Is this rendering architecture blog post from 2015 still accurate?


I am reading through an old blog post on cesium’s site: Graphics Tech in Cesium - Renderer Architecture – Cesium
I wanted to know how accurate this article is in regards to the modern cesium library. If Cesium has diverged, in what ways? Thank you for any help you can offer.

Hi @shane_benlolo, one of the most significant changes since that old post is the re-design of how we load and render glTF models. This was introduced in the 1.97 release.

For a high-level explanation of the changes, see Peter Gagliardi’s blog post about the new architecture, especially the sections on rendering a model and the model rendering pipeline.

1 Like


Thank you for the response. I have a follow up question if you don’t mind:

I tried cloning cesium and implementing this tutorial: Entity API Performance – Cesium

It’s a pretty old tutorial, and after following the directions and profiling the performance difference, my performance was actually a bit worse, while the blog shows the time getting cut down to about 1/10th the original.

Am I implementing incorrectly? Or is this tutorial no longer relevant? Here is the changes:

Hi Shane,

That is a fairly old tutorial, but the contents should still be relevant.

We would like to document and address any significant performance hits. I’m curious if the performance hit is due to one type of entity in particular-- For instance just labels or billboards.

Would you be able to give each type of entity a try and see if the performance effect is isolated, or generalized to all entities?