CK2 Dev Diary #18: Optimization and modding

[Moah]

I thought this would be one of the additions in terms of performances... Ironman could really use this.

Ironman is still compressed, this is unavoidable for various reasons. However we changed the type of compression used and fixed several issues we had with compression. The difference in speed with uncompressed is negligible now, even with late game saves.

 

[Moah]

Here's an example from the deaths script:

death_dungeon = {
    long_desc = "DEATH_DUNGEON"
    can_nokiller = yes
    sound_list = {
        female = {
            "death_prison_female_01"
            "death_prison_female_02"
            "death_prison_female_03"
        }
        male = {
            "death_prison_male_01"
            "death_prison_male_02"
            "death_prison_male_03"
        }
    }
}

long_desc is the generic description for that death cause.
can_nokiller means a death caused by someone else can have an unknown perpetrator.
sound_list is the list of sounds associated to that death cause. One will be picked at random based on the gender of the victim.

We also have the following flags:

execution = yes
violent = yes
murder_unknown  = yes

I think the two first ones are self-explanatory and "murder_unknown" means it's a death cause specifically for when the murderer is unknown.

Edit: And here are the localisation associated to that death reason:

DEATH_DUNGEON;died in the dungeons of $KILLER$;;;;;;;;;;;;;x
DEATH_DUNGEON_NO_KILLER;died in prison;;;;;;;;;;;x

 

[Moah]

@Moah: Thank you. Can I ask a few followup questions?

This is probably obvious, but sounds are optional, correct?

Is the "_NO_KILLER" key is used when the killer is unknown or when the death reason doesn't have a killer (like dying in an accident), or both?

Regarding the violent flag, I understand it means that the person died a violent death, but I was wondering how it would look different from a non-violent death in-game. For example if you have say death_poison without violent = yes, then would the death skull look different from a death_reason with violent = yes?

The violent flag triggers the red skull in the portrait.
The _NO_KILLER is used when you have a death cause that normally has a killer, like the dungeon or a murder but can end up with an unknown killer.
Here's another exampe:

death_duel = {
    violent = yes
    long_desc = "DEATH_DUEL"
    can_nokiller = yes
}
DEATH_DUEL;was slain by $KILLER$ in personal combat;;;;;;;;;;;;;x
DEATH_DUEL_NO_KILLER;was slain in personal combat;;;;;;;;;;;;x

compared to:
death_hashshashin = {
    violent = yes
    long_desc = "DEATH_MURDER_HASHSHASHIN"
}
DEATH_MURDER_HASHSHASHIN;died at the hands of the Hashshashin;

In the case of death_hashshashin you don't have the can_nokiller flag since the killer's name doesn't appear in the description, and thus it doesn't matter if for some reason we can't find it.

Edit: As for the sounds, if the death cause doesn't have any sound listed, it will use the default death sounds (the one we had in conclave).

 

[Moah]

The violent flag triggers the red skull in the portrait.

I must correct myself here: the violent flag is not used for the red skull, it is used when the death cause doesn't have a sound list, to chose a violent death sound instead of a normal one.

 

[Meneth]

Nobles are just too reluctant to marry lowborn, even when there's nobody else available.

Unmarried rulers now get more desperate as they get older, lowering their requirements. That should hopefully help counter this.

 

[Meneth]

Is there any way we can tone down this sort of thing, if we don't care about performance that much? I've never had a problem with it before all these massive performance changes, and have always felt the game has too few, not too many random courtiers. This seems like a massive thing to be forced on us, even more "Unimportant" characters disappearing.

It's always been there, but will now be configurable. You can set the court size threshold at which it happens in defines.lua.

 

[Meneth]

That's brilliant, thanks. The impression I got from this thread was the plan was to make it more aggressive? The fact we can edit it is great though!

Yes, the threshold is 20 in 2.5.2, while it is 10 in 2.6.

 

[Meneth]

Bummer. That means a player can't get achievements

It should've been configurable in the game configuration menu that was teased in an earlier DD. But I guess it's important from an engine initialization perspective, which is why it's kept in defines.lua.

Do note that player courts are unaffected, so your own courtiers won't be randomly killed off.

 

[Darkrenown]

So, do you think we faked the video of the two versions running side by side, or...?

 

[Moah]

Could they set that up as a game option though?

Game options are fully modable, so it should be possible, in the same was that the "explode map" from CK2 could be made into a game rule rather than a decision.

 

[Darkrenown]

Unless you've got a tiny screen I'm not sure why you can't see the clock in both games in the video there, and they are both running on the same machine so the specs aren't really relevant.

 

[Meneth]

It could just be my eyesight. I'm on a six inch screen as well, mind you.

Specifications are relevant in comparison to my own computer, although I imagine other consumers may be interested in seeing how the video will stack up against them.

Don't know what the specs were. @rageair should know.

As to the rest, from the description:
"Both campaigns started on January 1st, 769, and ran for 2 minutes, 15 seconds.

The 2.5.2 game got to April 772; just short of 1200 days into the game.
The 2.6 game got to late January 775; just over 2200 days into the game.
All told, in this sample 2.6 was roughly 80% faster than 2.5.2.
The yearly autosaves likely contributed significantly to this; with no or rarer autosaves the difference we've observed has usually been about 40%."

 

[Moah]

Don't know what the specs were. @rageair should know.

As to the rest, from the description:
"Both campaigns started on January 1st, 769, and ran for 2 minutes, 15 seconds.

The 2.5.2 game got to April 772; just short of 1200 days into the game.
The 2.6 game got to late January 775; just over 2200 days into the game.
All told, in this sample 2.6 was roughly 80% faster than 2.5.2.
The yearly autosaves likely contributed significantly to this; with no or rarer autosaves the difference we've observed has usually been about 40%."

You can also check the youtuber videos made on the preview build:
https://forum.paradoxplaza.com/forum/index.php?threads/reapers-due-preview-videos.962646/

 

[Moah]

Thank you @Meneth and @Moah, that's really helpful. If @rageair wouldn't mind sharing the specifications as well then I'll definitely be picking the expansion up.

The optimizations are in the patch anyway, so you can always check the game at release before deciding one way or another.

 

[Meneth]

Meneth - is it possible you can change (or ask someone) to alter the viewing angle just a tiny bit? SO when we zoom all the way in, the viewing angle will be flatter. This would allow us to see more of the map when zoomed fully in and we would also be able to better see all the 3D units we pay extra for.

I know it can happen since there are plenty of Paradox videos that swoop and zoom around the map. Or maybe just make it an option we can check in case it breaks the game a bit (clipping etc). I realize this is not what you're there for but this little change/addition would be amazing.

I'm not on the project any more, so nothing I can do about it sorry.

I think if you connect a controller, that'll let you pan. Not 100% sure about that.