The objective of this assignment was to mimic fire as observed in real life. My fire observation was a standard disposable lighter.
I used a Panasonic DVX-100 on progressive mode to capture several different fire examples to get a range of how fire reacts, as well as multiple examples of a lighter. I then researched on the internet and found high-speed photography of a lighter being lit. Although I eventually decided not to mimic the high-speed photography footage, it was important in my understanding of the relationship between the gas, sparks, and flame.
Since a simple metaball can just be colored to look like fire, my focus was to get a very real motion with subtle movements just as my source footage expressed. For this, I devoloped a simulation of fire, as opposed to creating fire in 3d.
| Fire Simulation Relationship |
|
| figure 1.2 |
By creating needs and attributes to each group of particles in my simulation, my intent was to have a system that was self sustaining, where it would reduce itself when was is lacking any of the necessary elements, and increase when it was given ample requirements. Finally, I had begun to put what
"Smokey the Bear" had taught me to good use. '
Calling upon my programming roots, I decided to work in Houdini in a manner similar to a flow-chart by using netboxes.
| Fire "Flow-Chart" Netbox Organization in SOP |
|
| figure 1.3 |
By using netboxes, I was able to take several POP operations and group them according to what the group, as a whole, represented. In programming, this is very similar to a function, only in houdini, each group feeds into the next to allow the simulation to continue.
One could read the figure 1.3 as follows:
Gas exists and is being created.
Sparks also exist and are being created.
When the two combine, under certain circumstances, there is an ignition of flame.
The inital flame creates a constant flame based entirely on it's lineage.
This simulation created not only something quite interresting, but also many things to consider. After applying a metaball surface to the simulation, and creating a makeshift x-ray shader, I ended up with the following (after several tweaks of color values this way and that).
Saturation, Opacity, Edges, and Color were among the few of the things that those who would give me an honest critique told me were wrong with it, but a good start at that. First to get a more "wispy" feel to my fire, I created a few expressions to control opacity, particle velocity (to enhance motion blur), as well as particle scale. Extensive overhauls were done to the lower half of the pop network, and the flame igniton was all but left out, but it still remained important in deciding the shape of the constant fire and when it was to ignite.
After a seemingly a million tiny tweaks, I generated this image.
| Revised Render |
|
| figure 1.5 |
Although the colors were far closer to what I wanted, they still didn't have that feeling of burning. I attempted to copy the fire meta, and scale it up slightly and attach a seperate shader to create a glow, but I found that in compositing, I could achieve much better results and it only took a fraction of the time to create, and each frame had half the geometry, since the meta didn't need to be copied.
I ended up turning in the version with the color correction and glow applied in post compositing (see figure 1.6). Although I'm fairly happy with the work, considering it was my first particle system in Houdini, I am saving all necessary files to go back and touch up my simulation, as well as possibly using it in a composited scene to see if that would help elminate the "CG look."
Even after the short period between when I created this and the creation of this website, I feel as if the level of work I created here is far below what I'm capable of. I will revisit the problem of creating realistic fire in the future.
