Pyro test

[jacys]

there is a test about pyro with DOPs.

first version then trying make it better.

Pyro_test01.zip

Edited April 23, 2009 by jacys

[jacys]

one thing i was curious.

I turn pyro to uprez, but i didnt see any thing.

where i done wrong?

i use lowrez it's ok.

[old school]

You have to cache the low res volumes to disk. If you inspect the high res DOPs they reference the File SOP in the low res object. The file cache options is in the render SOP that fetches in all the volumes.

Edited April 24, 2009 by old school

[jacys]

You have to cache the low res volumes to disk. If you inspect the high res DOPs they reference the File SOP in the low res object. The file cache options is in the render SOP that fetches in all the volumes.

so, I have to create a ROP Output Driver to cache it ?

or other way?

[jacys]

You have to cache the low res volumes to disk. If you inspect the high res DOPs they reference the File SOP in the low res object. The file cache options is in the render SOP that fetches in all the volumes.

http://vimeo.com/4313485

new version here, render for almost 10 hrs.

Edited April 26, 2009 by jacys

[stevenong]

That looks pretty good. What is the upres divisions? Also, did you use Raytrace Shadows or Deep Shadow maps?

Did you turn Scattering on in the Pyro shader? To speed up your renders, you can either turn Scattering off, which will change the look, or change the Atten Method to "Approximated, No Shadows".

Cheers!

steven

[jacys]

That looks pretty good. What is the upres divisions? Also, did you use Raytrace Shadows or Deep Shadow maps?

Did you turn Scattering on in the Pyro shader? To speed up your renders, you can either turn Scattering off, which will change the look, or change the Atten Method to "Approximated, No Shadows".

Cheers!

steven

I use Deep Shadow maps and turn Scattering off.

up-res(uprez, we also say that hi rez in other software.) is another resolution in Pyro, but you have to simulate low-rez and cache it for up-rez pyro first.

[Jason]

I use Deep Shadow maps and turn Scattering off.

up-res(uprez, we also say that hi rez in other software.) is another resolution in Pyro, but you have to simulate low-rez and cache it for up-rez pyro first.

Depending on the scale of your scene, the "Volume Step Size" parameter on the Mantra ROP can have a significant effect. Basically, you can find the sweet spot where you use the largest step size possible before your quality degrades. There is a spot where a smaller step size doesn't improve the quality too much but really starts to slow stuff down.

Unfortunately Mantra gives no hints about whether you are undersampling or oversampling or volume too much as it does with Displacement Bounds -- (which would be awesome, SESI hint! ) - so you have to find it yourself by trading off quality for speed.

[jacys]

Depending on the scale of your scene, the "Volume Step Size" parameter on the Mantra ROP can have a significant effect. Basically, you can find the sweet spot where you use the largest step size possible before your quality degrades. There is a spot where a smaller step size doesn't improve the quality too much but really starts to slow stuff down.

Unfortunately Mantra gives no hints about whether you are undersampling or oversampling or volume too much as it does with Displacement Bounds -- (which would be awesome, SESI hint! ) - so you have to find it yourself by trading off quality for speed.

Thanks for this great class here!

I will try rocket launch next time!

I expect it will save more 4 hrs simulation.

[jacys]

Thanks for this great class here!

I will try rocket launch next time!

I expect it will save more 4 hrs simulation.

http://vimeo.com/4334933

I try to add a force to simulate the rocket launch.

It simulates about 4 hrs. Nice quality and cost.

Trying to do more test then make a demo reel.

Edited April 26, 2009 by jacys

[Raymond Chua]

http://vimeo.com/4334933

I try to add a force to simulate the rocket launch.

It simulates about 4 hrs. Nice quality and cost.

Trying to do more test then make a demo reel.

Wow, looks good but seems like a bit expensive. Is there a way to shorten time to simulate and render?

[jacys]

Wow, looks good but seems like a bit expensive. Is there a way to shorten time to simulate and render?

Buy a new computer.................

___

btw,

I will try exploding ball next time.

Edited April 27, 2009 by jacys

[dbukovec]

You have the same problem with fire, like, i have just tested the fire tool from the shelf, and main problem was, tht the fire volume has a stationary noise with horizontal holes in it, which looks very ugly. Anyone knows from where that noise layer comes from? or why it is not moving with fluid?

[Fomal]

You have the same problem with fire, like, i have just tested the fire tool from the shelf, and main problem was, tht the fire volume has a stationary noise with horizontal holes in it, which looks very ugly. Anyone knows from where that noise layer comes from? or why it is not moving with fluid?

I Agree, this is a bit awkward. The source noise has been added to distriute the fuel a bit more random but does not move with the source since it doesn't have the rest attribute associated with the build field. There might be an easy way out of this since the rest attribute is available in the add_noise vop (inside the pyro solver) Problem is that this tracks the contour, not the source field.

For me, the best way is to use a seperate front build in sops to distribute the fields and update the solver. I'm currently looking into that, This way your feeding it a predefined source that corresponds with the emitter but doesn't get created in dops (by building the relationships). I'm using the sources uv space to get the right noise distribution.

Arg, work in progress!

Edited May 20, 2009 by Fomal

[Fomal]

For anyone interested, I've quickly cleaned up the exploding sphere scene. Used a seperate pyro sim to create a quick explosion, advected some particles based on that velocity and used that to shatter a sphere. Next I filled the sphere with all the necessary fields, so as the shards fly off there is an explosion in the beginning. Slowly fading out. I wanted to post the bgeos as well but the file becomes rather large. Hopefully it gives a good idea on how I did it.

shatter_exp_sphere.hipnc

[dbukovec]

Thanks for the scene file!

[wolf_cub_one]

I Agree, this is a bit awkward. The source noise has been added to distriute the fuel a bit more random but does not move with the source since it doesn't have the rest attribute associated with the build field. There might be an easy way out of this since the rest attribute is available in the add_noise vop (inside the pyro solver) Problem is that this tracks the contour, not the source field.

For me, the best way is to use a seperate front build in sops to distribute the fields and update the solver. I'm currently looking into that, This way your feeding it a predefined source that corresponds with the emitter but doesn't get created in dops (by building the relationships). I'm using the sources uv space to get the right noise distribution.

Arg, work in progress!

Fomal,

Have you made any progress in regards to the noise of the flame not moving? I separated the fields so that I have one for density and one for heat. I am able to fade the smoke, but when I merge back the smoke and flame, the noise of the flame is not moving. If I don't separate the fields, I can still fade my smoke but it also fades my flames with the flame noise moving correctly. Just wondering how you were able to bypass the issue. Any hints would be appreciated.

Thanks

[wolf_cub_one]

Nevermind. Figured it out. Really simple. Since I separated them into different fields, I just had to merge back all the other fields that I took away.