orderIndependentTranslucency issue?

Hi,

I'm displaying thousands of colored points (Entities) and it's working
fine except I think I have an orderIndependentTranslucency problem,
see attached.

Zoomed out, the points overlap a lot, and it seems that drawing order
(back to front or front to back) controls whether they display as a
pleasing solid gradient of color, or become almost invisible. It seems
as if they're being antialiased with the background, and the fringe of
background colored antialiasing is overwriting them as they're drawn.

Is this where this orderIndependentTranslucency switch comes in to
play? I'm using 1.24 from cesiumjs.org and creating the viewer like
this:

  viewer = new Cesium.Viewer('cesiumContainer', {
    contextOptions: {
      antialias: false
    },
    orderIndependentTranslucency: true
  });

but that seems to make no difference vs.:

  viewer = new Cesium.Viewer('cesiumContainer')

It's not a very high performance graphics card, so maybe OIT just
doesn't work on this setup?

Thanks, Terry

Capture.PNG

Hello Terry,

It’s hard for me to tell exactly what is going on from the picture you posted.

We do have an issue filed for OIT problems with polygon: https://github.com/AnalyticalGraphicsInc/cesium/issues/4170

Maybe that’s related to what you’re seeing?

What does it look like if you set orderIndependentTranslucency: false?

Best,

Hannah

Terry - I think your issue is similar to what I dealt with concerning overlapping points:

https://groups.google.com/forum/#!msg/cesium-dev/5qv7Q_qudt8/7VlF5JBQAgAJ;context-place=forum/cesium-dev

I ended up having to use the hacked solution suggested in this github issue thread:

https://github.com/AnalyticalGraphicsInc/cesium/pull/3784

You have to edit the PointPrimitive shader file and add borders to all of the points to implement this fix, but I have found it to be successful. Apparently there is an issue with IE/Edge according to other associated issue posts, so you might have to weigh that.

[snip]

Terry - I think your issue is similar to what I dealt with concerning
overlapping points:

Redirecting to Google Groups

I ended up having to use the hacked solution suggested in this github
issue thread:

Messing with Z by emackey · Pull Request #3784 · CesiumGS/cesium · GitHub

[snip]

Thanks, that definitely looks like my issue, will follow up the threads
next week.

Cheers -Terry