Nutrition values depending on hunger of character are not continuous

[BevapDin]

The effective nutrition of food is altered based on the hunger of the character (see player::nutrition_for). However, the formula used there produces non-continuous values as shown in this table:

hunger	nutrition
0	10 (value from JSON)
99	10
100	6.7
150	10
300	20
301	8.6
2800	60
2801	46.7

That means if the hunger increases by one point from 99 to 100 or from 300 to 301, the nutrition that will be extracted from the food is actually decreased. This seems very wrong. In the range 100 to 149, the character gets less nutrition as when they would be not hunger at all.

For reference, this is the current code (slightly reformatted, hunger >40 is hungry, >100 is very hungry, >300 famished):

if( hunger < 100 ) {
    return nutrition;
} else if( hunger <= 300 ) {
    return hunger / 300.0 * 2 * nutrition;
} else if( hunger <= 1400 ) {
    return hunger / 1400.0 * 4 * nutrition;
} else if( hunger <= 2800 ) {
    return hunger / 2800.0 * 6 * nutrition;
} else {
    return hunger / 6000.0 * 10 * nutrition;
}

Btw. the item info shows the hunger-dependent values, not the plain value, is that intended?

Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

[kevingranade]

So many things wrong here...
One, it shouldn't be adjusted by the food at all, if there's an
adjustment it should happen when the nutrition is applied to the player, in
the player (or character) class.
Two, it is definitely intended to be a steady increase, not a discontinuous
one.
Three, I'm sceptical of the whole concept of nutrition increasing based on
how hungry you are. (Acknowledging that hunger isn't what we normally call
hunger, it's more like starvation level at the point when this stuff is
kicking in).
Four, (a side effect of 'one'), the post-adjustment values definitely
should not be used.

Personally I'm for nixing the whole thing.

[Rivet-the-Zombie]

Three, I'm sceptical of the whole concept of nutrition increasing based on how hungry you are.

So am I; the whole thing is silly.

[chaosvolt]

Yeah, it seems...hmm. Is it meant to represent some psychological factor? Seems like the only psychological factor that would affect a food stat would be morale, for example desperation leading to eating food normally regarded as unpleasant.

[kevingranade]

If anything, you should perhaps suffer from lowered metabolism when starving, which would mean you don't get hungry as quickly, but also are impaired in other ways, like slower stamina regeneration and healing. That's just a guess though, I don't know if that's how it really works.

[Coolthulhu]

If anything, you should perhaps suffer from lowered metabolism when starving, which would mean you don't get hungry as quickly, but also are impaired in other ways

I like this idea.
Though our current hunger system (inc hunger once per x turns) is a bit hard to adjust. The simplest way would be to make it all floats, but there may be a more elegant one.

[Midaychi]

Unreal world has hunger separated out kind of neat.
They have a 'fullness' meter that operates a lot like cata's current nutrition system does. You drink or eat something and you get fuller, and sometimes get more full if you go a long while without eating or eat in small amounts for along time.
The second meter is their overall healthiness of their diet, which tracks how many calories and nutrients on average you're clocking a day compared to how many calories and nutrients you're using just survivng the environment and doing work.
If you eat poorly in that game, you can easily get into a situation where you're wasting away but full, and vice versa.

[kevingranade]

Yea we're planning on doing something similar to that, but it's deeply off-topic here :P

[tivec]

Take a look at #11514.

Back then, we found it very strange that one would need to eat 10+ cooked meat just to raise from famished to very hungry. The system there was conceived during that process, to make it a bit less unrealistic.

[chaosvolt]

While I see the gameplay reasoning for that, the other issue (besides the realism concern) is that the net effect of that scaling isn't continuous, it's less of a ramp-up and more of a roller coaster.

[caland]

I think the concept of the code was if hunger is less then 100 return base nutrition value.
and if hunger is between 100 - 300 it should scale to return 1 time to 2 times the nutrition value.
and 300 - 1400 scale from 2 times to 4 times the value. etc.

if that is the case the code should be:

if( hunger < 100 ) {
return nutrition;
} else if( hunger <= 300 ) {
return ((hunger-100)/200 + 1) * nutrition;
} else if( hunger <= 1400 ) {
return ((hunger-300) / 1100 * 2 + 2) * nutrition;
} else if( hunger <= 2800 ) {
return ((hunger-1400) / 1400 * 2 + 4) * nutrition;
} else {
return ((hunger-2800) / 3200 * 4 + 6) * nutrition;
}

which give a consistence increase.

[Zireael07]

@caland: That is a good idea.