[workspace] Upgrade meshcat to latest commit

[mwoehlke-kitware]

This change is 

[mwoehlke-kitware]

Someone else should perform the full test, as I skipped the XR/AR and gamepad portions. Here are my results/observations:

The blue box should spin clockwise about the +z axis.

The blue box, as viewed from above (which is closest to the camera position set by the script), spins counterclockwise.

The lights should have dimmed.

If anything, the light seems brighter. (I wasn't expecting a change, however, and have no way of directly comparing. Also, at this point I didn't realize that things don't always update until the camera is moved and I don't think I moved the camera before advancing to the next stage.)

An environment map has been loaded from a png.

It did, but was not visible until I rotated the camera.

The Cornell box has been replaced by a room with brick walls loaded from a jpg.

Again, the background was white until I moved the camera.

Reloading the page should always succeed.

Reloading caused Firefox to crash. 😒

Things seemed to be loading back to the previous state just before the crash, however. Also, after restarting, everything appears to be as it was before trying to reload. (I guess restarting the browser is a better 'did it reload' test than just a forced refresh, anyway...)

Click the ButtonTest button a few times.

This does not seem to do anything. (However, later the test reports that the buttons were clicked several times.)

Next, we'll test gamepad (i.e., joystick) features.

Besides that I don't have a handy gamepad, "Gamepads are not supported outside of a secure context".

Now we'll test the WebXR functionality.

Nope, not installing a German mystery extension.

[jwnimmer-tri]

Here are my results/observations: ...

It would be worth disambiguating which observations are new to this branch compared to master, versus which ones also happen on master. Things that are novel here deserve a close look, but things that also exist on master are probably latent problems that we haven't noticed yet, which could be filed and deal with aside from this PR.

[mwoehlke-kitware]

For comparison, my results on master...

• The blue box rotates in the same direction (which I would describe as "counterclockwise").
• I suspect I'm seeing the same behavior with the "dimmer" lights. (See analysis, below.)
• The Cornell Box background is not visible (I think it is not visible in the shiny box's reflections, either, although it's difficult to tell without moving the camera) until a repaint is forced (e.g. by moving the camera or resizing the browser). To be sure I wasn't just seeing lag, I waited over two full minutes.
• Same with the bricks.
• Considering my previous results, I did not test reloading.
• Knowing what to expect, I can verify that the number of times the buttons are pressed is correctly recorded. As before, there was no immediate feedback.
• I also did not retest results which were not previously anomalous, or which I didn't test previously.

With respect to the lights, this time I have prepared "before" and "after" screenshots. Here is the previous state:

Here is the state after the lights became "dimmer":

Careful pixel-peeping (and use of a color picker) shows that the lighting is, in fact, dimmer, however the effect is extremely subtle and the optical illusion of the lighter background makes it very hard to tell using only the Eyeball Mk-I.

Therefore, for the tests I did perform, it appears that the upgrade did not change anything.

[jwnimmer-tri]

The sha bump here, even with manual testing involved, is still fine for just a single reviewer.

+@EricCousineau-TRI for both reviews per schedule, please.

[mwoehlke-kitware]

@EricCousineau-TRI, reminder, please run the manual meshcat tests before approving. (Especially the gamepad/XR/AR parts.)

[SeanCurtis-TRI]

-@EricCousineau-TRI +@SeanCurtis-TRI

[SeanCurtis-TRI]

On the shopping list:

• blue box rotation: agreed, definitely counter clockwise. It's clear from the code that it is getting positive rotation around the Wz axis. So, perhaps it would be most precisely articulated thus.
• dimming light: yep. While it unambiguously reduces the intensity of the ambient light, that is almost offset by the environment going white. When we added environment maps, that's the end result we got. I may have to revisit that logic in meshcat.
• For the numerous occasions where reported visual changes weren't observed until something forced a redraw, while I very rarely experience that on chrome on my machine, I wonder if it's a more common problem in firefox. But we should definitely have a note saying, "Depending on your browser, the background may not redraw until you bump the camera." (or some such thing)

Otherwise: X2

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: complete! all discussions resolved, LGTM from assignee SeanCurtis-TRI(platform)

[mwoehlke-kitware]

While it unambiguously reduces the intensity of the ambient light...

Yes, that was my impression also; the ambient light is (subtly) reduced, but the illumination where directional light is present is nearly unchanged. (Especially as those parts are mostly over-exposed.)

[SeanCurtis-TRI]

RE: the environment. If the scene had only consisted of objects with a "certain kind of material", they would've become dimmer even as the background got lighter. But the green box presented front and center is most affected by the white background; it gets illuminated, spoiling the effect.

I've submitted an issue for meshcat and would love any and all thoughts:

meshcat-dev/meshcat#175