genr

I'm trying to create two planes one in front of another, using texture with alpha chanel for the front plane so that I can see through the holes to look at the back plane. It works fine in the jit.window, but when I switch to fullscreen mode, the front plane turn to be behind the back plane. I tried to change the depthbuffer setting, only one of jit.window and fullscreen would work, but not both. Is there any known reason for this difference?

strange-difference-between-jit-window-and-fullscreen

Without knowing what your patch looks like, it's hard to say what's

going on. It could be any number of things. Things work fine in this

patch. It was thrown together quickly, so there are some things I

wouldn't do if performance was an issue. Anyway, it'll give you a

Without knowing what your patch looks like, it's hard to say what's
going on.  It could be any number of things.  Things work fine in this
patch.  It was thrown together quickly, so there are some things I
wouldn't do if performance was an issue.  Anyway, it'll give you a
start.

hi wes, sorry for not including the patch in my last post, thanks for the example, i tried to modify my patch based on your example, still not working. i extract that part with the depth problem from a huge messy patch. you need to use mouse to make some swirl in the jit.pwindow, it will apear in front of the another plane in the jit.window, but when switch to fullscreen, the swirl is hidden behind the other plane. if i set the depthbuffer of the jit.window to be 1, the result is the reverse, the swirls are in the front in the fullscreen mode, but hidden in the jit.window.

hope you could help me find the problem, thanks.

So your problem was that you didn't have @depth_enable 0 for the

videoplanes. Sometimes @blend_enable 1 will make this not necessary,

but if one of your planes is opaque, this will not necessarily be the

case. Also, you may want to draw the video planes in an explicit

order with either @automatic 0 or @layer .... Thirdly, since you're

doing water simulation, you may want to check out xray.jit.water.

It's faster, more accurate and provides better control of wave

parameters. The convlution method in your patch explicitly links wave

speed to the convolution kernel size. You can get a UB version of

http://www.mat.ucsb.edu/~whsmith/xray-mac-UB.zip

So your problem was that you didn't have @depth_enable 0 for the
videoplanes.  Sometimes @blend_enable 1 will make this not necessary,
but if one of your planes is opaque, this will not necessarily be the
case.  Also, you may want to draw the video planes in an explicit
order with either @automatic 0 or @layer ....  Thirdly, since you're
doing water simulation, you may want to check out xray.jit.water.
It's faster, more accurate and provides better control of wave
parameters.  The convlution method in your patch explicitly links wave
speed to the convolution kernel size.  You can get a UB version of
xray.jit.water from http://www.mat.ucsb.edu/~whsmith/xray-mac-UB.zip .

thank you very much for your help. your mdification of setting the depth_enable to 0 still not work here, but it works when i draw the planes in an explicit order following your first example.

i also tried your water object, it is much better than that one i got from the jitter example. i have some questions about the water simulation want to ask you. as you have seen in my patch, i want to use the simulation to draw the swirls explicitly, the problem is that, the resolution is very low, i can see pixels in the swirls, the quality is very bad, i tried to increase the matrix size, but that killed the speed, when i tripple the size, the simulation almost not working any more. also, in your water simulation, how can i increase the initial radius of the swirls?

thank you very much for your help. your mdification of setting the depth_enable to 0 still not work here, but it works when i draw the planes in an explicit order following your first example. 

There are a number of ways to do this, but you have to be careful to

http://en.wikipedia.org/wiki/Courant%E2%80%93Friedrichs%E2%80%93Lewy_condition)

otherwise, the simulation will blow up on you. You can increase the

timestep, increase the wavespeed,...whatever will warp the time of the

simulation. If you want smooth interpolation in opengl, you may want

to send the water simulation to the graphics card as 3d geometry via

jit.gl.mesh and use some kind of per-pixel lighting to smooth things

out. You're never going to be able to simulate water at a super high

resolution, it's an expensive thing to do as there are 2 spatial

derivatives and 1 temporal derivate in a simple model. THere are

however clever tricks to get around this. Something like a Phong

shader would do nicely here. You can even calculate transparency

based on depth in the shader if you just want the swirls. This could

be setup to blend away the surface if it is at its default height

And another wat to increase the initial radius of the swirls would be

to stimulate with a circle instead of a point like in the help file.

I've used 20x20 matrices before and added the values stored in it to

the water matrix at the location where I wanted a swirl. I did it

with a javascript, but it can be done in a patcher. I'll post that

> thank you very much for your help. your mdification of setting the depth_enable to 0 still not work here, but it works when i draw the planes in an explicit order following your first example.

> i also tried your water object, it is much better than that one i got from the jitter example. i have some questions about the water simulation want to ask you. as you have seen in my patch, i want to use the simulation to draw the swirls explicitly, the problem is that, the resolution is very low, i can see pixels in the swirls, the quality is very bad, i tried to increase the matrix size, but that killed the speed, when i tripple the size, the simulation almost not working any more. also, in your water simulation, how can i increase the initial radius of the swirls?

There are a  number of ways to do this, but you have to be careful to
not violate the CFL condition
(http://en.wikipedia.org/wiki/Courant%E2%80%93Friedrichs%E2%80%93Lewy_condition)
otherwise, the simulation will blow up on you.  You can increase the
timestep, increase the wavespeed,...whatever will warp the time of the
simulation.  If you want smooth interpolation in opengl, you may want
to send the water simulation to the graphics card as 3d geometry via
jit.gl.mesh and use some kind of per-pixel lighting to smooth things
out.  You're never going to be able to simulate water at a super high
resolution, it's an expensive thing to do as there are 2 spatial
derivatives and 1 temporal derivate in a simple model.  THere are
however clever tricks to get around this.  Something like a Phong
shader would do nicely here.  You can even calculate transparency
based on depth in the shader if you just want the swirls.  This could
be setup to blend away the surface if it is at its default height
(i.e. not displaced).

And another wat to increase the initial radius of the swirls would be
to stimulate with a circle instead of a point like in the help file.
I've used 20x20 matrices before and added the values stored in it to
the water matrix at the location where I wanted a swirl.  I did it
with a javascript, but it can be done in a patcher.  I'll post that
patch tommorrow.

On 10/17/06, xiaohua sun  wrote:
>
> thank you very much for your help. your mdification of setting the depth_enable to 0 still not work here, but it works when i draw the planes in an explicit order following your first example.
>
> i also tried your water object, it is much better than that one i got from the jitter example. i have some questions about the water simulation want to ask you. as you have seen in my patch, i want to use the simulation to draw the swirls explicitly, the problem is that, the resolution is very low, i can see pixels in the swirls, the quality is very bad, i tried to increase the matrix size, but that killed the speed, when i tripple the size, the simulation almost not working any more. also, in your water simulation, how can i increase the initial radius of the swirls?
>
> thanks,
>


I'm looking forward to your swirl patch :)

It seems that I got the low resolution becasue my method is way too simple. I did calculate transparency based on the depth and set those parts at default height to be 0. But I simply used this infomation to create a matix as texture for the surface. Seems that I'll have to look at the shaders which I didn't have the gust to get into before.

It seems that I got the low resolution becasue my method is way too simple. I did calculate transparency based on the depth and set those parts at default height to be 0. But I simply used this infomation to create a matix as texture for the surface.  Seems that I'll have to look at the shaders which I didn't have the gust to get into before.

http://mat.ucsb.edu/~whsmith/WaterSploosh.zip

In this patch I'm using a fragment shader here to do the images

processing based on the water simulation. I'm also using JS to handle

user interaction. You can draw in the pwindow and a "splash kernel"

created with jit.expr will leave a 20x20 pixel splash in the water.

The JS uses jit.matrix src/dstdim stuff for efficiency. Not that I

grab the values at the splash location and add the splash kernel to

them. I found that just setting them to an absolute value created

funky artifacts in the stimulation since it broke up the continuity.

To change how the shader is alpha blending based on the water, have a

look at the gl_FragColor = vec4(color.rgb, displacement); line and

replace displacement with your own function of displacement. Right

now it's set to only show the video where the water is in peaks and

In this patch I'm using a fragment shader here to do the images
processing based on the water simulation.  I'm also using JS to handle
user interaction.  You can draw in the pwindow and a "splash kernel"
created with jit.expr will leave a 20x20 pixel splash in the water.
The JS uses jit.matrix src/dstdim stuff for efficiency.  Not that I
grab the values at the splash location and add the splash kernel to
them.  I found that just setting them to an absolute value created
funky artifacts in the stimulation since it broke up the continuity.
To change how the shader is alpha blending based on the water, have a
look at the gl_FragColor = vec4(color.rgb, displacement); line and
replace displacement with your own function of displacement.  Right
now it's set to only show the video where the water is in peaks and
valleys.

Awesome patch! the water already looks very alive just with the video. The shader file will be a great template for me to start playing with shaders. one more question, I understand your reason of setting the splash matrix to a signle value 1, just wondering whether there is a way to make the splash area less flat, ignore this question if I'm being too greedy :) I'm already very grateful for your help.

Sure, do something like jit.expr "hypot(snorm[0], snorm[1])" and pipe

it into jit.op @op >p or something similar. You could also build a

gaussian with jit.expr. Many possibilities.

> Awesome patch! the water already looks very alive just with the video. The shader file will be a great template for me to start playing with shaders. one more question, I understand your reason of setting the splash matrix to a signle value 1, just wondering whether there is a way to make the splash area less flat, ignore this question if I'm being too greedy :) I'm already very grateful for your help.

Sure, do something like jit.expr "hypot(snorm[0], snorm[1])" and pipe
it into jit.op @op >p or something similar.  You could also build a
gaussian with jit.expr.  Many possibilities.

On 10/19/06, xiaohua sun  wrote:
>
> Awesome patch! the water already looks very alive just with the video. The shader file will be a great template for me to start playing with shaders. one more question, I understand your reason of setting the splash matrix to a signle value 1, just wondering whether there is a way to make the splash area less flat, ignore this question if I'm being too greedy :) I'm already very grateful for your help.
>
>
> genr
>


Thank you very much for the suggestions!


i tried to access your example with water splash but the link is broken. (I think it's normal because this thread is from three years ago).

Could you post the example again if you still heave it .

strange difference between jit.window and fullscreen