PDA

View Full Version : FXWars! NUKE!: Eric Axelson, Freeform


Axelson
02-16-2010, 09:29 PM
This is my first fx challenge.

I am mainly doing this to get a better understanding och maya fluids.
I have tested some things out for the nuke, much work needed but i feel its going in the right direction and im starting to understand how to control it, wich attributes that u can adjust to go forward and not backwards.

Axelson
02-17-2010, 06:44 PM
The biggest issue that i have right now is to get the nuke to slowdown once the blast has expanded as i want to, it rises to quickly and hits the container walls.

Is the only trick here to key the sim rate or should i also scale up the container?
the container res is now 100 x 150 x 100 wich makes it pretty heavy on my old computer.

i will post a playblast tonight.

Katachi
02-17-2010, 06:58 PM
The biggest issue that i have right now is to get the nuke to slowdown once the blast has expanded as i want to, it rises to quickly and hits the container walls.

Is the only trick here to key the sim rate or should i also scale up the container?
the container res is now 100 x 150 x 100 wich makes it pretty heavy on my old computer.

i will post a playblast tonight.

Hi,

I donīt own Maya fluids, but isnīt there something like a physical scale option? So where you can define the real physical scale of the simulation (as in respective real world size of the fluid container dimensions or voxel dimensions)? Not sure what "sim rate" is but if it is the time step of the simulation you could probably key the time step at a certain point. Should work quite fine. But if you have the chance for setting the physical scale that would of course be much better.

Axelson
02-18-2010, 12:48 AM
What im reffering to is simulation rate scale, this seems to be able to slow down the simulation. not really sure how it works yet, if its maintaining buoancy and just slowing it down or how it works.

did a test with it anyway. started out on with a sim rate scale of .5 then after frame 50 to 65 it goes down to .25

what im missing is the sucking effect, like in the center basepart of the cloudshape it should suck the smoke up, this causes the swirl effect from the side as from what i can understand. not really sure how to speed up that part and still maintaining the slow movement of the big cloud part.

more tests to come, first playblast of what i got:

http://vimeo.com/9537708

mustan9
02-18-2010, 02:34 PM
what im missing is the sucking effect, like in the center basepart of the cloudshape it should suck the smoke up, this causes the swirl effect from the side as from what i can understand. not really sure how to speed up that part and still maintaining the slow movement of the big cloud part.

I think you need to create a second source for fuel. Like a thin cylinder to create the stem for the mushroom cloud. That will add heat to the center of the cloud shape causing it to turn inside out. Let this second fuel source burn for a while and then die out.

This is what I've seen other people do, but I haven't tried it myself.

Aikiman
02-18-2010, 10:04 PM
what im missing is the sucking effect, like in the center basepart of the cloudshape it should suck the smoke up, this causes the swirl effect from the side as from what i can understand. not really sure how to speed up that part and still maintaining the slow movement of the big cloud part.


thats the difficulty using Maya fluids, seems to be much more easily achievable using FumeFX, but with Maya you really have to know your stuff. If I knew the answer Id help you out.

Axelson
03-03-2010, 12:04 PM
Well im very close to giving up on this one now, doesnt matter wich direction i drag it in it still want go where i want to.

For example: using a sim rate of .5 the speed is pretty much okay but the shape wont hold up fore more than say 70 frames, then the shape will go nuts.

On the other hand, usign a simrate of 2 the shape is nice all the way until it collides with the top of the container, the speed thou is way to high and needs to be scaled down in trax.
Another problem that i have noticed is that i cant really seem to get the wispy feeling, i want those details very small since it should be a very large explosion.

The main shape should be the mushrom and then small details with wispy smoke. trying to acheive this without success. i tried to crank up the temperature turbulence and that helped slightly, will test with more later. The regular turbulence tab seems very dangerous to touch, put it on 0.1 and it all goes nuts. =/

lets see if i can do something on it now.

mustan9
03-03-2010, 01:29 PM
Have you thought of rendering it at a higher frame rate, rather then trying to get the sim to slow down. I plan to render mine at 60/fps and play it back at 24/fps.

You might have to increase your subframe sampling for the simulation if you make the sim rate that low.

Axelson
03-03-2010, 04:26 PM
I have now started over with a new setup. simrate is @ 2 and i have keyed bouyancy fraom 1 to around 0.1 a couple of frames after it starts emitting.
Still cant get that nice wispy smoke around the edges, have a look at the playblast here the explosion in the starts need pretty much work too, its getting sucked up pretty fast and i guess the way to fix that is to use another emitter for the shockwave.

Here is the playblast: http://www.vimeo.com/9887216

Any tips and ideas to make it more turbulent at the top? im thinking that the regular turbulence that is on the fluid container seems very unstable. as soon as i up it more then 0.02 it goes all over the place, i have tested different frequencies to get smaller turbulence but no luck. is the frequence depending on the container size?
My container is only about 20 high and 10 in x and z, is there any difference if the container is say 10 10 10 and resolution is 20 20 20 or the container is 30 30 30 and res is 60 60 60 ? of course it will take longer to simulate but except for that?

mustan9
03-03-2010, 05:33 PM
The turbulence is dependent upon the container resolution. You need a high resolution to get small details into the mushroom cloud and make it look less smooth.

I'm not sure you should animate the buoyancy because lower buoyancy will flatten out the simulation. Causing the cloud to spread apart more.

I think someone here (vik?) posted a Maya scene file as an example somewhere.

stooch
03-03-2010, 07:36 PM
keep your sim rate low not high. even the goodlooking sims so far on this challenge look too fast to me. nukes are typically 1-2 km in diameter.... think about that for a second.

i suggest you use sub frame stepping, when you cache, go into your frame step size in options and make it something like .5 then cache that. your sim will be twice as slow and it will also have half the chance of freaking out.

also generally, if you are using force fields, you should use them in "blips" ie, turn it on and quickly turn it off, you will notice that the velocities they impart will stay as they are (assuming you have low damping).

also your dissipation and dissolve must be very, very low. to "suck in", use a radial field... again, blip the intensity, if you use velocity draw you can easily see how the directional arrows remain affected in the direction of the radial field even after its back to 0.

this is a very important concept because if you just keep pumping fields into the sim they will cause it to become unstable.

after testing, i found that putting a number of fields into my fluid box and then blipping their intensities at certain intervals and sequence, gives me the most controllable fluid sime (although in the end they ended up inadequate for me).

andybyrne462
03-03-2010, 11:01 PM
keep your sim rate low not high. even the goodlooking sims so far on this challenge look too fast to me. nukes are typically 1-2 km in diameter.... think about that for a second.

i suggest you use sub frame stepping, when you cache, go into your frame step size in options and make it something like .5 then cache that. your sim will be twice as slow and it will also have half the chance of freaking out.

also generally, if you are using force fields, you should use them in "blips" ie, turn it on and quickly turn it off, you will notice that the velocities they impart will stay as they are (assuming you have low damping).

also your dissipation and dissolve must be very, very low. to "suck in", use a radial field... again, blip the intensity, if you use velocity draw you can easily see how the directional arrows remain affected in the direction of the radial field even after its back to 0.

this is a very important concept because if you just keep pumping fields into the sim they will cause it to become unstable.

after testing, i found that putting a number of fields into my fluid box and then blipping their intensities at certain intervals and sequence, gives me the most controllable fluid sime (although in the end they ended up inadequate for me).



this is good advice.

I'm sticking with volumetrics for the most part in my version, while only relying on fluids for subtle effects. Mainly because volumetrics are easier to direct...especially for something that SHOULD barely be moving. I just hope I can finish in time :eek:

Axelson
03-04-2010, 06:58 AM
Thank you for the input about the cache, do you mean that i should put the "evaluate every frame to 0.5 AND then scale the cache in trax? cause just setting the evaluate every frame to 0.5 doesnt change anything in speed.

Or am i messing with the wrong attribute?

http://vimeo.com/9905326

this is the sim i put on last night. with a very low bouyancy and sim rate at 2.0
gonna to another one now with standard bouyance and low simrate as you said.

stooch
03-04-2010, 08:55 AM
yeah that last sim you did is a tad too fast. i think its dissipation is too fast and there is too much boyancy. also you might want to keep that emitter going for a while, it needs to keep feeding the density, otherwise it will thin out.

i would release a long pulse of density and a short burst of heat and or fuel.

then i would let the heat dissipate. Fuel is nice beacuse you can tune it to keep adding heat to your sim over a longer period, long after your heat emission stopped, and because it adds some really nice swirls and details. you just have to keep things SLOOOWWW... slow reaction rate, low boyancy on density (most of the boyancy should be from heat). Slow heat dissipation too, its a nuke... so there is alot of heat to dissipate.

then simulate it at .25 frame step. save every evaluation and simulate on 1/4 of your timeline. That would effectively give you 1/4 speed. which should put you closer to where you need to be. (might even want to entertain .2 frame step for 1/5th speed)

Axelson
03-04-2010, 11:04 AM
Thank you very much stooch! you have helped very much! Its coming along a lot better now.
still have a lot of tweaking to do but now im atelast workign in the right direction.

PLayblast in an hour or something. =)

Axelson
03-04-2010, 12:04 PM
here is the lates playblast.

Much better in my opinion.
what do you guys think? i havent touched the shading part yet. want to nail the simulation part first.

http://www.vimeo.com/9908880

stooch
03-04-2010, 11:59 PM
better. you might want to dissipate the heat sooner, there has to be a difference between down velocities and/or drag and upward velocities. only when you have a delta between the two, will swirling convection currents will become apparent. try adding a slightly negative boyancy on density too, so that it sinks down a little bit and enhance any swirling, in addition to reinforcing the stem part (which is looking a little thin imo).

Axelson
03-05-2010, 11:20 PM
latest test.

did som short test in comp to just for testing how the color and stuff will be able to change.

http://vimeo.com/9938135

Aikiman
03-06-2010, 12:02 AM
latest test.

did som short test in comp to just for testing how the color and stuff will be able to change.

http://vimeo.com/9938135

Thats a private video there. I agree with Stooch your heat needs dissipating whether you achieve this with fuel or temperature settings. I prefer that method which is dynamic, others prefer keyframing the incandescent ramp to kill the colour, either way works. Im not sure I would do it in post though, might be a little differcult. The idea of keyframing the incandescent is an interesting one because the heat and temperature is still there in your explosion and hasnt burnt out as it would be in the dynamic method.

Axelson
03-06-2010, 03:08 PM
I tried setting a negative bouyancy for density, but the fluid just stayed at the bottom then. even thou if i raised the heat bouyancy.

Here is the video that i last posted, now when im watching it it feels like i have been going in a little wrong direction. this is probably because i have tried to get the small details in.

anyway, here it is http://vimeo.com/9938135

more to come.

stooch
03-06-2010, 09:13 PM
I tried setting a negative bouyancy for density, but the fluid just stayed at the bottom then. even thou if i raised the heat bouyancy.

welll... the obvious answer to that is that you used too much negative boyancy and not enough heat...

basically, if you have negative boyancy, as soon as there is 0 heat in a cell, it will start adding a negative vector. so you have to have a slow enough heat dissipation that allow the dissipating heat to sustain the fluid for a while before the negative boyancy wins out and start pulling the values down into the next cell.

anyway your new sim is looking nice. wondering if the scale still feels a little small.

Axelson
03-07-2010, 09:00 AM
Thanks for the ideas. if im remembering correct the settings used for dissipating the heat was superlow when i tested with neg boyuancy.

Density bouyancy was at -0.3
Heat bouyancy was at 3
Heat dissipation was at 0.05

with those settings i couldnt get it to lift at all.
As i said before im gonna go a little back and start on over on one version from a couple of days ago.
I think it holds the mushroomshape much better.

Updates later this evening.

stooch
03-07-2010, 01:42 PM
Density bouyancy was at -0.3
Heat bouyancy was at 3


...yeah that would cancel each other in one frame. depending on the reaction speed.

WeylandYutani
03-07-2010, 01:57 PM
Thank you for the input about the cache, do you mean that i should put the "evaluate every frame to 0.5 AND then scale the cache in trax? cause just setting the evaluate every frame to 0.5 doesnt change anything in speed.

Or am i messing with the wrong attribute?

http://vimeo.com/9905326

this is the sim i put on last night. with a very low bouyancy and sim rate at 2.0
gonna to another one now with standard bouyance and low simrate as you said.

All you have to do is cache the simulation with 4 or 5 substeps per frame. Once you have the cache done, scale it up in Trax 4 or 5 times. This will effectively increase the simulation time as many times as you scale the cache in Trax. Just make sure to not over scale the time more than the substeps per frame you have in your cache file. If you do, you will get weird noise in the incandescence texture.

stooch
03-07-2010, 08:48 PM
All you have to do is cache the simulation with 4 or 5 substeps per frame. Once you have the cache done, scale it up in Trax 4 or 5 times. This will effectively increase the simulation time as many times as you scale the cache in Trax. Just make sure to not over scale the time more than the substeps per frame you have in your cache file. If you do, you will get weird noise in the incandescence texture.

I wouldnt over sample the cache. that would make for a HUUUUGE cache file.

I would oversample the sim and save 1:1 frames into cache, try not to use trax for stuff like this.

Axelson
03-08-2010, 06:21 PM
This is what i got right now.

http://vimeo.com/10007328


The stem part in the center looked better before i upped the res of the container. i think its going in the right direction atleast. on this one i have a turbulence field as a volume sphere at the top. with a magnitude of 15 to break it up a little.

Comments are as always helpful and i must thank you guys who have helped go in the right direction!

Aikiman
03-08-2010, 06:48 PM
Eric,

its probably still too fast for a nuke but apart from that it makes a great explosion, nice balance and timing and forms well in shape. Nice work.

Axelson
03-08-2010, 08:09 PM
Thank you for the input. i will see if i can take it down a notch in speed and still maintain the the good detail. need to start on the shockwaves soon, someone whispered deadline =)

Axelson
03-10-2010, 11:24 PM
Here is the next step. i just need to put some damping on it so it doesnt destroy the shape in the end as it does now.

Video: http://vimeo.com/10075096

There is nothing fancy going on here really. Just starting to get friendly with fluids.
For you guys out there that want some info.

here goes:

container size: 40 60 40
container res 100 150 100

Sim rate scale 0.8
High detail solve on all grids exept for velocity.

Buoyancy is at 0.4

Velocity scale is upped by 1.3 on both x y z

Temperature has the turbulence, nothing at the regular turbulence tab.

This is the most important things i guess for getting something similar.
Sorry for the bad updates but my computer is slower then watching paint dry.

CGTalk Moderation
03-10-2010, 11:24 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.