EMU618社区

 找回密码
 立即注册
搜索
查看: 3438|回复: 5

[讨论] 【光明力量—古代封印】关于成长率的问题~~

 关闭 [复制链接]

该用户从未签到

发表于 2007-10-4 15:33:02 | 显示全部楼层 |阅读模式
在日攻看到某人帖出的国外高人们的研究~~全是鸟语……谁能翻译一下??

Hitting a dead end with my OpenGL... so I did some tests on this one. Might be relevant to anyone making a SF clone... or whatever.

Each character+class combination has a base stat, a projected stat, and a stat growth type for each of the statistics (HP, MP, ATT, DEF, and AGI).

The base stat is used when a character is first spawned. It is the value the character is given for that class at Level 1.

The projected stat is sort of the intended lowest value that stat can be at level 30. The amount of actual lowest stat gain you can get over this time is simply Projected Stat - Base Stat.

The growth types are linear, early, late, middle, late+early and none. These are just names I've given the growth curves, for the peak growth levels.

After level 30, every stat that CAN grow, will do so at a 50/50 chance of +1 or +2 each level. For many characters this is a (bad) kick in the stats (I know Bowie mellows out as a swordyguy after 30), for others (Kiwi) it can be a massive gain. Simply multiply 1.5 times the level over 30 to estimate stat gain over that time. This is true as of basic classes and promoted classes.

Following is a chart that's a rough estimate of the rates of growth for the various growth types. The format is Level, Percentage of the Projected Gain (Projected Stat - Base Stat) attained, and the Rate of Change in (parentheses). The RoC just makes it easier to see the slight difference between these types.

Linear

Level - Percentage (Rate of Change)
1 - 0
5 - 16 (+16)
10 - 34 (+18)
15 - 52 (+18)
20 - 70 (+18)
25 - 87 (+17)
30 - 106 (+19)

Late

1 - 0
5 - 11 (+11)
10 - 26 (+15)
15 - 43 (+17)
20 - 61 (+18)
25 - 81 (+20)
30 - 102 (+21)

Early

1 - 0
5 - 17 (+17)
10 - 36 (+19)
15 - 54 (+18)
20 - 71 (+17)
25 - 86 (+15)
30 - 101 (+15)

Middle

1 - 0
5 - 12 (+12)
10 - 29 (+17)
15 - 48 (+19)
20 - 69 (+21)
25 - 87 (+18)
30 - 101 (+14)

Early+Late

1 - 0
5 - 18 (+18)
10 - 35 (+17)
15 - 49 (+14)
20 - 62 (+13)
25 - 79 (+17)
30 - 101 (+22)

These numbers were derived by taking Jaha, changing his projected HP to 109 (Making his projected gain a simple 100, which yielded these nice whole percentage numbers since the game won't give you decimal gains, obviously), and leveling him to level 40 over and over with the various growth types, recording his HP at every_single_level. I set medical herbs to cause a level up on use without a possibility of breaking - would have driven me nuts otherwise, clearly, clearly.

I also leveled him up to level 99 as a Gladiator once to see if promotion made any difference to how this system works - it doesn't. Percentages appear about the same. The 1.5 gain after level 30 held true.

A more accurate estimate could be yielded by repeating the test over and over and averaging the results, but I'd rather not do it myself.  

This post has been edited by Space King on Tue Jun 26, 2007 6:20 am

貌似归结起来成长率有一个上限~~前面长得高了~后面就难长了……

该用户从未签到

发表于 2007-10-4 15:38:38 | 显示全部楼层
打一个死胡同,我的opengl ...所以,我做了一些测试,就这一个。可能相关的人作出的sf克隆… …或什么的。

每个字符+类结合了统计基础,统计预测和统计生长型,每个统计(惠普,国会,垂体瘤att , 〜 ,碘化银) 。

基地stat是用来当一个人物,首先是引发。它是价值的特点是由于这一类1级。

预计stat是一种预定的最低值,统计,可以在一级30 。金额实际最低stat的增益可以渡过这个时间是纯粹统计预测-s tat的基地。

生长类型是线性,早,晚,中,晚+早无。这些都只是我的名字已经获得的生长曲线,为繁忙的增长水平。

经过30级,每一个统计,可以成长,将做一个50/50机会+1或+2每个层面。许多字,这是一个(坏)踢在统计(我知道鲍伊mellows作为swordyguy三十日后) ,其他(奇异果) ,它可以大量收益。简单乘以1.5倍的水平超过30估计stat的增益超过时间。这是真正为基础班及晋升班。

以下是一个图表,这是一个粗略的估计增长率为各类型生长。格式水平,百分比推算增益( stat的预测-基地统计)达到了,而且变动率(括号内) 。中华民国只是使它更容易见到略有差异,这些类型。

Level - Percentage (Rate of Change)
1 - 0
5 - 16 (+16)
10 - 34 (+18)
15 - 52 (+18)
20 - 70 (+18)
25 - 87 (+17)
30 - 106 (+19)

Late

1 - 0
5 - 11 (+11)
10 - 26 (+15)
15 - 43 (+17)
20 - 61 (+18)
25 - 81 (+20)
30 - 102 (+21)

Early

1 - 0
5 - 17 (+17)
10 - 36 (+19)
15 - 54 (+18)
20 - 71 (+17)
25 - 86 (+15)
30 - 101 (+15)

Middle

1 - 0
5 - 12 (+12)
10 - 29 (+17)
15 - 48 (+19)
20 - 69 (+21)
25 - 87 (+18)
30 - 101 (+14)

Early+Late

1 - 0
5 - 18 (+18)
10 - 35 (+17)
15 - 49 (+14)
20 - 62 (+13)
25 - 79 (+17)
30 - 101 (+22)


这些数字得出以亚哈,改变他的预测,惠普109 (使他的预测得到一个简单的100 ,而取得这些尼斯整个人数比例自游戏,不会给你点数上扬,显然) ,并平整他40级遍地与不同生长类型,记录他在惠普every_single_level 。我定中草药导致一级对使用未经可能性打破-将带动我坚果,否则很明显,清楚。

我也为平地,他到99级作为一个斗士,一旦看到,如果推广作出任何差别如何制作品-它没有。百分比似乎大致相同。 1.5增益水平后, 30日举行的真实。

一个更准确的估计可能产生的重复测试遍地平均结果,但我宁愿不要自己了。

这个职位一直由编辑空间国王周二2007年6月26日上午06时20分




这是GOOGLE的自动翻译.   虽然语句不很通顺,但大体的意思是可以看的明白了

该用户从未签到

 楼主| 发表于 2007-10-4 15:41:03 | 显示全部楼层
:L …………基本能看懂的说……

签到天数: 48 天

[LV.5]常住居民I

发表于 2007-10-4 15:41:05 | 显示全部楼层
A more accurate estimate could be yielded by repeating the test over and over and averaging the results, but I'd rather not do it myself.  

This post has been edited by Space King on Tue Jun 26, 2007 6:20 am
只翻译了这点


一个更准确的估计可能产生的重复测试遍地平均结果,但我宁愿不要自己了。

这个职位一直由编辑空间国王周二2007年6月26日上午06时20分

该用户从未签到

 楼主| 发表于 2007-10-4 18:49:47 | 显示全部楼层
研究了一下rom~~从地址1ee2f0开始~~分别是人物的能力~projected值~魔法种类~习得等级~


修改了projected值(也就是上限参考值)后~攻防素有很大机会都可以+2了~~:lol

该用户从未签到

 楼主| 发表于 2007-10-5 08:57:00 | 显示全部楼层
呃……中文版的修改过后偶尔会死机的说……:yumen:
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

Archiver|手机版|小黑屋|国治模拟精品屋 ( 沪ICP备15012945号-1 )

GMT+8, 2024-11-14 15:29 , Processed in 1.070313 second(s), 18 queries , Gzip On.

Powered by Discuz! X3.4

© 2001-2023 Discuz! Team.

快速回复 返回顶部 返回列表