Dolphin, the GameCube and Wii emulator - Forums

Good morning
Not sure if you saw my latest reply from pm conversation, so I copy it here as well:
So I think we are heading in the right direction for sure!

I tuned the code to an extreme, so it's really obvious (for me) what is going on. I think what we need is to make the transition between the stop and max speed of the accell animation more smooth or linear?
Because now it almost looks like there are 2 speeds in the accell (stop and max), but nothing in between. Hopefully what I try to say make some sense

Here I tried to draw what we need. Sorry for my bad drawing

Somehow managed to RUN the code from your link and saw the shape of your current tunning.

So if that is what is actually going on on the accels with the modification, we might need actually something simple like this, I think.

hmm, so this might be the closest I managed to get?

#R version 3.3.2

y <- rnorm(20)
y[1] <- 0.016
for (i in 2:20) {
x = y[i-1]
y[i] <- 1.0 * ( x + 0.1 * x )
}
plot(y)
print(y)

Hmm sorry. Turns out that plotting stuff isn't accurate enough. I would have to snoop through more of the game's math to make it accurate. Here's my current test:

$gecko accel from stop2
c2b228e8 00000006
c03f129c c282ff00
fc01a040 41a00020
c29d006c c2a2fefc
c2c2fef8 ec21b028
ee940572 ec340072
ec36082a 00000000
043420c0 3f8ccccd
043420bc 3f000000 //ramp up multi
043420b8 3f000000 //base speed

the math is:
constant modifier = 1.1ish or 1.9ish.
distance moved goes from near zero to about 0.7

new modifier = base speed + (Constant Modifier - base speed) * distance moved * multi

awesome!
Testing...

Here is a simpler: base speed + position * constant

$gecko accel from stop simple
C2B228E8 00000005
C03F129C C282FF00
FC01A040 41A00018
C29D006C C2A2FEFC
C2C2FEF8 EC340072
EC36082A 00000000
043420c0 3f8ccccd
043420b8 3f000000 //speed

If neither of these are close. I'll try to figure out more of the math happening.

Tested both.
I like the $gecko accel from stop2 more, because it gives me more freedom to tune things.

But in both cases I think the problem we still have is that the character gains the speed too slow and then there looks to be a "jump" from the slow to the fast phase of the animation.
At least that what my eyes see   

https://www.youtube.com/watch?v=IjVJn1JCXaY


It should be really smooth close to this progress:



But yeah we are definitely getting there!
I can already feel that once we sort this out the game will look and feel so MUCH better. Because the gameplay (and skating in general) is basically based on these accelerations. Players in a hockey game accelerate very very often. It's not some tiny detail at all, it's pretty essential and has an effect on everything.

Ok, because we have to base it off position, the further along the animation is, the more accel we get. So I can try to do:
base + position / (position + 1), which should counteract it a bit.

$gecko accel from stop3
c2b228e8 00000006
c03f129c c282ff00
fc01a040 41a0001c
c29d006c c2a2fefc
c2c2fef8 eeb5a02a
ee94a824 ec36a02a
60000000 00000000
043420c0 3f8ccccd
043420bc 3f800000 //probably ignore
043420b8 3f600000 //base

Or maybe
base + position^2

$gecko accel from stop 3.1
c2b228e8 00000005
c03f129c c282ff00
fc01a040 41a00018
c29d006c c2a2fefc
c2c2fef8 ee940532
ec36a02a 00000000
043420c0 3f8ccccd
043420b8 3f600000

It's hard to test thoroughly because the animation is so short. U remember I tried to prolong the animation the last time I worked on it and it had too many safety checks.

Great, currently driving a car, give me an hour or two, please
Thanks!

Ok, I'm not certain any of these will work and they each require testing various values to see if it could.

Also, did you say if I needed to filter what gets affected by this mod better?