Fluid shader Alpha

[Schoenberger]

Hi

I have an alpha version of the fluid shader to play with.
No samples, no help, not all features finished, but it renders.

You can render:
- Latest alpha of emFluid
(more features in cooperation with Eric planed)
- Maya .mc cache files (you need to add a geometry shader to the scene)
- (not implemented yet: Phoenix cache files)
- (not implemented yet: Field3D cache files)
- (no response yet: FumeFX... so perhaps no implementation)

How it works:
Plug the fluid shader into a picker and the picker into density of the volume.
Plug the fluid shader into the color of the volume shader if wanted.

Use either fluid_value_color or fluid_value_float to get a value of the fluid.
Plug these shaders directly into the fluid shader.
Or use a gradient or fcurve inbetween.

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Schoenberger]

Doh'

The links:
www.BinaryAlchemy.de/download/110314__baVolume_v4alpha
0.01___Softimage_2010.zip
www.BinaryAlchemy.de/download/110314__baVolume_v4alpha
0.01___Softimage_2011.zip

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Steven Caron]

i figured as much...

[Steven Caron]

thanks for this, but are we allowed access to the emFluid version you are referring to? if not, i can't test it. i actually dont own the latest version, but i do own version 3.

can you post a render for those who can't test it?

s

On Mon, Mar 14, 2011 at 12:04 PM, Schoenberger <X...@digidragon.de> wrote:

[Schoenberger]

emFluid Alpha 008 should be available for download.
I am currently doing some render tests on my own.
But they are not that special.
As I said before, you need a good/detailed sim, the shader can not do much.
(if you have texture coordinates it can do more, testing that with
Maya, will probably post a video later)

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[javier gonzalez]

Thanks, but i have problems with the links, it seems broken.

2011/3/14 Schoenberger <X...@digidragon.de>

[Schoenberger]

Just tested, they work fine.

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Robert Chapman]

oh - it only works with Alpha version of emfluid? not the curent download version 3.01? 

also can we not set density attributes per particle or build the structure necessary for the shader in ICE? or is it my lack of understanding and you need the volume cell structure already present per particle...

Nice one though Holger, cant wait to see something! :)

[Stefan Kubicek]

Afaik they entered some strategic alliance with Cebas/Final Render whshortly after PhoenixFD from Chaos Group was released.
That might also influence their stance towards other renderers, but tbh that's just a wild guess.

>> - (no response yet: FumeFX... so perhaps no implementation)
>>
>>
>>
>>
>

--
-------------------------------------------
Stefan Kubicek Co-founder
-------------------------------------------
keyvis digital imagery
1050 Vienna Wehrgasse 9 Austria
Phone: +43/699/12614231
--- www.keyvis.at ste...@keyvis.at ---
-- This email and its attachments are
--confidential and for the recipient only--

[Robert Chapman]

sorry Holger, totally didnt realise that emfluid 4 demo was available for download here -
http://www.mootzoid.com/wb/pages/softimagexsi/emfluid4.php

just downloaded 3.01 thinking it was the latest version - doh! going to go try give it a shot now, thanks

[Gene Crucean]

Does anyone know if there is any builtin way to get fluid data out of blender? Plugin maybe?

On Mon, Mar 14, 2011 at 12:04 PM, Schoenberger <X...@digidragon.de> wrote:

--
[Gene Crucean] - [VFX & CG Supervisor/Generalist]
** Freelance for hire **

[Robert Chapman]

nope can't get it working with 4.0 either , has no effect - I guess theres another version after this that it works with ?

[Schoenberger]

Two test videos with Maya fluids and simulated texture coordinates:
www.BinaryAlchemy.de/ftp_file/baFluidCacheFile_B.mov
www.BinaryAlchemy.de/ftp_file/baFluidCacheFileA_.mov

Another hint:
You can mix the default particle shader and the fluid shader.

@Robert Chapman

> also can we not set density attributes per particle or build the structure
> necessary for the shader in ICE? or is it my lack of understanding and you
> need the volume cell structure already present per particle...

There are no particles. You have the fluid grid. And the fluid grid
can only be generated by a fluid simulator.

Perhaps it would be possible that you create an array of floats in ICE
and then use that instead of a fluid grid, but that's something
someone else has to figure out. Then I can support it.

@Gene Crucean:

> Does anyone know if there is any builtin way to get fluid
> data out of blender? Plugin maybe?

Forgot that format. Perhaps these cache files will also be able to
load. As I can get the specification of the file easily.

@Ben
> Those won't work if I click on them obviously.
Spaces..., need to remove them, usally I never use them.

@Robert Chapman

nope can't get it working with 4.0 either , has no effect

I guess theres another version after this that it works with ?

I will setup a sample scene.

regards,

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Steven Caron]

exactly! if we could have a field/voxel data type your fluid shader expects we could fill it using our own methods. i haven't had much luck building field3d on windows but my idea is to make a custom ICE attribute which exposes the field3d field types ( specifically the sparse type ).

steven

[Steven Caron]

are singleton array's supported with the scalar_attribute shader? if so then we should be able to bring in any array of data into the render tree and pass to your fluid shader, right?

s

[Schoenberger]

Ok, just tested the emfluid version which you can download.
It seems to be an older version.
So we have to wait for Eric (On vacation for two weeks).
Bad that there is no version display somewhere or am I missing something?

> i haven't had much luck building field3d on windows

I have to implement it into my source, so I will check if there is an
issue with that. Bad that I could not use it for my shader, as I need
one big block of memory for a geometry shader.

> are singleton array's supported with the scalar_attribute shader?
> if so then we should be able to bring in any array of data into
> the render tree and pass to your fluid shader, right?

The fluid shader would need to get the data directly. Not the
scalar_attribute shader. But that would not be an issue.
If you have an ICE array of float, I could access it.

Requires some changes to define the name of the ICE float array and an
ICE attribute to get the size of the grid.

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Steven Caron]

i am not sure i follow... are you suggesting you would get it directly from ICE yourself within the shader? if so that means we would have to reserve an attribute name correct?

s

[Mathieu Leclaire]

What was the resolution of the fluid and how long did it took to render?
Will it render with motion blur?

-Mathieu

[Schoenberger]

> i am not sure i follow... are you suggesting you would get it directly from
> ICE yourself within the shader? if so that means we would have to reserve an
> attribute name correct?

Yes, should be working that way.
If you create an float array, per Object.
I don't know if I would have to copy the array in memory, but that is
something else (without a copy it could be a bit slower)

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Schoenberger]

> What was the resolution of the fluid and how long did it took to
> render?

The fire simulation (baFluidCacheFileA_.mov) had a pretty big resolution.
All detail is simulated, there is no texture on the density, only on
the color.
Rendering was also slow because of the very fine detail in the color.
Last frame: 215x177x136
Render time in small HD: 01m42s on a 4x3.0Ghz

> Will it render with motion blur?

3D fluid motion blur not yet, but is on the list.
Additional to a vector render for post-blur.

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Robert Chapman]

@ Holger  There are no particles. You have the fluid grid. And the fluid grid can only be generated by a fluid simulator.

Perhaps it would be possible that you create an array of floats in ICE and then use that instead of a fluid grid, but that's something someone else has to figure out. Then I can support it.

interesting! so if we did create our array of floats we could, in theory, populate it with data from ICE particles!? or 'advect' our own into this...voxel buffer format? that connects to your fluid shader.

and seeing something like this visually, it would be something perfect for ICE to do no?

http://webstaff.itn.liu.se/~jonun/web/Teaching/2009-TNCG13/Siggraph08/talks/2306-abstract.pdf

back to 2008 wishful thinking again! :D
 

[Steven Caron]

that linked paper is kinda what i am thinking, using ICE like they are using FELT, but in the end we are converting everything to a ICE array like they are converting to a voxel buffer.

s

[Robert Chapman]

yeah maybe I should have just learned Houdini back in 2008 eh!  ;)  but getting back on topic, I really would rate the ability in ICE to read, visualize, manipulate, write and now also Render with Holgers shader a common cache format like field3D for volumetric stuff like this...in Softimage that is.

[Schoenberger]

Attached an other rendering. Only density used.
Resolution was something with 200 in height, I have attached a second
picture that shows the fluid cells.
Render time 22 seconds, one light.

Another note:
I think fluids are better for stuff like cigarette smoke.
Otherwise particles (moved by a fluid sim) are probably easier to texture.

> FELT
Hmm, the FELT paper shows particles which are moved by vector fields.
And in the end you render the particles. You do not render the vector field.
It is the same as emFluid did all the time. It moves particles with a
vector field.
You should be able to do that example easily in ICE. No need to cache
that artist field into a grid:
Create a fluid sim with emfluid as base. (not required, but they had it)
Add a turbulence force.
Add a directional force with falloff (based on distance to a null,
requires some simple math)

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

[Steven Caron]

yes, i know emFluid advected particles. but you are incorrect.. they ARE rendering the field. quoted from the document...

"...These particles will ultimately be used to populate the voxel buffer
used in the final render, which is itself a FELT scalar density field."

i am going to reserve any further feedback until i can get my hands on the right emFluid and the shader. then we can talk in more detail about how users want to interact with these tools to produce effects.

also, those images you posted are great!

s
 

[Robert Chapman]

looking good Holger! only 22 seconds to render, much faster than I was expecting, so the quality of this is dependent of the fluid box resolution? but you are smoothing it afterward in the shader?

[Stefan Kubicek]

...And now Hair Farm:
http://www.cebasstation.com/index.php?pid=news_next&nid=421

[Ciaran Moloney]

Looks great! Nice work so far.

[Byungchul Kang]

Hello Schoenberger

Please link address again.

I can't download.

 

Kang

 

------ Original Message ------

Date: Tuesday, Mar 15, 2011 04:10:00 AM
From: Schoenberger <X...@digidragon.de>
To: <soft...@listproc.autodesk.com>
Subject: Re: Fluid shader Alpha


Doh'

The links:
www.BinaryAlchemy.de/download/110314__baVolume_v4alpha
0.01___Softimage_2010.zip
www.BinaryAlchemy.de/download/110314__baVolume_v4alpha
0.01___Softimage_2011.zip

Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

Quoting Schoenberger :

> Hi
>
> I have an alpha version of the fluid shader to play with.
> No samples, no help, not all features finished, but it renders.
>
> You can render:
> - Latest alpha of emFluid
> (more features in cooperation with Eric planed)
> - Maya .mc cache files (you need to add a geometry shader to the scene)
> - (not implemented yet: Phoenix cache files)
> - (not implemented yet: Field3D cache files)

> - (no response yet: FumeFX... so perhaps no implementation)
>
>

> How it works:
> Plug the fluid shader into a picker and the picker into density of
> the volume.
> Plug the fluid shader into the color of the volume shader if wanted.
>
> Use either fluid_value_color or fluid_value_float to get a value of
> the fluid.
> Plug these shaders directly into the fluid shader.
> Or use a gradient or fcurve inbetween.
>
>
>

> Holger Schoenberger
> technical director
> The day has 24 hours, if that does not suffice, I will take the night

VFX Technical Artist     Byungchul  Kang

COREENT from South Korea

[BLOG]  http://cgndev.com

[Vimeo]  http://vimeo.com/cgndev

[TEL]     +82-2-561-5335

[E.Mail] cgn...@nate.com

[Robert Chapman]

Hi Kang

you just have to copy the entire line into the browser - there is a space in the url which some mail readers are using as a hard return - so again

http://www.binaryalchemy.de/download/110314__baVolume_v4alpha 0.01___Softimage_2011.zip
http://www.binaryalchemy.de/download/110314__baVolume_v4alpha 0.01___Softimage_2010.zip

                                                                                                   ^ space is here   

[Byungchul Kang]

 Hi Robert

Yep! Thanks!

 

Kang

 

------ Original Message ------

Date: Tuesday, Mar 15, 2011 10:03:14 AM
From: Robert Chapman <tekan...@gmail.com>
To: <soft...@listproc.autodesk.com>
Subject: Re: Re: Fluid shader Alpha

VFX Technical Artist     Byungchul  Kang

COREENT from South Korea

[BLOG]  http://cgndev.com

[Vimeo]  http://vimeo.com/cgndev

[TEL]     +82-2-561-5335

[E.Mail] cgn...@nate.com

[Schoenberger]

"...These particles will ultimately be used to populate the voxel buffer
used in the final render, which is itself a FELT scalar density field."

 

Hmm, maybe ok. I have only read that they are using a "voxel renderer" to render the particles.

They create a voxel grid (no matter if the format is FELT or not)  for rendering and fill that with the particle density.

But the paper itself is about using velocity to drive particles.

And you cannot use that at render stage.

 

Having access to voxels for rendering is of course something different, there you are right.

 

 

>also, those images you posted are great!

It's mostly the sim.

Therfore I will test more with sims (exspecially the texture coordinate thing) to have more hints for a good look in the help files.

 

 

Holger Schönberger

technical director
The day has 24 hours, if that does not suffice, I will take the night

 

 

 

---

From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Steven Caron
Sent: Tuesday, March 15, 2011 12:09 AM

To: soft...@listproc.autodesk.com
Subject: Re: Fluid shader Alpha

[Schoenberger]

>much faster than I was expecting

The speed can be compared to very dense untextured particles. (perhaps bigger particles)
But as you usally texture a lot of smaller (small for obstacles) your particles, your render gets very slow.

 

> so the quality of this is dependent of the fluid box resolution?

Yes, mostly. As you can see in the cell image.

I have not done enough tests with the texture support, but so far it has a very uneven resolution/distribution.

As you can see in the video.

The texture is a colored honeycomb with black seperators between the combs.

So far I do not see honeycombs...

 

>but you are smoothing it afterward in the shader?

The cells have to be interpolated. Compare it to a 50x50 pixel image that you blow up to 300x300.

The shader supports

  -None (nearest neighbour)

  -Linear

  -Cosine

  -Cubic

  -Cubic Fast

  -Hermite

  -Hermite Fast

But I think you will always use Cubic Fast, otherwise you see the cells.

And so far there is no visual difference between cubic and cubic fast.

Cubic took 28s, Cubic fast 22s (my own invention :-).

 

I will try to change the contrast of the hermite to get a better result, perhaps then it is useful.

At the moment it sharpens to much.

 

 

Holger Schönberger

technical director
The day has 24 hours, if that does not suffice, I will take the night

 

 

 

---

From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Robert Chapman
Sent: Tuesday, March 15, 2011 12:13 AM

To: soft...@listproc.autodesk.com
Subject: Re: Fluid shader Alpha

[Byungchul Kang]

Good result!!

I did download and try to render.. but i can't see volume! my god....

 

please show your node if has anyone success~ oh please.

 

 Kang

------ Original Message ------

Date: Tuesday, Mar 15, 2011 08:02:57 AM
From: Schoenberger <X...@digidragon.de>

To: <soft...@listproc.autodesk.com>
Subject: Re: Fluid shader Alpha

Attached an other rendering. Only density used.
Resolution was something with 200 in height, I have attached a second
picture that shows the fluid cells.
Render time 22 seconds, one light.

Another note:
I think fluids are better for stuff like cigarette smoke.
Otherwise particles (moved by a fluid sim) are probably easier to texture.



> FELT
Hmm, the FELT paper shows particles which are moved by vector fields.
And in the end you render the particles. You do not render the vector field.
It is the same as emFluid did all the time. It moves particles with a
vector field.
You should be able to do that example easily in ICE. No need to cache
that artist field into a grid:
Create a fluid sim with emfluid as base. (not required, but they had it)
Add a turbulence force.
Add a directional force with falloff (based on distance to a null,
requires some simple math)



Holger Schoenberger
technical director
The day has 24 hours, if that does not suffice, I will take the night

Quoting Steven Caron :

> that linked paper is kinda what i am thinking, using ICE like they are using
> FELT, but in the end we are converting everything to a ICE array like they
> are converting to a voxel buffer.
>
> s
>
> On Mon, Mar 14, 2011 at 3:29 PM, Robert Chapman wrote:
>
>> @ Holger There are no particles. You have the fluid grid. And the fluid
>> grid can only be generated by a fluid simulator.
>>
>>
>> Perhaps it would be possible that you create an array of floats in ICE and
>> then use that instead of a fluid grid, but that's something someone else has
>> to figure out. Then I can support it.
>>
>>
>>
>>
>> interesting! so if we did create our array of floats we could, in theory,
>> populate it with data from ICE particles!? or 'advect' our own into
>> this...voxel buffer format? that connects to your fluid shader.
>>
>> and seeing something like this visually, it would be something perfect for
>> ICE to do no?
>>
>>
>> http://webstaff.itn.liu.se/~jonun/web/Teaching/2009-TNCG13/Siggraph08/talks/2306-abstract.pdf
>>
>> back to 2008 wishful thinking again! :D
>>
>>
>>>
>>>
>

VFX Technical Artist     Byungchul  Kang

COREENT from South Korea

[BLOG]  http://cgndev.com

[Vimeo]  http://vimeo.com/cgndev

[TEL]     +82-2-561-5335

[E.Mail] cgn...@nate.com

[Schoenberger]

>but i can't see volume! my god....

The current version does only work with maya cache files.

You have to apply the geometry shader  fluid_loader.

Then to the geometry object the particle_volume, the fluid shader and at last the particle_value_float shader.

 

 

Holger Schönberger

technical director
The day has 24 hours, if that does not suffice, I will take the night

 

 

 

---

From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Byungchul Kang
Sent: Tuesday, March 15, 2011 2:58 AM
To: soft...@listproc.autodesk.com
Subject: Re:Re: Fluid shader Alpha

[Mike Donovan]

Extremely promising ... cant wait to get my hands on this for some testing.
________________________________________
From: softimag...@listproc.autodesk.com [softimag...@listproc.autodesk.com] on behalf of Schoenberger [X...@digidragon.de]
Sent: Monday, March 14, 2011 7:02 PM

To: soft...@listproc.autodesk.com
Subject: Re: Fluid shader Alpha

Attached an other rendering. Only density used.

Quoting Steven Caron <car...@gmail.com>:

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________

________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________

[Morten Bartholdy]

Holger This is fantastic news! You guys rock :)

MB

> Hi
>
> I have an alpha version of the fluid shader to play with.
> No samples, no help, not all features finished, but it renders.
>
> You can render:
> - Latest alpha of emFluid
> (more features in cooperation with Eric planed)
> - Maya .mc cache files (you need to add a geometry shader to the scene)
> - (not implemented yet: Phoenix cache files)
> - (not implemented yet: Field3D cache files)
> - (no response yet: FumeFX... so perhaps no implementation)
>
>
> How it works:
> Plug the fluid shader into a picker and the picker into density of the
> volume.
> Plug the fluid shader into the color of the volume shader if wanted.
>
> Use either fluid_value_color or fluid_value_float to get a value of the
> fluid.
> Plug these shaders directly into the fluid shader.
> Or use a gradient or fcurve inbetween.
>
>
>

> Holger Schoenberger
> technical director
> The day has 24 hours, if that does not suffice, I will take the night
>
>
>
>
>

--
Best Regards

Morten Bartholdy
3D/VFX Supervisor

[Morten Bartholdy]

Beautiful smoke plume Holger.

MB