Support marker.opacity arrays in scatter3d

[alexcjohnson]

So we can get rid of multiOpacityTransform
scatter3d traces already support a single-valued marker.opacity, and they support opacity as part of a marker.color array (which is what this transform makes use of) so it should be pretty easy to add this capability.

[etpinard]

so it should be pretty easy to add this capability.

Not so fast. Mikola wasn't able to get this to work. The "opacity" in that multiOpacityTransform hack isn't "true" opacity. I don't remember the exact details, there should some info in old streambed issues.

[etpinard]

Related: #1267

[alexcjohnson]

I had an idea about 3D opacity... curious if something like this has been tried before:

We oversample 3D scenes, right? This idea depends on oversampling, and may work better if we increase it. What if instead of trying to model the opacity independently within each pixel, we fill pixels pseudorandomly - ie at 50% opacity we just fill 50% of the pixels at full opacity. Then a point at a different Z value and still 50% opacity fills some different 50% of the pixels, and the depth buffer sorts out ordering just like it does with fully opaque items normally.

The result would be a bit speckled... but averaged over some area (hence the oversampling) it seems like it would look just like real partial opacity. The pseudorandom generator should be a pure function of screen position (x,y,z), but we may want to do something to give it greater uniformity in (x,y) than a real random distribution, to reduce speckle.

@archmoj do you think this sounds plausible?

[archmoj]

Seems like a good idea @alexcjohnson
Let me think about it.

[gvwilson]

Hi - this issue has been sitting for a while, so as part of our effort to tidy up our public repositories I'm going to close it. If it's still a concern, we'd be grateful if you could open a new issue (with a short reproducible example if appropriate) so that we can add it to our stack. Cheers - @gvwilson