Spiral Knights - Movement speed and possibly other speeds

Movement speed and possibly other speeds

6 replies [Last post]

Thu, 09/26/2013 - 17:55

Zeddy

In this thread I'll be posting my findings on things that have speeds. Here are some old threads on the subjects:

Antistone and Culture doing laps in ATH.
An earlier Arsenal thread where I did some tests.
Lancer Knightz sword charge data.
Lancer Knightz handgun charge data.

This forum needs a spoiler tag so you can show/hide my data dumps. I went to Moorcroft Manor, running all the way from the west end to the east and back. Using an assistant (Thanks, Culture!) and two nitronomes, we measured the total distance to be 77 tiles, out of which I omitted the very left and the very right from my running field, making for 75 tiles per stretch. I ran 5 times back and forth with each, recording the run at 100fps with low settings, then divided the total amount of tiles ran (usually 5x2x75 = 750) by the time taken. Because I on average am required to turn around twice per lap, there is a bit over 1% margin of error accounting for the 8 tiles I don't always turn around perfectly on, whether in timing or distance.

These are the results, mostly known already:

Base movement is about 5.2 tiles per second. (Assuming my measurements of Moorcroft's length are correct.) I'm not sure why the deviation is so large from Antistone's speed. Maybe speed got changed since then?
Each level of MSI changes the speed by about 0.2 tiles per second, very close to 4%.
Charging Nitro reduces speed by 10%.
Charging BAB reduces speed by 40%.
Shielding reduces speed by 50%.
Charging Callahan reduces speed by 50%.
I'll be doing charge speeds and such as well. Lancer Knightz mostly have them, but they include time spent attacking.

Data dump below. You can stop reading the post from this point if you don't are. I mostly put it here so I can find it again later should I need it.
-2 MSI

length = 5 * 150
Total time = 155.44
Speed = 4.825
92.41%

-1 MSI

length = 6 * 150 - 8 (I derped up and turned four tiles early one lap)

Total time = 176.96
Speed = 5.041
96.55%

+0 MSI

length = 5 * 150
Total time = 143.66
Speed = 5.221
100%

+1 MSI

length = 5 * 150
Total time = 137.68
Speed = 5.447
104.33%

+2 MSI

length = 5 * 150
Total time = 132.9
Speed = 5.643
108.1%

+3 MSI

length = 5 * 150
Total time = 127.86
Speed = 5.866
112.35%

+4 MSI

length = 7 * 150 (Was going so fast I kind of lost track of laps)
Total time = 173.33
Speed = 6.058
116.03%

0 MSI while charging Nitronome

length = 5 * 150
Total time = 158.39
Speed = 4.735

90.67%

+4 MSI while charging Nitronome

length = 5 * 150
Total time = 137.56
Speed = 5.452

104.4% (90% of +4 MSI)

Thu, 09/26/2013 - 18:27

Bopp

good

Donkeyhaute and I are doing similar tests. We run from the west end of Emberlight (in a certain location) to the east end (in a certain location), without any turns. Our goal is to expand the earlier Lancer Knightz data, to have all weapons charging at all levels of MSI. We can't use your data directly, because you're measuring differently, but our data sets should check each other nicely.

One minor correction: Lancer Knightz' charge movement data don't include attack time. Those data are obtained by charging a weapon, and then running the race.

Thu, 09/26/2013 - 18:42

Zeddy

@Bopp

I meant charge speed as in time it takes to fill your charge bar. I've been meaning to make those measurements for a while since the charge bars got introduced. I've already done a few of these, and discovered I've been wrong about quite a few things in this area! Irontech, for instance, really does have a chargetime somewhere between Nitro and BAB, rather than being the same as Nitro. Quite embarrasing.

Testing charge speed movement at different levels of MSI isn't strictly necessary, and I shan't bother with it. If the charge reduces your movement speed by 50%, it'll do so for all levels of MSI.

Thu, 09/26/2013 - 19:09

Bopp

even better

Oh, the Lancer Knightz data on weapon charge times certainly include the attack time. Pure charge times would be great. So our projects are complementary.

In general, it seems that Donkeyhaute and I follow a different data strategy than you do. You collect data until you have figured out the pattern, and then you extrapolate. Is that fair? We try to collect all of the data possible.

Your approach leads to insights more quickly. Our approach...wait, what does it do? I don't know, but it takes a long time, so it must be valuable...

Thu, 09/26/2013 - 19:57

Zeddy

Pretty much. I gather a chunk of data, compare it, find a pattern, gather some more data to confirm (that is: try to disprove) my theory, and then make long-winded posts in Arsenal about my findings.

For instance, when I did the GTH project, I gathered every single number for every single level of (positive) health bonus. When time came to do it again for the Lost Souls project, I already knew I could linearly extrapolate between the max and no bonus values, and as such didn't bother to gather any values in between, except a few ones to confirm the rule.

I'll also try to guess how the designer was thinking at points. For instance, all values on my poison chart except duration were entirely made up by me after thinking up a nice-looking pattern that looked close enough to my gathered values that they seemed plausible enough to be the real deal. They have yet to prove wrong, all values tested always falling within 1 damage point of the excpected.

Since I don't yet have a complete grasp of the atomic values of SK, and since I don't have a firm grasp on which particular rounding mechanic SK uses for its numbers, I wouldn't put numbers up on the wiki unless I'd manually gathered them first. One day, though...

Fri, 09/27/2013 - 12:50

Culture

Charts

I posted a chart containing the data Zeddy collected. The blue dots are observed measurements. There is a dotted linear regression with the equation: speed = 0.207*msi + 5.2348.

We are measuring velocity as "tiles per second," OOO could be using some other set of units which could contribute to funky numbers like 5.2348. That is in addition to the method of measurement which cannot be perfect since we are measuring small changes within the confines of the game client.

So... doing some slight rounding, we have: speed = 0.2*msi + 5.2 which is the gray line. This is a great approximation of what was measured. I think the most important part is that each unit of MSI consistently showed a 4% boost in speed. If we ever get a chance at MSI MAX (+6) then we'll likely be running around with a 24% boost.

Fri, 09/27/2013 - 13:53

Bopp

added to Abilities

I've noted this study at the wiki page "Abilities" (where there was already a study indicating 4%).

Just to clarify, the functional relationship

SPEED = m * MSI + b

does not capture the idea of MSI increasing speed by a fixed percentage, but rather by a fixed constant --- in your case, m = 0.2 tiles. You predict varying percentage increases, between 4.2% (for MSI-2 to MSI-1) and 3.4% (for MSI+3 to MSI+4). To get ahold of a percentage increase, you want a power-law relationship like

SPEED = b * c^MSI.

Taking the logarithm of that equation, we obtain a linear relationship

log SPEED = log b + MSI log c.

The linear least squares best-fit values of log b and log c are 1.65346 and 0.0379847, respectively. Thus the best-fit power-law relationship is

SPEED = 5.22505 * 1.03872^MSI.

For all practical purposes, the conclusion is the same as in your calculation: Base speed around 5.2, and 3.9% increase per level of MSI. Cheers.

Search

Movement speed and possibly other speeds