Allow to set maximum star opacity at daytime

[Wuzzy2]

Fixes #9887. Successor of #10541.

This PR adds a new Lua API feature to player:set_stars: day_opacity. This is a floating point number between 0.0 and 1.0. It sets the maximum opacity of stars at daytime, with 1.0 meaning fullbright stars (same as night) and 0.0 being fully invisible.

Technically, the mechanism to do this is very simple: A value of X just means the star opacity is capped at X at daytime, meaning it cannot get more opaque during the day.

The default value is 0.0 and this feature is backwards-compatible. I've tested a 5.5.0-dev client on a 5.4.0 server and it worked.

Note: The word "opacity" here means the opacity of the stars BEFORE star_color is applied. This means "full opacity" here means that the full effect color of star_color is applied, it does not neccessarily mean that the stars are actually fully opaque. Stars never get more opaque than the alpha value of the star_color, like before. This is NOT new behavior, I just pointed that out for clarity.

To do

Ready for review.

How to test

Function test in 5.5.0-dev:

• Activate the luacmd mod
• Start DevTest
• /lua me:set_stars({star_color="#FF0000FF"}) (this is optional, but red stars are better for testing; also note the default star_color is not fully opaque, so don't be surprised if day_opacity=1 doesn't make the stars fully opaque when using the default star_color)
• /time 12000
• Check sky (expected: no stars)
• /lua me:set_stars({day_opacity=1})
• Check sky (expected: visible stars)
• /lua me:set_stars({day_opacity=0})
• Check sky again (expected: no stars)
• /time 0
• Check sky again (expected: visible stars)
• /lua me:set_stars({day_opacity=1})
• Check sky again (expected: visible stars)
• /time 12000
• /lua me:set_stars({day_opacity=0.75})
• Check sky again (expected: reduced opacity stars)

Feel free to try out other opacity values.
Also try to let a full day/night circle complete (under various conditions) to check whether the transition is still smooth in all cases. You could do /set time_speed 300 to speed things up.

Compability test:

• Launch a 5.4.1 server
• Connect with your compiled -dev client to it
• Watch the sky at night and at day; for compability reasons, the stars should behave in the default way (visible at night, not visible at day)

[ghost]

Looks promising from a quick test, but haven't given this a full pass yet.

That said, this needs to be rebased and fixed to work with master. This is missing changes to compile with the latest IrrlichtMt, and if we rebase to get them the code no longer compiles without modification due to #11983. The fix is fairly trivial though, it's a 1-liner.

[Wuzzy2]

The rebase is complete.

[ghost]

Just a quick note--I saw the notification, and will be reviewing this weekend. I'm taking a break from reviews this week since I'm participating in a game jam, but this is at the top of my list and will probably get looked at again on Saturday.

[ghost]

Tested, seems to work as intended.

• Singleplayer - no issues
• New server, old client - changing old client's daytime opacity does nothing, as expected
• Old server, new client - obviously impossible to change daytime opacity, both players see the same star opacity as before

I did see one odd thing--at time speed 720 (2 minute cycle), the stars 'warp' to new positions around sunrise. At time speed 72 though, this did not occur. It could be an existing behavior that wasn't visible before, I'm not sure if we want to investigate it now or not worry about it for the time being.

[ghost]

One other thing I'd like to raise before approving: Should we have night_opacity or some variant thereof? It may sound redundant at a glance, but I suspect we can't presently make daytime stars brighter than nighttime ones even with this change.

I'm not sure how useful this feature is, but it seems like it could be worthwhile to have the option available for the sake of flexibility. Any opinions from others would be appreciated.

[Wuzzy2]

I did see one odd thing--at time speed 720 (2 minute cycle), the stars 'warp' to new positions around sunrise. At time speed 72 though, this did not occur. It could be an existing behavior that wasn't visible before, I'm not sure if we want to investigate it now or not worry about it for the time being.

This does not seem to happen reliably, tho, only sometimes. And it's also the sun and moon that suddenly "jump". So I think this must be an old, unrelated issue. Won't touch.

One other thing I'd like to raise before approving: Should we have night_opacity or some variant thereof? It may sound redundant at a glance, but I suspect we can't presently make daytime stars brighter than nighttime ones even with this change.

Won't do. I don't see a need right now. But if it will be needed, there is always chance for another PR.

[rubenwardy]

tested