[Swifty]

Let's try this again. This time I'm lucid enough to include a patch. The last soft particles build didn't work so well because the unified shaders kept overriding my soft particle shaders. Hopefully this fixes things.

UPDATED: http://www.mediafire.com/?gcd7nfhb85n1htt

Soft sphere quad rendering has been extended to most effects. The only things left that this shader can be applied to are beam muzzle glows and glow points.

Example screenshots here: http://imgur.com/a/zv3ox

[Spoon]

I'm definitely seeing the soft particles now. Looks pretty nice 'cept for all the artifacts I'm getting:

White squares around the borders on some effects and a few (really hard to screencap those) green and yellow circles around a few explosions that just flash by during the first few frames

And when beam piercing start happening...

[Kolgena]

Omg, those fireball artifacts fit right in with WoD!

I'll be sure to test this when it's more complete (with fireballs support, which may/may not be this weekend)

[sigtau]

Dunno if this applies to FSO in the way it applied to the game engine I'm accustomed to working with, but when you get white borders around billboards and so forth, it's fixable by clamping the textures to the edge of the billboard (rather than have them use the same repeat mode as the other shaders).

Does that help at all?

[Zacam]

So far, this has been working out really well. (I'm using the suggested modifications as proposed by Valathil via IRC)

Code: (soft-f.sdr) [Select]

float depthOffset = radius - sqrt(
pow( radius * ( abs(0.5 - gl_TexCoord[0].x) * 2 ), 2 ) +
pow( radius * ( abs(0.5 - gl_TexCoord[0].y) * 2 ), 2 )
);

Sadly, the nVidia Cg Compiler loves being phenomenally stupid and thinks the instance of "2" in the shader are being declared as INT causing stupid warnings about implicit casting.
We know that it's not, obviously, but it would probably be bad for it to assume it's treating it like an INT rather than a float, so I turned the 7 occurrences of "2"  into "2.0" to make it shut up.

[Valathil]

Sadly i found a problem yesterday but couldnt get swifty to respond. This radius based approach is really a great idea no doubt. BUT the geometry batcher in use for particles only saves the radius per particle texture. So what you get is everytime a new particle is drawn you overwrite the radius and use it for all other particles with the same texture. Its almost like a random number generator for the radius ergo you get horribly flickering particles which is kinda sad cause there is no way to fix it without forgetting the geometry batcher alltogether and porting it over to a geometry instancing framework or vertex attributes which i heard he is working on.

[Sushi]

Yet another reason why using particles for explosions is a bad idea...

[Swifty]

Yet another reason why using particles for explosions is a bad idea...

Err, what? What Valathil brought up is being fixed right now.

[Sushi]

Yet another reason why using particles for explosions is a bad idea...

Err, what? What Valathil brought up is being fixed right now.

The fact that we use particles for a lot of explosions (such as missile and flak explosions) has a number of annoying consequences. Ship explosions cast light, for example, but missile explosions do not (no particles cast light). Particles are orientation-less: if you are rolling while viewing them, the explosion rolls with you. I think particles also don't render if they are just barely off the edge of the screen, so they can sort of "pop" into existence in an odd-appearing way.

That said, last time I looked at particles was about a year ago, so perhaps some of those annoyances have already been fixed (or will be soon). 

[Bobboau]

you have implemented this on thrusters, right?

[Swifty]

The fact that we use particles for a lot of explosions (such as missile and flak explosions) has a number of annoying consequences. Ship explosions cast light, for example, but missile explosions do not (no particles cast light). Particles are orientation-less: if you are rolling while viewing them, the explosion rolls with you. I think particles also don't render if they are just barely off the edge of the screen, so they can sort of "pop" into existence in an odd-appearing way.

That said, last time I looked at particles was about a year ago, so perhaps some of those annoyances have already been fixed (or will be soon). 

Funny, while looking through the fireballs rendering code right before I read your response, I noticed that explosions and shockwaves were intended to rotate independently of the player's orientation. That has definitely not been happening. A simple one line change seems to have fixed it. But it involves adding the negative value of the Physics_view_bank angle value to the rotation of the bitmap. Which is weird because nebula bitmaps are oriented just fine when adding the positive value of Physics_viewer_bank. They're definitely calculating their orientations differently...

But yes, particles don't have their own orientation or up vector. The geometry batcher that the particle system uses has support for rotated billboards but I don't think anything is using it. We could possibly code in weapon hit explosions to exclusively have orientations while leaving the less noticeable smoke/fire/spark particles to share their up vector with the player.

you have implemented this on thrusters, right?

Not yet but I can do that real quick.

[BlasterNT]

Ooh, particle thrusters and Valathil's refraction shader allows for http://www.youtube.com/watch?v=PC4y2T0-skM&feature=player_embedded#at=33

*Edited link for better timing

[Dragon]

You mean the distortion at the start? That would indeed be great.

[BlasterNT]

I do mean that.  And all other sorts of haze effects*.  Ahh, the possibilities! 



*haze in vacuum, lol

[Dragon]

It could be a great addition to atmospheric mods, as well as to normal FS (if you're willing to stomach the lack of realism).

[MatthTheGeek]

REALISM ?

IN MAH FREESPACE ?

[Jeff Vader]

[Valathil]

You do know that thrusters exhaust gas as a means to generate thrust dont you. So having thrusters with heat distortion in vacuum is perfectly realistic btw

REALISM ?

IN MAH FREESPACE ?

It's more likely than you think!

[MatthTheGeek]

Aww come on. It was cooler when it wasn't realistic.

[Valathil]

You still have perfectly contained cylindrical beams and non newtonian physics dont cry.