Track this frame mode causes solar system objects to render badly

[patudom]

Context - this is for an Eclipse-related Cosmic Data Story.

In Solar System view, we want to show how the Moon moves between the Sun and Earth to cause an annular eclipse. This is tricky to do in solar system mode because if you have the camera centered on Earth, Earth itself will block the view of the Moon and Sun. If you have the camera centered on the Moon, you can make a static view of the Moon/Sun work, but when you run time forward, it's hard to keep the viewing angle directly in front of Earth (and due to parallax issues, it ends up looking like the Moon is not moving in front of the Sun)

In the desktop client, we can set this up by tracking the frame of a particular city in the eclipse path as described here: cosmicds/noncode-issues#17. However, if we try to follow the same steps in the web client, the tracked frame view goes haywire.

For me, in Chrome on a Mac, the Moon is not visible in front of the Sun with Planet Size = Actual, which you need for the eclipse proportions to be correct. If you increase the Planet Size, it looks like the Moon is being rendered BEHIND the Sun, and the imagery has randomly shaped missing chunks that kind of dance all around the object.

For @johnarban, the Sun didn't appear at all in this view. (That was also the case for me on Safari).

@pkgw, do you think this is a problem that is fixable within the next couple of weeks, or do you have suggestions for other options for creating the type of view described?

Thanks!

[pkgw]

Yikes! This seems like a pretty serious rendering engine issue that should be hunted down, but I suspect that it would be tricky to diagnose and fix. Unfortunately "next couple of weeks" does not feel likely to me ☹️

The only alternative I can think of right now, which may or may not be realistic either, is some custom code to manually steer the WWT camera to achieve the effect you want. If you're basically creating a single animation, that might be feasible. If you want a more interactive experience there might be a clever way to implement things to make it work, but my intuition says that it would be a lot harder to get something going.

[patudom]

@pkgw, thanks for looking into this. I thought the short timescale might be optimistic, so we will fall back on a Sky view-only MiniDS for the October eclipse (for which we will need #266) and try to get the larger 3D solar system Data Story out for the April eclipse. Does that seem more manageable?

[pkgw]

Hopefully? I'd recommend budgeting around a week of @Carifio24's time to dig into things and try to understand what's going on, and then hopefully a similar amount of time to devise a fix, but it depends on what exactly is going on under the hood.

[patudom]

Sounds good, thanks @pkgw. @Carifio24, let's plan to have you tackle this after we get through the Sep/Oct CosmicDS crunch.