Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[WxTech]

Here are a couple of screenies showing a new treatment I'm trying for aircraft fires. Previously, my .mat files had  this typical entry (as do virtually all fire effect .mat files since year zero):

  tfBlendAdd 1

What this does is brighten the particles by adding to them the brightness of any superimposed texture, whether in front (if having some transparency so as to be seen through) or behind. This includes scenery, other effect particles, objects, etc. The problem here is that when two or more of an effect's particles overlap, the brightness adds pretty considerably. And when even more are superimposed, the brightness tends toward a washed out 'flat' aspect lacking detail.

In-game fires generally have the particles initially crowded more closely together, before they have a chance to separate via acceleration (such as when VerticalAccel is in operation, or the particle emitter is in motion and GasResist is not set to 1.0). The result for plane fires is the particles at and near the emitting source being crowded together, and hence overlapping, leading to a washed out, overly bright yellow with little if any detail seen over much of the particle.

By setting

  tfBlendAdd 0,

there is none of this additive brightening, and so the particles preserve a normal brightness and present full detail. This is a valid approach for fire because fire has a not large optical depth, meaning it has a fair degree of opacity. If you line up exactly a number of fires, only the nearest one is seen; those beyond are essentially invisible due to the opacity inherent to the flame.

Both these images illustrate the new approach, for both the aircraft fires from wing/engine AND the separated fuel leak fireball a bit in the background for the second capture. They also show the new textures I've created for the aircraft fires, they having much more detail than in the current effects pack.

[Click the pics to see full size, 1280 pixels wide.]

[Atoka220]

I like it, it seems to give a more natural look, especially with the new texture

For comparison here are 3 shots from the current 1.3 effects mod




I myself haven't seen such huge fires yet but what I noticed, we see them more in color, opaque and in detail as you said while camera recordings usually make them seem bright yellow or straight up white, like the effects adding up in game
Like this photo I took shows the same detail loss and overall whiteness in the flame, while I saw the fire in it's full color and detail

Also just managed to capture a real life phoenix lol

To be honest, I've never noticed it, as your effects as of 1.3 already look darn immersive. But now that you brought up this experiment made me realize really how much detail is lost in overlapping and seeing your new screens made me think through it.

Of course flame brightness also depends on the overall energy being released meaning more energy - more brightness, brighter colors. But that mostly goes for explosions as they have a momentary flash of bright white light followed by a fireball going from orange to red before fading out. Then it also comes to energy dissipation. White light has the most energy while red has the least, that's why explosions and fires go from very bright white to red before dissipating.

I think I've gone too far with the latest paragraph. 

Long story short, I think your new effect looks much closer to what we would see in real life and I'd like to see it in your new Fx pack version

[WxTech]

As your photo of the 'Phoenix'    illustrates, cameras have a very much more limited dynamic range than the eye. Whereas the eye can accommodate a scene having quite bright and rather dim elements, a camera will have one or the other--or both--overexposed to white and underexposed to black.

To be honest, I'm still on the fence as to which approach is the more effective in the context of the game visuals. More experimenting to do...

[Atoka220]

That's right

What I tried to explain there, the overlapping textures adding up in brightness and losing details looks like what a camera would see
While having them consistent would probably be like much more what our eyes would see
But again, I've never seen a burning aircraft (nor burning fuel) with my own eyes, so I can't tell for sure which would look truly better and realistic
Aside this, the textures are neat, I'll take them for sure. For the blending part, I'll leave it to your decision

[Orge Schwab]

Of course flame brightness also depends on the overall energy being released meaning more energy - more brightness, brighter colors. But that mostly goes for explosions as they have a momentary flash of bright white light followed by a fireball going from orange to red before fading out. Then it also comes to energy dissipation. White light has the most energy while red has the least, that's why explosions and fires go from very bright white to red before dissipating.

I think I've gone too far with the latest paragraph. 

Not at all, you are spot on. I could give numerous examples of combat reports where attacking pilots witnessed 'a sudden white flash from the fuselage'. One explanation is that the aircraft's pressurised oxygen bottles had been hit. In German bombers, they were usually positioned in the mid or rear fuselage, as they represented a clear danger to the crew, but in Allied heavies they were often sited close to individual crew positions (albeit covered in chainmail to protect the crew from flying fragments).
I have also read reports where a pilot witnessed a blazing cockpit fire in a bomber that had been attacked, and there are survivor testimonies confirming that this could happen if the crew were on oxygen when their aircraft was hit. If an incendiary bullet hit an air line, you essentially had a white-hot blow torch in the cockpit, which was very hard to extinguish.
If an aircraft was carrying incendiary bomblets (phosphorus?) and one of them was hit, again, a bright white flash would occur sometimes leading to a much larger explosion as other incendiaries caught fire (this would be a bomb-load dependent effect, I suppose).
A further variant sometimes mentioned in combat reports is the 'violet flash' from the fuselage - which I assume to have been some kind of electrical short as a vital piece of equipment was hit.

[genXgamer]

G'day WxTech

Glad this topic has come up, this is one of the things I wanted to contact you about and get your option on last year.
Switching to B.A.T. I thought the aircraft fires were a great improvement but displayed too much red.

I can only go off F-111 fuel dumps and footage of Concorde crash.
You can always take a look at how modern flight sims appear in videos on YouTube.

[WxTech]

Another crack at it. I've gone back to the 'standard' "tfBlendAdd 1" .mat file setting. Note that there are two different textures present here, for different fire origins. The textures have had the central portions made less color saturated, which eliminates the exaggerated yellow hue.

[genXgamer]

G'day WxTech

I personally think this looks better, what do others think?

Perhaps a poll?

[WxTech]

The strong yellow and red (the yellow in particular) began to grate on me, hence the desire to try a less color saturated fire. Near pure, primary colors are not common in typical combustion, where the spectrum is largely that of a black body. When there is not present any notable emission comprised of discrete spectral lines (as from the ignited chemicals present in tracers and signal flares), the colors for burning are not unlike those of the surfaces of stars.

I've been altering other fires where appropriate.

[WxTech]

I spent a whole day working over just about all other fires (for vehicles, ships, ship guns, buildings, fire objects). Besides texture hue and tone adjustment, this re-jigging includes the .mat file parameters which control vertical acceleration, wind, emit frequency, initial velocity. I graphed all these parameters as already existed in my current mod, then constructed a tidied up look-up table of all relevant properties as they scale with the initial fire particle size. This rationalizing has resulted in a an internally consistent behavior for all sizes of fire, the table including starting fire particle sizes from 3m to 40m. From just the starting particle size, all other parameters are read from the table, to reliably create a uniformly appearing and behaving fire.

[bomberkiller]

I personally think this looks better, what do others think?

It looks wonderful!

Many thanks to WxTech!

Best regards

 

Gerhard

[baggo]

Glen, I agree with you this looks more like raging flames than previous versions, your definitely on to something here, a very good improvement I think. Kev.