The new release has the addition of Canvas2D, which is described as:
"basically an entire 2D accelerated engine available out of the box."
While Babylon.js is of course a full featured 3D engine, developers often need to handle 2D content, hence the development of Canvas2D. This is 100% WebGL based, and can be used for drawing, animating and interaction with 2D content.
Rendering is another area that has been improved. You can use a refraction channel for standard material and physically based rendering material, and this has now been improved so that you can use a refraction texture to automatically generate a live cubemap of the environment and use it as a source for the refraction channel. Once you've done this, materials will then apply optic deformation to simulate real-time refraction. There's a demo of refraction textures in action on the babylong.js website.
Another addition that is described as 'mind blowing' in the blog post on the Eternal Coding blog about the new version is support for HDR (High Dynamic Range) maps for PBR (Physically Based Rendering). PBR rendering techniques simulate real life lighting, and HDR uses a larger dynamic range for the lighting calculations so scenes look more realistic.
Also on the lighting front is the ability to use multiple dynamic lights simultaneously - Babylon.js v2.3 was limited to four dynamic lights in use at the same time, but the developers say that as more people have machines with powerful GPUs, they have decided to remove this constraint. You can now have an unlimited number of dynamic lights for standard material and PBR material, within the limits of the GPU.
Other improvements include more flexible physic engines plugins, and an improved Unity3D exporter. This has had several missing pieces added, so that skeletons are now exported and animations are supported, and PBR material is generated when the exporter finds compatible shaders. Support for parallax mapping and parallax occlusion mapping has been added, meaning you can simulate volume with textures only.
Rumors that Google was acquiring the data science community Kaggle were confirmed at the Google Cloud Next Conference yesterday. This confers the benefit of the ability to store and query large datase [ ... ]