapper)

cons are performance and experimental nature (browser support is varietal)

I also considered Unity in case webGL is not able to satisfy needs in near future, but when I look at pros and cons of Unity,

pros are single(ish) codebase for multiplatform, new webGL exporter, stable release and native performance

cons are steeper learning curve, frequent involvement of 3D elements (steep learning curve again and/or increasing asset cost), not as web friendly as webGL and radical change in game design

What I don't like most at Unity is radical change in gameplay as doubt it is common practice to include other HTML elements to an Unity game, I don't want to make gameplay "casual" (I mean in "any creature with a few neurons that can tap on a screen can play the game" style)

So my question is,

Would you recommend webGL or Unity? And in case of using webGL what would be ideal setup in your opinion from singleish codebase and performance perspective?

Thanks in advance and my apologies due to writing style.