Oddly enough, or not I guess, I too have never found it useful to use particles for anything but velocities and the occasional odd or end. I suppose when your source isn’t visible it is a different story. I have always used mapped simple or object sources, I believe you have far greater control with mapped emission.
I think for the Particle Source to be truly useful it would have to take particle shapes into account, basically turning it into a group object source.
You could do that if you Mesher the particles. Not sure if their velocities would still be intact however…
A quick question regarding WT: do you apply the WT only to the fire or also to the fuel?
I’m currently testing the WT on a 1000x350x750 grid, with WT grid detail scale set to 3.0, resulting in a grid of 3000x1050x2250.
I’ve read in the manual that the Strength is based on the grid size so the larger the grid, the larger you should set the Strength. Can you give me a ballpark figure what strength-value should be set? I’ve tried values from 1 to 1000 but I seem to be running around in circles regarding the right Strength/Treshold and not getting decent results.
The only thing I want to do is eliminate the voxel artifacts. I even applied the WT with all channels unchecked which upsamples without applying WT, but no luck.
David
EDIT: I’ve added some screengrabs (partial grabs of the same part of the image, the full resolution of the images is 5500x4125) to illustrate the “problem”. The first one is a render of the original 1000x350x750 grid (no WT), where you can see the voxel artifacts. The second one the settings which lead to the third image (rendering of the 3000x1050x2250 WT grid). You see that a lot of detail is lost when adding the WT. Due to me doing something wrong 
Threshold is the important value, it determines in broad terms how many places the effect will be present. Strength is respectively how much will take place at each point.
Not really sure there is a ballpark figure although I would assume that the generic settings are based off of a default grid of 100x100x100 with a spacing of 1.0. So if a grid were 5 times larger I would expect the “base” setting to be 5 times higher strength. This is purely speculative as I have not tested it to see.
Yes, I’m playing with the strength and treshold.
The manual states that 2.0 is the default for a 100x grid and 20 for 1000x so I’m making some educated guesses based on that. But it still looks “ugly”. The thing that boggles my mind is that most of the times the WT results in a change of look at the edges (the flames become more transparent at the edges).?Also see the screenshots in my previous post. I assume this is what the extra WT-detail is trying to do, but not the look I’ m after. I’m playing with the treshold (raising it) so there is less detail.
Maybe it’s because of the high resolution that the settings are a bit of world and rather hard to determine. But I keep searching…
Hi.
I have one problem and can’t find the right (production proven i mean) method.
99% explosions i see at vimeo, youtube is large scale with just big amount of expansion, fuel and temp. But i want to achieve something like that:
Especially in the very beginnig. And then explosion begins at the impact point, expands and curls those streaks of smoke.
But i don’t know how far should i go with particles to sculpt this? Only the tips with several trail particles and let fluids do the rest? Or use long trails?
I tried to sculpt “cones” with particles here - https://vimeo.com/42033844
but the problem is if i inherit particle velocity by fluid it starts to curl (i think as result of removing divergence from velocity field) and if i don’t - smoke just don’t grow in the right direction.
And second - smoke (without combustion, fuel) is strictly tied to particle. as soon as particle dies it stops to evolve.
Hope someone could point me to some info - methods, quantities (maybe i’m doing right things but with low particles number, low grids etc).
I thought to use krakatoa to render high quantities of points with greater mass to behave as sand, ground fracture and in addition emit some smoke from them. But i don’t know how effectively comp these two later.
Thanks in advance. Any help is greatly appreciated.
So I’ve been playing around with this Wavelet Turbulence stuff, and for some reason, it seems to make the edges of my smoke look very “hard” instead of the nice soft falloff I had with the previous sim. Anyone know why this is?
Try lowering the threshold of your fire and smoke, right now it looks for more the defined stuff to wavelet, so by lowering the thresh hold it will push more through the wavelet when it runs, rather than only the denser areas, that’ll keep it more consistent.
Same here with the WT.
My orginal high res sim (1500x-2000x grid) is very smooth with nice gradients, but with some voxels artifacts (due to not enough voxels). When I apply the WT it becomes rather smeary and dirty a long with the hard edges. I tried a lot of WT-settings but not really satisfactory.
A few observations I would like to share:
-When you run the the WT without any checked channels, you get the most “smeary look” I assume this is because everything is subdivided (based on the WT scale) but no actual WT is applied to “clean up” the smeary voxels.
-I tried the most strange WT settings (even up to 1000 in WT strength) to get higher detail, but always I get a dirty look, it seems I can not get the WT to be “small” enough to be crisp at 6400x4800 renderings (this is a FFX grid (after WT) in the range of 3000x-4500x). I found this strange since the original sim (without WT applied) is already very smooth with only small voxel artifacts, applying WT which subdivides this high res grid by 2 or 3 results in much less smoothness but more artifacts, strange…
-With higher tresholds only the fast moving parts of the fluid (in my case fire) are affected by the WT. The other parts are only subdivided based on your WT scale (resulting in the smeary look). So I can not use this technique to preserve my soft edges while at the same time upping the res to remove the voxel artifacts of the original sim.
-The WT detail only looks good when zooming out my renders so at approx 3200x2400 it looks almost nice (you get very detailed flames, almost “noisy”).
_my current conclusion is that I have to do it without WT and trying to up the scale to immense levels (3000x grids look nice up to 4000x3000 rendering resolutions, anything higher exposes the voxel artifacts too much) to resolve the voxel artifacts, but it seems that the 64GB in my machine even isn’t enough to pull this off (I need renderings at 6400x4800)… I really don’t know how to solve it honestly. I’ve been trying to solve this for 4-5 weeks now (even ordered a new 64Gb machine) but I think I’m pushing FFX too far. The only solution I see is more ram but that is not an option …
Out of curiosity have you guys re-adjusted your render AFC’s at all? You can pretty much eliminate ranges of the simulation with spikes in the curves. Just a thought, maybe it will help maybe it won’t. I have noticed that after a wavelet my AFC’s are not the same as the original.
Have you tried rotating the grid on any XYZ axis so it is not looking straight on to the camera?
Something that just came to mind, if you where to figure out what the size of your print pixel is and matched to your grid size at the camera distance, maybe your artifacts would be least noticeable?
EDIT: hmm didn’t notice your posted pics until I visited the thread, I have been following this via email, now I see what you mean by smeary. Interesting indeed, are you using any substeps? How close to the grid is your camera? it almost looks like it is in it, which in all reality is bad. What I posted above still applies.
Although not a FumeFX sim (they used Flowline), MPC released a breakdown of a CG explosion used in Prometheus. It is probably the best looking CG explosion I’ve ever seen. If I could get mine looking that good, I’d be quite happy. Check out the breakdown here… (WARNING: SPOILERS)
http://www.cgsociety.org/index.php/CGSFeatures/CGSFeatureSpecial/prometheus
Argh, bummer, can’t look at it until tomorrow, I still have to see the movie, no spoilers 
The left picture the original sim and the right one is with WT applied. I think that if there are any visible banding (due to not enough steps) this is emphasized with WT. The original sims took use of multiple steps (but enough, I was testing if WT could cover that up). I’m currently simming the grid at high res with high quality and enough steps.
Than I’m going to take into account your tip about the AFC-curves to try to solve the voxel artifacts with the AFC-curve & to try to rotate it a little bit. Thanks for those tips.
Adjusting the pixels to the print pixels is difficult since it the image will be used in multiple resolutions/media.
The cam is not in the fire. The pictures attached are just a cut out out of a 4000x3000 rendering
Maybe i’m completely wrong, but i think ffx is too limited to produce something directable.
No access to fields (scripting is very slow), no volume math, no tweakable shaders (no BBR shader)). WT is just wavelet noise with only couple exposed params.
Effectors in 3.0 are more like 2.2 stuff. I think something like Magma mod for volumes is way more like 2012 year plugin.
Yes, i’m very disappointed, especially cause i can’t find the way to “sculpt” good explosion (not big atomic-scale one)) and nobody can’t help me and answer my questions couple of posts before=)
Yesterday I tried something that may help with the hard edges. I have not been able to run enough tests to be conclusive, but this is what I did. If my original sim has a smoke opacity of 1.0, and I wavelet turbulence it up to with a Grid Detail Scale of 2, then I change the opacity of my smoke to 0.5. If i did a Grid Detail Scale of 4, I would lower my smoke opacity to .25, and so on. This probably won’t help in all situations, but I was making some clouds and it seemed to work nicely.
Your smart enough go use Houdini, it has all you are asking for or use something else, it is as simple as that. As Kreso told me it is an artist tool, fair enough, I think exposure is decent for what it is, of course there could always be more. You get what you pay for, FumeFX or Phoenix for that matter are reasonably priced gaseous fluid simulators.
I haven’t “sculpted” many explosions or else I would have answered you. If I had answers to your questions you would have them. Your intelligent and capable enough to figure this shit out just like the rest of us have or have tried to do. sorry :shrug:
You always help me with valuable info or joke. thanks=)
And yes. i’m trying Houdini right now)
I should add fumefx is do it well, i just need more control and some leaked pro setups))
You can also try lowering the Step Size % and increasing jittering in the rendering tab. Maybe that will get rid of of those artifacts.