new waveshape

[chkman]

Great! The thing keeps slamming the center. It should work well will most of the flowfields.

Are you submitting your waves to SoundSpectrum? You should, they will at least be part of the 'extras' page and there is a chance they could be part of the main package. I am just trying for the extras page but the flows that JayPro worked on might well be good enough for the main.

I do see a couple places that your wave could be optimized tho. Any expression used more than once could be put in a var, like B1+2*pi/3 so that it is calculated once but the result is used several times . Also, multiplying is faster than division so /3 could be *.333333 and /10 could be *.1. I am leaving now but will look at it this evening....

And yes, hitting the corners is good for flows like the gravity wells etc.

[Linvincible]

hey, thanks for the optimization tips

here it is in better

Aspc=0
ConB=1


A0=".6667*pi"
B0="abs(mag(s))"
B1="t-B0"
B2="B1+fft(s)"
B3="B1+A0"
B4="B1-A0"
B5="B2+A0"
B6="B2-A0"

X0="(abs(cos(B1))+abs(sin(B1)))*(s*cos(B2)+(.1+.1*B0)*cos(-t))"
Y0="(abs(cos(B1))+abs(sin(B1)))*(s*sin(B2)+(.1+.1*B0)*sin(-t))"


X1="(abs(cos(B3))+abs(sin(B3)))*(s*cos(B5)+(.1+.1*B0)*cos(-t))"
Y1="(abs(cos(B3))+abs(sin(B3)))*(s*sin(B5)+(.1+.1*B0)*sin(-t))"

X2="(abs(cos(B4))+abs(sin(B4)))*(s*cos(B6)+(.1+.1*B0)*cos(-t))"
Y2="(abs(cos(B4))+abs(sin(B4)))*(s*sin(B6)+(.1+.1*B0)*sin(-t))"

Pen="fft(s)^.5+.05*B0"

LWdt="3+20*B0"


Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260

[chkman]

Good going!

I did a little more to it. B7, B8, B9 because each is used more than once and cos and sin use more CPU than multiply or divide. B10 is only 1 multiply and one add but it is used 6 times. B11 and B12 coz each is used three times.

I didn't do s*cos(B2) and the others coz each is only used one time. Could have done B10*B11 and B10*B12 as well but I didn't notice it sooner. It would save 4 multiplications.

A0=".6667*pi"
B0="abs(mag(s))"
B1="t-B0"
B2="B1+fft(s)"
B3="B1+A0"
B4="B1-A0"
B5="B2+A0"
B6="B2-A0"
B7="abs(cos(b1))+abs(sin(b1))"
B8="abs(cos(B3))+abs(sin(B3))"
B9="abs(cos(B4))+abs(sin(B4))"

B10=".1+.1*B0"
B11="cos(-t)"
B12="sin(-t)"

X0="B7*(s*cos(B2)+B10*b11)"
Y0="B7*(s*sin(B2)+B10*B12)"

X1="B8*(s*cos(B5)+B10*B11)"
Y1="B8*(s*sin(B5)+B10*B12)"

X2="B9*(s*cos(B6)+B10*B11)"
Y2="B9*(s*sin(B6)+B10*B12)"

Pen="fft(s)^.5+.05*B0"

LWdt="3+20*B0"

Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260

[JayPro]

Also, try adding this:

NCap=1

This will make sure that any waveshape like this that has the ConB set to 1 looks like a solid mass and not a bunch of dots that give the impression of being soldered together.

[chkman]

ERRORS!!!! ERRORS!!!!! Sound the Alarm!!!!!

I was trying JayPro's suggestion about the NCap. I wanted to see what difference it would make when I realized that the newer waves weren't doing the sound thing anymore. That is, no fft zig-zagging just straight lines spinning around. I noticed that when you optimized, linvincible, that you forgot the A0 factor in B1. But that wasn't it. The big problem was that you were using B vars instead of C vars. The B vars are evaluated at the beginning of the 'frame'. In other words, only at the beginning of the draw cycle. The C vars are re-calculated for every value of 's'.

You were using B0="abs(mag(s))" all along. I guess that gives you B0="abs(mag(0))" every step of the way. Which might be okay depending on what you want but it won't change during the calculations. Using t in a B var shouldn't matter since the clock probably won't tic during the calculations but mag(s) will change coz s changes. Likewise, you would want a C var for using i in colormaps, i changes every iteration. But go to D vars for using x or y in flowfields.

So when you put fft(s) into a B var it would have the same value the whole time like using fft(0). Here is the updated wave, but I am still wondering what effect the NCap has.....


A0=".6667*pi"
C0="abs(mag(s))"
C1="t-C0+A0"
C2="C1+fft(s)"
C3="C1+A0"
C4="C1-A0"
C5="C2+A0"
C6="C2-A0"
C7="abs(cos(C1))+abs(sin(C1))"
C8="abs(cos(C3))+abs(sin(C3))"
C9="abs(cos(C4))+abs(sin(C4))"

C10="(.1+.1*C0)"
C11="C10*cos(-t)"
C12="C10*sin(-t)"

X0="C7*(s*cos(C2)+C11)"
Y0="C7*(s*sin(C2)+C12)"

X1="C8*(s*cos(C5)+C11)"
Y1="C8*(s*sin(C5)+C12)"

X2="C9*(s*cos(C6)+C11)"
Y2="C9*(s*sin(C6)+C12)"

Pen="fft(s)^.5+.05*C0"

LWdt="3+20*C0"

NCap=1

Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260

[Linvincible]

there was something wrong, some variables had to be "C" vars otherwise the blades were straight

NCap=1
Aspc=0
ConB=1

A0=".6667*pi"

B0="abs(mag(s))"
B1="t-B0"
B2="B1+A0"
B3="B1-A0"
B4=".1+.1*B0"
B5="B4*cos(-t)"
B6="B4*sin(-t)"

C0="B1+fft(s)"
C1="C0+A0"
C2="C0-A0"
C3="s*(abs(cos(B1))+abs(sin(B1)))"
C4="s*(abs(cos(B2))+abs(sin(B2)))"
C5="s*(abs(cos(B3))+abs(sin(B3)))"

X0="C3*cos(C0)+B5"
Y0="C3*sin(C0)+B6"

X1="C4*cos(C1)+B5"
Y1="C4*sin(C1)+B6"

X2="C5*cos(C2)+B5"
Y2="C5*sin(C2)+B6"

Pen="fft(s)^.5+.05*B0"

LWdt="3+20*B0"

Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260

Thanks for all your help guys!

L'invincible

[willrob]

This last version by chkman is very effective. The main difference one sees is the response in the rotating blades themselves: they react more dramatically to the beat, transforming from straight lines into waves as they turn.

[chkman]

L'invicible,

Okay, but again, using B0="abs(mag(s))" is the same as B0="abs(mag(0))" If you want to see the mag values you need to change it to a C var.

In comparing the last two waves there is not a big difference since the wave is doing so much spinning and such. With fft the effect is on the leading edge of the lines but with mag it goes in both directions. I think fft is always 0 or positive but mag will go negative.

Willrob,

Thanks, but I was just getting the wave back to what it was....

[Linvincible]

hey chkman,
looks like I was preparing my post when you posted yours, quite funny!

L'invincible

[chkman]

Yeah, both typing in about the same time.

[JayPro]

Just a thought:

If any of you are a bit concerned about the minimal rotation that occurs with this, jacking up the numbers in B4 will make it rotate around screen. Just keep it below 1; I tried .9 .

Also adding fft() to the first "s" to C's 3, 4 and 5, as will make the arms of the shape bend as the sound gets louder. Try replacing that func with mag(), too.

[Linvincible]

title: H bars, you'll see why if you try ;o)


Aspc=0
NCap=1
ConB=1


A0="100" // Thickness

C0="2*s-1"


Y0=" C0"
X0=" -0.875"

Y1=" C0"
X1=" -0.625 "

Y2=" C0"
X2=" -0.375 "

Y3=" C0"
X3=" -0.125 "

Y4=" C0"
X4=" 0.125 "

Y5=" C0"
X5=" 0.375 "

Y6=" C0"
X6=" 0.625 "

Y7=" C0"
X7=" 0.875 "



LWdt=" A0 "

C1=" fft(s)"

Pen= " C1"

Meta="reactive=5 detail=5 density=5 morphable=3"

Vers=260

oh and I updated another one, the scrolling spectrograph, so that it will work with more flowfields

NCap=0
Aspc=0
ConB=1

B0="abs(mag(s))"
B1=".2*t"
B2="B0*.15"
B3="0.99-B2"


C0="2*s-1"
C1="2*(B1-trnc(B1))-1"


X0="C1-B2"
Y0="C0"

X1="B3"
Y1="C0"

X2="-B3"
Y2="C0"

X3="C0"
Y3="B3"

X4="C0"
Y4="-B3"

Pen="fft(s)^.5+B0"

LWdt="15+10*B2"


Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260

L'invincible

[chkman]

Cool! I like the H-Bars. They will keep any flow field busy.

I was watching the Spectrograph with some Rovastar flows. When it passes the center it really lights up the flow. But the screen is clear by the time it passes again. Interesting. But it doesn't light up the wallpapers too well. I wonder how it would look with all three of the vertical ones scrolling? Just a thought...

[Linvincible]

I guess I could add something in the center...
but it's better to keep the frame as some other flowfields use it

[Linvincible]

NCap=0
Aspc=0
ConB=1

B0="abs(mag(s))"
B1=".2*t"
B2="B0*.15"
B3="0.99-B2"
C2="2*pi*s"
C3=".5*fft(s)"


C0="2*s-1"
C1="2*(B1-trnc(B1))-1"


X0="C1-B2"
Y0="C0"

X1="B3"
Y1="C0"

X2="-B3"
Y2="C0"

X3="C0"
Y3="B3"

X4="C0"
Y4="-B3"

X5="C3*cos(C2)"
Y5="C3*sin(C2)"

Pen="fft(s)^.5+B0"

LWdt="15+10*B2"


Meta="reactive=5, detail=5, density=5, morphable=3"

Vers=260