[CF-Devel] Weather level 5 (Re: Weather needs smoothing)

crossfire-devel at archives.real-time.com crossfire-devel at archives.real-time.com
Sat Jul 31 01:35:19 CDT 2004


Just my two cents on the topic of weather...

>
     
      Weather code, IMO, need a complete redo. The guy who coded this, i think had too great ambitions for a MORPG. When you look at code you see it simulate seasons, humidity, temperature, wind, gulf stream (no you are not dreaming), altitude, and am not sure but probably earth rotation.
     
     
Well, I basically agree with the idea of a redo. The code seems to have been written without any planification of some sort. This resulted in a somewhat patchy code that became a bit hard to follow and eventually improve.

The question of too high ambitions is different, though, IMHO. The parameters included in the weather simulated model weren't different from what you could find in some computer Wargames of the early 90ies. And if programmers and computers of those barbarian dark ages were able to manage them, I suppose it should still be possible today.
Yet the question of usefulness of some parameters needs to be asked - it may be more accurate to simulate the gulfstream effect, but if it has no significant impact on the game itself, it is a bit useless.

>
     
     But i think there were great things in this code that should be kept!
     
     
>
     
      1) terrain change as climate change (eg snowed mountains)
     
     
Definitely a good idea.

>
     
      2) plants are growing (a good tihng to make the world alive)
     
     
Again, an interesting one, but that needs some refining. The current implementation is basically incomplete (trees aren't included in the system, plants grow but - AFAICT - never die, etc.)

>
     
      3) snow and rain change as you change location (but current calculation is too complicated, all i need as player is too see climate change and perhaps moutians area colder and mor humid than sandy areas). This could be complete random player won't see difference. No need for climate predicitons.
     
     
My objection to this is that a (somewhat) deterministic system allows interesting game manipulations (which are not used ATM): adventurers could for example prepare themselves for an expedition in the Nagora Ice Desert by asking a Wizard to tell them if the weather will be good or bad. Weather forecasting may give you a tactical advantage in a couple of situations, so I'm not sure it should be thrown out of the window too quickly.
Now it is true that computation definitely needs to be improved in a way or another.

>
     
      But the following could be changed/added
     
     
>
     
      1) No gulfstream (simply reducing speed of climat echange at seaside is enough to simulate)
     
     
I'd say "better gulfstream" - which covers the idea of simplifying the concept.

>
     
      2) Better drawing for rain (I have seen player trying to fight those blops) Some small circles would be enough, perhaps adding smoothing to this would be 
     
     good (you just need to set a good smoothlevel (say big case on top of most things) and draw the smooth picture. No need to make it floor as smoothing 
isn't restricted to floor (any visible arch can be smoothed!!!)

My idea is that rain shouldn't appear on the tiles below the player - I think it would be much better if a topmost layer was used to represent rainfall (or snowfall). This way, it wouldn't be confused with items lying on the floor.

Now, one could object that you need to represent the fact that the floor is becoming wet, something my proposal wouldn't display. Well, do we represent the fact that a floor is dusty or covered with blood ? My own opinion is that an graphically non-intrusive representation would be far than enough for what's mostly a secondary parameter. Just adding an adjective to the description of the floor should suffice for most (if not all) purposes.

>
     
      3) No large scale calculation, i think creating randmoly some clouds and then moving the smoothly along the land until they are empty would be good enough to simulate some small climate.
     
     
Agreed, but again with a reserve against a completely random system. You should be able to explicitely define areas producing the clouds, as well as the quantity of those (either directly or through the combination of the various weather parameters).

>
     
      4) need some way to tell client 'hey it's raining' or 'hey it's snowing' so it can do some special effect (sound of water drop, snowflakes, etc...) And we can make the difference between 'it's snowing' and 'snow has dropped here a few minutes ago, it's full on ground'
     
     
See point 2) above.

>
     
      5) Perhaps create a life cycle for plant (a rose is not suppose to survive on full snow ground, grass starts to die put in sun for too long)
     
     
Agreed. I think it was the initial intempt, but for some unknown reason, it never got finished.

>
     
      The last 2 points are not the most important and may need lots of work, but the rest could be done quite easily (easy to say, difficult to do, like 
     
     usual).

4) and 5) are mostly consequences of a weather system rather than part of it, so it definitely gets lower priority IMO.

>
     
      Anyway, i dunno what's your opinion but mine is current weather code is jeopardy to activate on server and personnaly i don't want to touch it 
     
     anymore (too difficult to maintain).

So you already reached the point of making it work ? You went further than me, then :)
Seriously, the current weather code is unstable and nearly unusable. At best, it can probably gives clues on how to (re)implement it properly.

>
     
      Ok, time to rush to the theater for king arthur movie, bye all!
     
     
Not really a great movie, wasn't it ? :)


_______________________________________________
crossfire-devel mailing list
     
     crossfire-devel at lists.real-time.com
     
     
     https://mailman.real-time.com/mailman/listinfo/crossfire-devel
     
     
    


More information about the crossfire mailing list