Welcome back :)
I'll start off by saying this is probably going to be a shorter post. Mostly I'll be working on some of last week's feedback and nothing new.
Some comments regarding the 3D Model
So, I've been lagging a bit on the 3D modeling side of things. There isn't much to say in this week's blog other than I'm chipping away at it. The goal is to have it finished by week 8, though. I'm also going to need to create a simple rig for the helicopter to animate the rotors and blades spinning. I'm going to create a rig for the proxy too in case the model takes longer than expected. The rotor motion will be needed when I get to creating velocity fields for how the spinning blades will affect my pyro.
For the procedural floating islands, I won't be working on that much this week either. I'll mostly be doing more research on different methods of creating procedural cliffs and other such structures before applying this research into my project next week.
Now, Addressing Past Notes...
Collisions
I got them working! I don't know how but sometimes Houdini is strange like that. I had to turn down the VDB voxel size quite a bit to get the rotor blades in there, but collisions appear to be reading well now!
Main Pyro Shaping
Jumping into the main pyro, there are too main issues that stand out. 1) The smoke looks too blobby and 2) the flames near the source still move too quickly.
1) Smoke Shaping
For the first one, by blobby I mean, can you see how it kinda looks like a bunch of smoke spheres being emitted rather than a smooth looking smoke? That's what I mean.
My assumption here is that the issue is the velocity. I had to bump the source scale way up in the pyro node to get better shaping but I think that maybe that may have led to points where there was so much velocity shooting out that little mushroom clouds were forming everywhere.
To test if velocity was the issue, I doubled the pulse durations on my velocity noise nodes, to slow down the animated noise. I broke up my noises into two adjust vector nodes, one for direction and the other for length. Under the length one, I also turned down the element size to allow for more breakup in lengths (initially, both noise nodes were set to the same element size and offset
This...barely helped. I messed around for a while longer until I got some settings I liked: I first combined both my velocity noises into a single node for cleanliness then bumped up the element size way up. I also turned up the turbulence and its swirl size. This gave me much better results. However, I'm still getting some of those blobby "mushrooms", though it's reduced quite a bit.
2) Fire Shaping
For this section, I first increased the pulse duration and element size on m velocity noise. Then, I created a new pyro solver to start from scratch. I still had to bump up my velocity source multiplier quite a bit, but in the shape tab I only added a little bit of turbulence, shredding, and viscosity.
There is still, unfortunately, quite a bit of stepping that I'll need to further troubleshoot.
Initial Fire Fixes
The main issue I've been having with this particular element is the blobbiness around its source. I hopped on a call with my professor to get some help with this issue since it was giving me quite a bit of trouble. What we found out was that I had neglected to reduce the size of my particle size inside the pyro source node, which was the primary issue. After that, I added a point velocity node to add some velocity from the helicopter's motion to my flames. I had to use a wrangle to reduce this a bit since it was too strong. Then an attribute adjust vector was used to spread out the velocity a bit. Finally, I jittered the positions of my trailing points to get rid of blockiness. To do this, I used an attribute VOP to multiply the velocity and time increment (which would give the next frame position of the particle) then added that to its current position.
After some tweaks in the pyro solver, here are the results:
*Note: I did apply some of this knowledge to my previous smoke adjustments as well, namely the trail jitter.
The Missile Trail Round Whatever it is Now
I think I might have got it now! I got advice from my professor on this element in our call as well. First thing I did was clean up my network. I also got rid of all that P attribute noise since I realized I could have just added a POP wind and turned up the swirl size a whole bunch. Much easier 😅. Then, I adjusted my ramps a bit for the pscale and density scale by age. Or, normalized age. It's better to use nage since it calculates age per frame, rather than by second. The reason I hadn't done this previously was because it was giving me a very strange jitter, which I learned was due to my particle life variance being too high.
Finally, for the volumes. Instead of a VDB from particles, my professor recommended using a volume rasterize from particles. I had to make quite a few adjustments to my pscale and density afterwards, but eventually I got a look I liked.
Now, the important question: will it render?
Yes! It's rendering!!! Woohoo! 🥳
The Final Render
At this point, I feel ready to render. Here is my final for this week:
Now I did remove that secondary smoke element I had added last week for more coverage. With the updated primary smoke, it seems to be covering more screen that I felt the extra element was unnecessary.
Realizations
So, I'm righting this literally the night before I post this, but I think I may have figured out why my main pyro element was producing all those mushroom-y shapes and why it's not going away, as well as why my fire is moving so fast. See, my professor told me that to test if velocity is the cause, reduce the pulse duration on the noise. That barely helped and then that's why I turned up the element size, which at least reduced the size. But, look at this flipbook with noise animation turned completely off:
I think what's happening here is that my noise is "swimming" back as the helicopter moves forwards. And since it moves so quickly, it gives the appearance of fast moving noise. I think what I'll have to do is create a rest position for my noise to stop the swimming. Another issue is that the point velocity for my source I initially calculated also flickers a bit, which may be impacting the noise.
Additionally, I will be working on creating custom velocity fields for the pyro, namely to drive how the spinning rotors of the Samson cause the smoke to swirl. That will have to wait until I rig the model.
However, I unfortunately have run out of time to test this this week, so you'll have to wait for the next! See you then :)
Comentários