The weights of the five attributes of each level are different. The arena model has not been tested, but it is likely to be the same.
For example, the weight of Girl 1-1 is simple 1 cute 2 lively 3 pure 2 cool 1, and the weight of Girl 1-2 is simple 3 cute 1.5 elegant 3 pure 3 warm 1
I tested Girl 1-1 for many times by kissing her hair with red beans with straight A and a petal rain powder dress with straight A.. It is found that the ratio of simplicity: cuteness: liveliness: purity: coolness is roughly 1: 2: 3: 2: 1.
So each level has different weights for five attributes. I don't know if the matching arena has this weight. This can be used to test various levels with clothes of various attributes to get results. I guess there should be no weight in the arena, but students who want to make the list at each level can pay more attention.
second, the attributes are similar
the attributes of the similarity are not scored! The attributes of Xiangke are not scored! The attributes of Xiangke are not scored! (The important thing is to say it three times.)
No score means that this attribute is counted as , and it will not be deducted. In other words, even the worst clothes (except accessories, which will be discussed below) are worth wearing even if they only have one attribute. For how to prove this conclusion, please refer to the reverse engineering post
. Many tests have found that if a certain attribute does not match (for example, cool clothes are given), the score of that attribute will become or very close to . This can be seen by changing the accessories that slightly miss (C) to seriously miss (SS). Both C and SS have little influence on the results, and even no random disturbance is significant.
but it is very important to pay attention to this feature! ! Very important! ! Very important! ! It is of decisive significance to the later work
It is found in the test that miracle warmth will also appear the same as traveling around, and a dress will be given to F directly if it is not what Da Miao wants. In this case, the score of each item is 1/1 of the theoretical value. In other words, if an attribute is expected to get 5 points, it can only get 5 points because the dress doesn't match the eight characters.
three, F.
some dresses, tops and bottoms will cause F. In the case of F, all the scores are divided by 1.
the reason for f is still unknown. it is observed that some tag clothes can only be used for the level that requires the tag, but there are exceptions.
even if it is fixed, the score will change. At present, the maximum disturbance observed can reach plus or minus 5%. I am a little suspicious that I misread the number. < P > Fourth, the score of clothes < P > Every clothes will have a five-dimensional attribute, and then decide what level to give according to which range this score falls. Considering the random factors, the score of Class A clothes with the highest score may occasionally exceed that of Class S clothes with the lowest score.
from the perspective of gear division, the score of coat = the score of bottom dress = the score of dress /2
If the weight is 1, the score of hair A is about 8, while the score of dress A is about 32. How to measure the specific truth value, we will use conclusion 2 to help.
at a certain level, use hair that doesn't match at all, for example, in 1-1 girls, use fairy elder sister, and all five attributes don't match. At this time, the influence of the five attributes of hair on the total score is almost negligible (if any). At this time, you can test the scores of all dresses, not only to find out the scoring interval of each file, but also to get the true value of each dress.
for example, 1-1 girls, test hair = fairy elder sister, dress = petal rain powder (all A)
simple ~ = 39
cute (x2)~=597
lively (x3)~=96
pure (x2) ~. Pure 345, cool 312
That is to say, the A file of a dress is about 32 plus or minus 3, and by analogy, the A file of hair is about 8 plus or minus 75.
the more data we test, the more comprehensive conclusions we can get. In this way, you can write a perfect warm simulator, and you don't have to check the properties one by one.
The following are the scores of all the positions I have measured. Please note that due to the limited physical strength and clothes, There are some big errors in some data:
hair
SS: EST125s: 95 ~ 115a: 75 ~ 9b: 55 ~ 75c: EST55
dress
SS: EST5 ~ 52s: 38 ~ 42. (Some results are from @dianashusy)
ss: 49 ~ 52s: 37 ~ 45a: 3 ~ 36b: 23 ~ 28c: est 19 ~ 21
jacket
ss: est 2725th s: 175 ~ 21a: 14 ~ 17. 21a: 14 ~ 18b: 125 ~ 14c: 8 ~ 12
socks (some results come from @dianashusy).
ss: est 86s: 57 ~ 65a: 44 ~ 55b: 4 ~ 44c: 27 ~ 3
shoes
ss: est 1s: 77 ~ 95a: 64 ~ 75b: 53 ~.
ss: est 47 ~ 5s: 37 ~ 45a: 29 ~ 37b: 22 ~ 28c: est 19 ~ 21
makeup (too few samples)
ss: est 25 ~ 28s: 19a: 13b: Generally speaking,
4 ornaments: 6.5%
5 ornaments: 11%
6 ornaments: 18%
7 ornaments: 25%
8 ornaments: 31%
. Jewelry: 46% of the total score of jewelry will be deducted < P > So if some parts only have jewelry below 1 points, please don't wear it, otherwise the score will be lower < P > VI. Future Work < P >-Measure the attribute weight of all levels < P >-Find out the punishment law of jewelry < P >-Study the scoring law of tag related items.