Can this effect be replicated in LW (Gravity mist particle fx on solar panels)

[Planeguy]

And this is probably the only tutorial (Houdini) that did an amazing job in doing this effect. Probably the best one out there. This is what I'd like to achieve.

And this one, I believe was done with turbulencefd but not sure what package was it rendered on, the guy didn't share any info but when I looked turbulencefd on youtube, this anim showed up, so I assume it was done with it.

 

[prometheus]

Yes, that houdini tutorial is advanced, and unless you intend to follow it and doing it in houdini, I think it´s not that helpful to go in to detail study of it if you use something else than houdini, you would have to wrestle the platform differences a lot.

I noticed they used velocity vectors a lot, and also, the fluids are based on the particle velocity, so the motion derives from particles, and not from fluid simulation, mainly, but rendered with the fluid engine.
This is doable in both lightwave gas solver and in blender.

Gas solver actually has something that blender doesn´t, which is to drive the fluids with the particle velocity force ( which isn´t the same as emitting fluids per particle velocity)
As seen here (not the same type of simulation though)

As for the starship atmospheric reentry, seems a bit simpler, with fluids the thing is that you could parent a force pushing against the object shield, so it follows the rotation of the object no matter how it rotates the generated heat collision will look good I thin, that is the pushing force, then you need the collision force around the object to work so it pushes the heat strokes a bit disturbed to the sides.


Also as demonstrated in the Houdini vid, a lot of separate passes and treating various elements individually and combining it in the end.

 

[Planeguy]

Yes, that houdini tutorial is advanced, and unless you intend to follow it and doing it in houdini, I think it´s not that helpful to go in to detail study of it if you use something else than houdini, you would have to wrestle the platform differences a lot.

I noticed they used velocity vectors a lot, and also, the fluids are based on the particle velocity, so the motion derives from particles, and not from fluid simulation, mainly, but rendered with the fluid engine.
This is doable in both lightwave gas solver and in blender.

Gas solver actually has something that blender doesn´t, which is to drive the fluids with the particle velocity force ( which isn´t the same as emitting fluids per particle velocity)
As seen here (not the same type of simulation though)

As for the starship atmospheric reentry, seems a bit simpler, with fluids the thing is that you could parent a force pushing against the object shield, so it follows the rotation of the object no matter how it rotates the generated heat collision will look good I thin, that is the pushing force, then you need the collision force around the object to work so it pushes the heat strokes a bit disturbed to the sides.


Also as demonstrated in the Houdini vid, a lot of separate passes and treating various elements individually and combining it in the end.

Good observations! I saw a video where the vfx company that did the effect explained how they did the plasma effect, and if I remember well, they talked about something similar to what you said. I'll see if I can find the video and post it here.

And yeah, I don't plan to learn Houdini for this effect. Maybe in the future for general purpose, but I'm more comfortable with LW or 3dsmax at the moment.

So, what would be a basic node approach to do the base shape of the effect in LW using the gas solver?

 

[Planeguy]

Found it:

(if the video doesn't start at the time mark, play it at the 1:01 mark) but it's a short vid with lots of interesting info. Let me know your thoughts on the process and how it would compare to LW based on what you said above.

 

[Planeguy]

Here's this from the interview:

"...We've used layers of simulations, particle simulations, and pressure simulations that we've calculated based on the velocity of the spacecraft as it comes in to work out the points of high pressure to generate procedural volumes of geometry and particle stream sparks, heat and ablation, up to 28 passes each of which have got multiple passes within them for each of these shots throughout the re-entry sequence..."

You can click on CC on the video to read the text. There's lot of interesting info.

But basically, simplifying the the effects for something for lightwave of phoenixfd, this is what I gather I'd need to do:

Layers of sims (unsure how this would be done)
Particle sims (somewhat there)
Pressure sims (have to research this more in phoenixfd or lightwave's gas solver)
Velocity calculations on high pressure points (can this even be done in LW?)
Procedural geometrical volumes (shouldn't be too hard to figure out)

And the rest are render passes, that's the easiest part after figuring out the above.

Thoughts?

 

[prometheus]

Here's this from the interview:

"...We've used layers of simulations, particle simulations, and pressure simulations that we've calculated based on the velocity of the spacecraft as it comes in to work out the points of high pressure to generate procedural volumes of geometry and particle stream sparks, heat and ablation, up to 28 passes each of which have got multiple passes within them for each of these shots throughout the re-entry sequence..."

You can click on CC on the video to read the text. There's lot of interesting info.

But basically, simplifying the the effects for something for lightwave of phoenixfd, this is what I gather I'd need to do:

Layers of sims (unsure how this would be done)
Particle sims (somewhat there)
Pressure sims (have to research this more in phoenixfd or lightwave's gas solver)
Velocity calculations on high pressure points (can this even be done in LW?)
Procedural geometrical volumes (shouldn't be too hard to figure out)

And the rest are render passes, that's the easiest part after figuring out the above.

Thoughts?

I would warn you that there are tools/techniques in there that I don´t think you will be able to get working inside of Lightwave´s framework.
Procedural geometrical volumes? it´s kind of unnecessary I think, since there are only a few volume types you can use..

1. volumetric items, that one you can exlude, it´s not practical trying to do moving simulations, there are advection nodes to push static volume items, but then you are better of pushing it with the gas solver,

2. Vdb´s, can only be extracted from other 3D simulations (If we exclude cached simulations from the lw gas solver )

3. Gas solver, either directly use the simulated gas (which is build around openVDB) and has to be vdb cached for efficiency anyway.

The velocity force I mentioned to drive lightwave´s gas solver, it´s as far as I know of only using the initial velocity force from a particle emitter, it won´t allow for the fluids to follow the velocity vectors fully, just basic velocity direction force.

Any more advance velocity behavior to import in to lightwave and drive the fluids I think isn´t possible, but I am not quite sure with 2020.
I don´t think I would try and adapt the principles to that level from what they did in houdini, rather go for a simpler solution with getting fluids and forces right, combined with particle simulations that are post processed, If you are to get somewhere within a shorter period of time, otherwise months of researchs and trials.
I have to look through that vid more properly later though, to form a better overview and later views on if it´s doable, practical etc.

 

[prometheus]

Good observations! I saw a video where the vfx company that did the effect explained how they did the plasma effect, and if I remember well, they talked about something similar to what you said. I'll see if I can find the video and post it here.

And yeah, I don't plan to learn Houdini for this effect. Maybe in the future for general purpose, but I'm more comfortable with LW or 3dsmax at the moment.

So, what would be a basic node approach to do the base shape of the effect in LW using the gas solver?

Not sure, I rarely use the gas solver nowadays, and I haven´t worked that much with collision objects.
Before anything, you would need to learn to shut down the uplift fire forces the gas solver by default yields, bouyance, density scale etc and perhaps some others.
Then learn to make the basic force input to go in x-y-z directions based on a null location or mesh item location, by adding vector scale nodes and feed in to the gas solver force inputs, and perhaps try to get a normal force setup as well.

Questionable if you should aim for a normal force that is parented to the solar panel or a pod or similar to match the rotation, or if the force should be equal and quite large in area to simulate a true pressure force on an incoming high velocity object, there are various ways I think.
That´s the directional force, then you got the collision force which could be the actual object itself, or another mesh to serve as an invisible collision object, or several of them, and which settings that needs.

The problem is that the gas solver and the framework that works only in nodal setups, is to tedious for my liking, as well as the actual slow simulation time, as well as the openGL preview which simply sucks, that´s why I have no passion to go in and test it, at least not this week, month..maybe but can´t swear on it, I rather just set it up much faster in blender, but it´s me who can work that Uggly UI, you on the other hand may have to deal with the Gas solver as prefered method, if Phoenix simulations isn´t working.

If Only the LW team could have focused on getting GPU simulation for the Gas solver, and a voxel slicing method for the OpenGL display, that would have ment a lot..while still missing a better framework around the fluid vdb tool.

I am not sure yet wether or not my use of particles as velocity force would work and to use particle collision only instead of actual fluid gas solver collision, I think there may be an issue since using that is only forcing the gas solver to push the fluid based on the initial particle velocity and direction, but it may work..not sure, and in such case all the collision you have to work with is within the actual particle system and legacy collision forces that can use the bounce settings, probability roughness etc...if the fluids then adapt and follows that particle velocity force properly.

 

[prometheus]

More stuff falling through the atmosphere

 

[Planeguy]

Not sure, I rarely use the gas solver nowadays, and I haven´t worked that much with collision objects.
Before anything, you would need to learn to shut down the uplift fire forces the gas solver by default yields, bouyance, density scale etc and perhaps some others.
Then learn to make the basic force input to go in x-y-z directions based on a null location or mesh item location, by adding vector scale nodes and feed in to the gas solver force inputs, and perhaps try to get a normal force setup as well.

Questionable if you should aim for a normal force that is parented to the solar panel or a pod or similar to match the rotation, or if the force should be equal and quite large in area to simulate a true pressure force on an incoming high velocity object, there are various ways I think.
That´s the directional force, then you got the collision force which could be the actual object itself, or another mesh to serve as an invisible collision object, or several of them, and which settings that needs.

The problem is that the gas solver and the framework that works only in nodal setups, is to tedious for my liking, as well as the actual slow simulation time, as well as the openGL preview which simply sucks, that´s why I have no passion to go in and test it, at least not this week, month..maybe but can´t swear on it, I rather just set it up much faster in blender, but it´s me who can work that Uggly UI, you on the other hand may have to deal with the Gas solver as prefered method, if Phoenix simulations isn´t working.

If Only the LW team could have focused on getting GPU simulation for the Gas solver, and a voxel slicing method for the OpenGL display, that would have ment a lot..while still missing a better framework around the fluid vdb tool.

I am not sure yet wether or not my use of particles as velocity force would work and to use particle collision only instead of actual fluid gas solver collision, I think there may be an issue since using that is only forcing the gas solver to push the fluid based on the initial particle velocity and direction, but it may work..not sure, and in such case all the collision you have to work with is within the actual particle system and legacy collision forces that can use the bounce settings, probability roughness etc...if the fluids then adapt and follows that particle velocity force properly.

Indeed, I found the gas solver to be tedious to setup as well, but I'm a beginner with it too. It almost has a mind of its own, doesn't seem to be consistent. I'm gonna try first with phoenix, and then give the gas solver a test on this.

If both fail, then I'll download a trial of turbulenceFD and see if the trial allows export of VDB's so I can export them back to LW. They probably don't allow VDB exports (with a trial/demo version) if I have to guess.

 

[Planeguy]

More stuff falling through the atmosphere

The effect doesn't look as impressive as the one in Gravity, but it does look good and smooth.

 

[prometheus]

Indeed, I found the gas solver to be tedious to setup as well, but I'm a beginner with it too. It almost has a mind of its own, doesn't seem to be consistent. I'm gonna try first with phoenix, and then give the gas solver a test on this.

If both fail, then I'll download a trial of turbulenceFD and see if the trial allows export of VDB's so I can export them back to LW. They probably don't allow VDB exports (with a trial/demo version) if I have to guess.

If TFD doesn´t allow for vdb conversion with the demo trial, it´s their loss probably since less people are proned to test it, though I can understand there may be an issue since I believe it´s a standalone converter, but implementing a demo trial period for both of them in such case could make people more interested in it.

Otherwise embergen has a full demo trial for 30 days, Including the ability to export the vdbs during that time.
I bought embergen in the end of last year for around 170 USD, a campaign price at that time, but before the purchase I was testing it quite a bit and exporting to vdb to make sure I was satisfied with it.

Currently it´s been some test work with cases like ..
Clouds
Fire and smoke
Explosions
Solar flares

I have yet to work with other particles and fluid simulations on my to do list though.
Vectorgen is something I need to read up on, I think it could have some interesting stuff there with particle vectors to get in to embergen as well.


Houdini apprentice let´s you learn houdini forever, and export the vdb´s as well ..if you know how to, and use in your personal projects..but for your stuff to go commercial, indie licens is required.
For me it´s mostly a question about time, wether or not I should invest more on learning houdini, but making some particle animations, mixing with vdbs, or painting in cloud volumes in houdini and exporting to lightwave, no problems now.

 

[prometheus]

The effect doesn't look as impressive as the one in Gravity, but it does look good and smooth.

I agree completely, it´s nice ..but the gravity stuff is better, different objects and reactions though.

 

[Planeguy]

If TFD doesn´t allow for vdb conversion with the demo trial, it´s their loss probably since less people are proned to test it, though I can understand there may be an issue since I believe it´s a standalone converter, but implementing a demo trial period for both of them in such case could make people more interested in it.

Otherwise embergen has a full demo trial for 30 days, Including the ability to export the vdbs during that time.
I bought embergen in the end of last year for around 170 USD, a campaign price at that time, but before the purchase I was testing it quite a bit and exporting to vdb to make sure I was satisfied with it.

Currently it´s been some test work with cases like ..
Clouds
Fire and smoke
Explosions
Solar flares

I have yet to work with other particles and fluid simulations on my to do list though.
Vectorgen is something I need to read up on, I think it could have some interesting stuff there with particle vectors to get in to embergen as well.


Houdini apprentice let´s you learn houdini forever, and export the vdb´s as well ..if you know how to, and use in your personal projects..but for your stuff to go commercial, indie licens is required.
For me it´s mostly a question about time, wether or not I should invest more on learning houdini, but making some particle animations, mixing with vdbs, or painting in cloud volumes in houdini and exporting to lightwave, no problems now.

That was a great price for embergen! I'll report back if I can export vdb's with the demo.

 

[Planeguy]

I agree completely, it´s nice ..but the gravity stuff is better, different objects and reactions though.

There's so much detail in the entry sequence you have to play it several times to appreciate. It's a pity, all the work and time they spend doing that effect and it only lasts a few moments lol.

 

[prometheus]

That was a great price for embergen! I'll report back if I can export vdb's with the demo.

With embergen demo you definitely can, with turbulenceFD I don´t know, just to be clear since you mention embergen in the last comment, and not TFD.

There's so much detail in the entry sequence you have to play it several times to appreciate. It's a pity, all the work and time they spend doing that effect and it only lasts a few moments lol.

720 in resolution isn´t good enough to study it "Properly" though.
I watched Gravity on my 50 inch tv set, should have seen it on the big stage to appreciate it, though there was some stuff that came to my attention..
The first was Sandra Bullocks "moaning" when she was spinning around..I am sure it was intended to sound as she was in panic, but to my sick mind that sounded like something else
Reminds me of a media project recording a movie, which I made with some younger ladies, I was playing a serial killer, trying to strangle one of them, and we added the panic sounds in a studio after the initial shooting, and one of the girls laughed because it sounded like something else than panic..haha.
As far as education and shooling in film production/editing, the course was crap..and shouldn´t have been in the program, but we had a bit of fun doing it.

Then you have the reality VS good looking movies..and the Chris Hadfield debunk, I think he thought the visuals were great, but the physics was a tiny bit off
though it wasn´t as bad as armageddon..

Timestamp starting from the Gravity coverage..

Chris Hadfield quote: uhh...

 

[Planeguy]

With embergen demo you definitely can, with turbulenceFD I don´t know, just to be clear since you mention embergen in the last comment, and not TFD.



720 in resolution isn´t good enough to study it "Properly" though.
I watched Gravity on my 50 inch tv set, should have seen it on the big stage to appreciate it, though there was some stuff that came to my attention..
The first was Sandra Bullocks "moaning" when she was spinning around..I am sure it was intended to sound as she was in panic, but to my sick mind that sounded like something else
Reminds me of a media project recording a movie, which I made with some younger ladies, I was playing a serial killer, trying to strangle one of them, and we added the panic sounds in a studio after the initial shooting, and one of the girls laughed because it sounded like something else than panic..haha.
As far as education and shooling in film production/editing, the course was crap..and shouldn´t have been in the program, but we had a bit of fun doing it.

Then you have the reality VS good looking movies..and the Chris Hadfield debunk, I think he thought the visuals were great, but the physics was a tiny bit off
though it wasn´t as bad as armageddon..

Timestamp starting from the Gravity coverage..

I once tried embergen demo. Need to download the latest version and see if it would allow me to test it, or not. Some programs don't allow you to do that. Once the trial expires, it's over. You buy it or leave it lol.

Yeah, there were several inconsistencies with Gravity and this is coming from a regular dude who knows little about physics! I think they over did a lot of stuff that made no sense. But, visually speaking, it's very good, but I haven't seen it on a big tv so not sure how the flames would have looked with more resolution. 720p is the best res I could find on youtube.

I'm running another sim in phoenix. Still struggling to get those elongated glowy flames! Gonna have to arm myself with tons of patience and keep trying until it gets close enough.

Got a faster laptop though, with a 12 thread CPU. I'll be transferring all the data over it. Hopefully it's significantly faster than my old i7. PhoenixFD is a CPU calc sim, only. Though it previews through the viewport with GPU, and vray. Each sim about 20 mins. The scene is only 100 frames.

If you get to play with embergen and want to try this, let me know and I'll send you the solar panels.

 

[prometheus]

I once tried embergen demo. Need to download the latest version and see if it would allow me to test it, or not. Some programs don't allow you to do that. Once the trial expires, it's over. You buy it or leave it lol.

Yeah, there were several inconsistencies with Gravity and this is coming from a regular dude who knows little about physics! I think they over did a lot of stuff that made no sense. But, visually speaking, it's very good, but I haven't seen it on a big tv so not sure how the flames would have looked with more resolution. 720p is the best res I could find on youtube.

I'm running another sim in phoenix. Still struggling to get those elongated glowy flames! Gonna have to arm myself with tons of patience and keep trying until it gets close enough.

Got a faster laptop though, with a 12 thread CPU. I'll be transferring all the data over it. Hopefully it's significantly faster than my old i7. PhoenixFD is a CPU calc sim, only. Though it previews through the viewport with GPU, and vray. Each sim about 20 mins. The scene is only 100 frames.

If you get to play with embergen and want to try this, let me know and I'll send you the solar panels.

Quite busy the whole week, but maybe I get the chance to "fire " up Embergen, don´t think I need solar panels, dummy objects or a pod, or some other items from nasa I could use that, and then discuss findings about it perhaps.

As for demos, I used an older demo as well, then after they released a later version there was no problem with installing that..so I think you should be able to install the newest version there to try for 30 days, but after that you can not test more I think, so make sure you set a side a lot of time only for testing that.
I think they also are opened minded to expand on test period if you ask politely and blow smoke and fire up their...you know, to make them feel good.

Faster laptop, interesting, then you could perhaps render clouds faster in Lightwave, would be interesting to see difference in speed if you can test on both machines later on when time permits.

One thing to think about, not sure how you approached it, did you do a full velocity movement of the solar panel so it moves in high speed? and if such what speed did you have on it, or did you use the solar panels just fixes staticly with only rotation movent and camera movement, you could approach it that way as well.
Trying to simulate the whole scale of such a high speed moving object can be difficult or pointless actually, at least if you are to import it within a container domain in other fluid systems.

Here´s a clip in a 1080 resolution from a bit later in the movie when the station goes down, I bookmarked a time a little bit in the clip where the re-entry effects start to arrive,
And at 1:09 some more friction effects, and in 1:57 it starts to get really heated and break down spinning when re-entring.

 

[prometheus]

Ah..and I should be in bed now, but this effect is a bit fun to test in embergen.

Quick setup with emission object and collision object, no actual object import for this quick test, and the items are still, with only fluid emission turned down all bouyance and other forces within the actual emission, one line force to push it, strength can be increased as you wish to make it faster etc.
Test of repulsion makes a bit more swirl when the collision hits.
I didn´t hide the collision object, the sphere in this case, but that is easy to do in embergen and only work with the tweakings of the physics for the collision object, repulsion etc.
Smoke generation is turned off, as well as off in the render settings, only fuel and flame is used.

There are so many ways to push the fluids, no vorticity, increase vorticity, add turbulence force, wind chaos, pressure, pressure chaos etc.
But keeping it as simple as possible from scratch when working with collision force is probably a good approach, so turning down as many parameters as possible that causes undulations, basicly you want a straight flame initially.
I actually forgot to showcase some other settings that made some interesting results in the flame.
No narrations before bedtime, just a teaser.

Then there are some options, either export to vdb or make a pixel perfect camera match, and render the fluids, Or..use the particle render in embergen and render that as well,
It has some nice character to it that could complement the fluid gas.
The shading could be adjust much better and also more dissolve to fade faster, final effect should be soo much more subtle, and work on several sections to sell the effect of affected panels or other objects re-entering, together with setting camera shots up of course., but it will have to do for a teaser, just messing around for a first test....

 

[prometheus]

Jason key on youtube does some nice experiments with Embergen.
here´s a comet trail, principles on how that is made can be tweaked for re-entry objects in atmospheres as well, also check his Embergen Realtime VDB clouds in Unreal, and yes..they can be animated as well, working with godrays in realtime to a quality I really haven´s seen before in real time.

 

[Planeguy]

Quite busy the whole week, but maybe I get the chance to "fire " up Embergen, don´t think I need solar panels, dummy objects or a pod, or some other items from nasa I could use that, and then discuss findings about it perhaps.

As for demos, I used an older demo as well, then after they released a later version there was no problem with installing that..so I think you should be able to install the newest version there to try for 30 days, but after that you can not test more I think, so make sure you set a side a lot of time only for testing that.
I think they also are opened minded to expand on test period if you ask politely and blow smoke and fire up their...you know, to make them feel good.

Faster laptop, interesting, then you could perhaps render clouds faster in Lightwave, would be interesting to see difference in speed if you can test on both machines later on when time permits.

One thing to think about, not sure how you approached it, did you do a full velocity movement of the solar panel so it moves in high speed? and if such what speed did you have on it, or did you use the solar panels just fixes staticly with only rotation movent and camera movement, you could approach it that way as well.
Trying to simulate the whole scale of such a high speed moving object can be difficult or pointless actually, at least if you are to import it within a container domain in other fluid systems.

Here´s a clip in a 1080 resolution from a bit later in the movie when the station goes down, I bookmarked a time a little bit in the clip where the re-entry effects start to arrive,
And at 1:09 some more friction effects, and in 1:57 it starts to get really heated and break down spinning when re-entring.

"One thing to think about, not sure how you approached it, did you do a full velocity movement of the solar panel so it moves in high speed? and if such what speed did you have on it, or did you use the solar panels just fixes staticly with only rotation movent and camera movement, you could approach it that way as well.
Trying to simulate the whole scale of such a high speed moving object can be difficult or pointless actually, at least if you are to import it within a container domain in other fluid systems."

I haven't tried that in phoenix. Not sure how it would work, but might be worth checking!

Oh, and yeah, cloud rendering would be in! I got this specific laptop because the GPU can be swapped with a faster one. Only thing I dislike it's the trackpoint (it's not as precise as Lenovo's).