Page 2 of 2

Re: New features.

PostPosted: 24 Sep 2017, 00:47
by Ben
Hey guys, some news here for mapmakers:

New feature is coming: universal transitions from any surface to any. No much need for different tiles transitions anymore, it can be done from any to any, even transitions between 3 or 4 different terrains are possible.

Here how it will looks like: https://puu.sh/xH7iY/5c81a72fdf.mp4

Image

For now only 1 type of transitions is available - soft transition, similar to current stone to grass. I will add other types also, f.e. sharp (same as iron to water f.e.) and maybe some others.
Looks pretty bad right now, no offence. At this stage, I hope it isn't implemented. The "transition tiles" don't seem to blend into the rest of the terrain correctly, making it look pretty beta-like.

With some more effort, it could like pretty nice. How is this being done? Using masks?

Re: New features.

PostPosted: 24 Sep 2017, 01:38
by Rey
Looks pretty bad right now, no offence. At this stage, I hope it isn't implemented. The "transition tiles" don't seem to blend into the rest of the terrain correctly, making it look pretty beta-like.
With some more effort, it could like pretty nice. How is this being done? Using masks?
Yes it was made with masks.
As far as I know there were problems with multiterrain transitions and also many missing transitions. This feature should add ability not to use stones + long pines in between to fix it, but use masks instead. It does not mean current transitions will be replaced with automatic one.

About how it looks - I took that mask from stone to grass transition. So its just copy of it. If you put new mask between stone and grass you will get exactly what we have now.
Obviously, that every mask can't be good for every 2 types of surfaces. That could be improved by adding more types of masks, as I said before.

Re: New features.

PostPosted: 05 Oct 2017, 04:51
by Rey
Hey guys, I've just finished new army stats charts, added 'lost' and 'defeated' options.

Image

Want to know your opinion on new 'Army Power' value - its a sum of all warriors strength, where strength is some abstract value (in points).
So f.e. militia is 1 point and knight may be 6 or smth.
Then on 'Army power' chart player with 10 knights will have same value (60points) as player with 60 militia.
For now I set some approximate values as:
  Code:
1, 2.4, 5.2, // ut_Militia, ut_AxeFighter, ut_Swordsman 2.2, 4, // ut_Bowman, ut_Arbaletman 2, 4, // ut_Pikeman, ut_Hallebardman 3.3, 6, // ut_HorseScout, ut_Cavalry 5.3, 1.5, 1.5, // ut_Barbarian, ut_Peasant, ut_Slingshot 5.3, 2.1 // ut_MetalBarbarian, ut_Horseman
FYI, KMR units parameters:

Image

Any suggestions?

Re: New features.

PostPosted: 06 Oct 2017, 13:52
by Ben
Yay! More stats!

As for your question of the "army power," when concerning stats, I think that the economical cost should be the sole factor of a unit's value, not the strength. For example, a crossbow man is very powerful and cheap and easy to produce. A scout is very expensive and very weak. In my opinion, a scout should still represent higher "army power" in stats.

Re: New features.

PostPosted: 06 Oct 2017, 22:14
by Rey
I think that the economical cost should be the sole factor of a unit's value, not the strength
On the 'Army' page we have to talk not about unit's economic cost, but of its value on the battlefield IMO. For economic stats we already have 'weapons produced' on economic page, but plain numbers, without production 'weight' though.
From your example: I can't say scout is very weak - he is not good in terms of cost/value because of expensive horse.
But in terms of his value on the battlefield its only depends of his stats, which are the best for leather unit.

Re: New features.

PostPosted: 08 Oct 2017, 15:28
by Strangelove
Not sure if thats what you've been looking for rey, but I tried to calculate a bit more accurate Unit Power for you, however it still seems arbitrary...

Here are the results:
unitpower.png
Here the formular:
  Code:
(Health x 30 + (Attack + AttackvsCavalry/4) x AverageAspd + (Defence + (DefencevsBow/4 + DefencevsCrossbow)) x 40)/10
The forumlar is kind of self explanatory, however here a few details: I devided the Attack vs Cavalry by 4 cuz ~1/4 of KaM units are considered Cavalry. The same thing is done with the Defence vs Bow/Sling. Defence vs Crossbowman is already 1/4 of the Bow/Sling defence value, so I left it at that. The total Attack is then multiplied with the avarage attack speed, calculating average DPS.
Why i multiplied the Health by 30? I calculated how much health each unit has in respect to its damage output. 30 is the average across all units. I have to admit that this is also very arbitrary...
(For all those calculations i left out the TPR Units and only applied the formular on them afterwards.)



EDIT:
If we want to be even more specific we could also calculate Defence more accuratly, meaning in relation to Attack power like we did for Health. An average for Attack/Defence for all units would be ~24, insted of the 40 I used previously. We could also calcule the AttackvsCavalry, DefencevsBow and DefencevsCrossbow Values more precicly: Instead of applying a correction value of 1/4, we multiply by 2/9 (2 of the 9 Units are affected). I also didn't favor DefencevsCrossbow this time.

Here the new formular:
  Code:
(Health x 30,5 + (Attack + AttackvsCavalry/9 x 2) x AverageAspd + (Defence + (DefencevsBow + DefencevsCrossbow)/9 x 2) x 24)/100
This time I devided the result by 100 and added one decimal to match Rey's chosen values.

New values:
unitpower2.png

Re: New features.

PostPosted: 08 Oct 2017, 21:48
by Ben
I think that the economical cost should be the sole factor of a unit's value, not the strength
On the 'Army' page we have to talk not about unit's economic cost, but of its value on the battlefield IMO. For economic stats we already have 'weapons produced' on economic page, but plain numbers, without production 'weight' though.
From your example: I can't say scout is very weak - he is not good in terms of cost/value because of expensive horse.
But in terms of his value on the battlefield its only depends of his stats, which are the best for leather unit.
Okay, that's fair enough. I agree with your reasoning.

With your version, "army power" could be useful for making a "kill quality" stat, which would be even better in my opinion.

Re: New features.

PostPosted: 10 Oct 2017, 11:49
by Rey
Another game statistic improvements have been just finished:
1. ability to see detailed (MP) stats for single player game
2. MP ware stats - added GDP option, to see what was an actual market price of produced wares. Works for 'All wares', 'Food' and 'Weapons'.
3. all MP stats are now possible to see not only for each player, but for each team as well.

Check it out:

Image

Re: New features.

PostPosted: 10 Oct 2017, 12:43
by Rey
Not sure if thats what you've been looking for rey, but I tried to calculate a bit more accurate Unit Power for you, however it still seems arbitrary...
That's a good way to calculate it with some formula, instead of just 'assuming' the values.
I calculated how much health each unit has in respect to its damage output
Can you explain it more detailed?

Some thoughts on your formula:
1. You calculated average values for parameters assuming that we can use unit types there (f.e. '1/4 of KaM units are considered Cavalry '). But its not 100% correct, because in real game all of the units are represented in different numbers. Some are more frequent and some could be rarely seen. F.e. we do not have as much scouts as axe fighters. Then lance carriers will get more points, because of that, while not seeing as many scouts as axe fighters in the real games (also scouts are much more expensive)
**Upd. Some crazy idea just came to my mind - what if instead of using 'average number by unit type' value we use real values from the current game, calc unit power depending on real game situation. F.e. calc all units of your opponents and see how many horse units they have. If no horses - then your lance carriers are useless.
Or may be its too complicated, unclear and confusing...

2. You missed Speed parameter, horse units should get their bonus points for higher speed. Similar for units with charge ability - axe fighters have more chances to catch bows then lance carriers.

3. I think attack speeds for ranged units are not correct. IIRC bowman's delay between hits is 24 ticks on average, crossbowman's is 40 and rogue's is 36.

Also I have no idea what is a measure value of attack speed - hits per 'some period of time', but what is that time actully.

Overall results has questions - Knight is only 1.2 points stronger then militia. He should be stronger IMHO.
But this is a right way to calculate. Just need to adjust coefs.

Re: New features.

PostPosted: 10 Oct 2017, 18:20
by Strangelove
Well, like I said, I tried to calculate values, however, I wasn't satisfied with the results. They still seemed pretty arbitrary.

I tried to bring all Values in line with the AttackPower of the Units, thats why i calculated an avarage across all units to get a coeff I can apply. Maybe thats the wrong way to do it?


Anyhow, to adress your points:
1. That 1/4 of the units are considered cavalry is ofc not entirly correct. The presice value would be 2/9, which is used in my updated calculation. That is of cause not really correct either, like you pointed out. However, I did not have any stats on how often a certain unit typ is used, so I just checked how many units of the 9 (TSK only) are considered cavalry units and that would make 2.
Calculating a different value for the units each game seems a litte over the top and confusing if you want to compare games, but it would certainly be more accurate.

2. You are right about that. I can include that. How to calculate charge would have to be discussed tho.

3. I used the values provided. I do not know the underlying gamecode.


The coefs can be adjusted anytime. The values seemed odd to me as well. Maybe there is something wrong with the approach I took?
What do you propose to make the values seems more accurate?

Re: New features.

PostPosted: 11 Oct 2017, 01:48
by Rey
However, I did not have any stats on how often a certain unit typ is used, so I just checked how many units of the 9 (TSK only) are considered cavalry units and that would make
Yes, we have to assume smth, I do not see other way.
Calculating a different value for the units each game seems a litte over the top and confusing if you want to compare games, but it would certainly be more accurate.
that is really crazy idea, not good to compare games indeed
I used the values provided. I do not know the underlying gamecode.
Yep, I just used screenshot I had, it was not accurate enought :rolleyes:
Maybe there is something wrong with the approach I took?
Approach looks neat in general, so we have to adjust coefs to get better.

Can you give a link for your table, so I do not need to recreate it again for testing with different coefs?

Btw, thank you very much for your help!

Re: New features.

PostPosted: 30 Dec 2017, 18:01
by Strangelove
However, I did not have any stats on how often a certain unit typ is used, so I just checked how many units of the 9 (TSK only) are considered cavalry units and that would make
Yes, we have to assume smth, I do not see other way.
Calculating a different value for the units each game seems a litte over the top and confusing if you want to compare games, but it would certainly be more accurate.
that is really crazy idea, not good to compare games indeed
I used the values provided. I do not know the underlying gamecode.
Yep, I just used screenshot I had, it was not accurate enought :rolleyes:
Maybe there is something wrong with the approach I took?
Approach looks neat in general, so we have to adjust coefs to get better.

Can you give a link for your table, so I do not need to recreate it again for testing with different coefs?

Btw, thank you very much for your help!
Whoops, sorry! Totally missed your rely! Yeah, I do have the excel-file somewhere.

Here you go:
KaMUnitPower.xlsx

Re: New features.

PostPosted: 12 Jan 2018, 14:54
by StefanTheGreat
Hi Ray! Thanks for your work with new kam. I like a lot of new things that you added, especialy ability to select ally's house/unit and see what he have,
I just post this viewtopic.php?f=24&t=3635&p=49915#p49915 with some ideas for new kam. I hope you will read this. :)

Peace! :D