This week I worked on some preliminary research on how fire and smoke look in reality and what sorts of effects are possible with real-time rendering today. I wanted to do this so I knew what was possible to do before I plan to do something that perhaps couldn’t be possible. I started with looking at what fire actually is, and according to HowStuffWorks.com and about.com [HowStuffWorks.com Fire] [about fire] it is basically the energy (light and heat) created, which is usually called “combustion”. So the fire you see is actually bright carbon atoms.
Looking at the images above, we see the shape of fire closely resembles the particle simulation I worked on except you can see the individual particles in the simulation. This is because, in reality, the particles are at the atomic level, and therefore there needs to be many many more particles in the simulation to get it to look like real fire. Also, in the simulation, each particle is represented as a 1×1 pixel dot on the screen, so further away particles will need to be smaller and closer ones be bigger. I tested to see how small I would need to make a large export of a screenshot and the result was that it varied depending on the density of the cloud. For example, with lots of spaced-apart particles. As a good mid-point to this, I decided that one could shrink the cloud down to 20% and it blended the particles together to appear close to a real fire effect.
[image of particle shrunk down]
I tested it in Unity with a fire effect and it looked okay after a bit of adjustments to the obscure shape.
[images of fire in Unity game]
Another test I did was to try and find some stock image effects in Adobe Fireworks that I could apply to the image like a glow effect or blur effect, but neither improved the image.
[images of blurred particle fireworks screenshot]
I will need to find some special image effects to see how they look. One that could improve the look of the cloud would be an effect that blends close particles together as large round balls, similar to a lens blur but keeping the solid edges of the particle. This may be impossible, but it would be good to test anyway.