I wouldn't worry... it doesn't look at all static- seeing a still shot... er, let's just say it doesn't even slightly prepare you for what the animation looks like.
Can't make it drift with the wind, though- CSimpleParticleSystem doesn't allow for that.
CSimpleParticleSystem(tests, sample code when it works)
Moderator: Moderators
Further tests have proven, at least to my satisfaction, that there is indeed an error in the z-buffer sorting. Size of the particles makes no difference.
Maybe the size of the entire sphere is being used to determine z-order? If so... then that's not going to work. Neither is some simplistic solution from the centroid of the camera.
I know it's expensive to do a z-buffer sort by particle, but it's necessary, or stuff like this will look good only from an OTA angle... and terrible otherwise
It's not like FX of this complexity are going to be running constantly- and for the lighter ones, adding a per-particle z-buffer sort probably won't make them that much slower. You could even make it optional, for FX of this type that don't need much slower calculations. But the way it works now, I see tons of clipping that looks extremely ugly when viewed from the sides.
I've tried various ColorMap settings to try to reduce this problem, but to no avail.
Maybe the size of the entire sphere is being used to determine z-order? If so... then that's not going to work. Neither is some simplistic solution from the centroid of the camera.
I know it's expensive to do a z-buffer sort by particle, but it's necessary, or stuff like this will look good only from an OTA angle... and terrible otherwise
It's not like FX of this complexity are going to be running constantly- and for the lighter ones, adding a per-particle z-buffer sort probably won't make them that much slower. You could even make it optional, for FX of this type that don't need much slower calculations. But the way it works now, I see tons of clipping that looks extremely ugly when viewed from the sides.
I've tried various ColorMap settings to try to reduce this problem, but to no avail.
Last edited by Argh on 10 Nov 2006, 13:00, edited 1 time in total.
The glow and fire not showing through makes it looks like a disc shaped rocket lifting off into space, it just doesn't look that much like a nuclear fireball than an oddly shaped cloud.. But its very close already, you've got the shape sort of right and it looks much better than the last attempt, which sort of fizzled out looking a bit weak and sparse this one I'd expect more of an oomph from.
Is that smoke going to be in your GPL BOS script base?
Is that smoke going to be in your GPL BOS script base?
The glow/fire thing is one of the problems I am referring to with the whole z-buffer thing. What's happening right now is that FX are either getting drawn entirely over ... or not. Per system, not per particle. Which doesn't work at all right, for the layered effect I want. I was aiming for "fuzzy at the edges, impenetrable in the center", like the previous one, but with the new toys available in CSimpleParticleSystem. For that to work, each particle needs to be drawn on top ones farther from the camera's centroid, with alphas being added to one another. That's not working right in this version.
The smoke/fire is another project entirely. It may share bitmaps with this, but probably not. These are pretty much custom-built for the larger scales.
However, the SizeGrowth/Mod tags + SpawnDelay should allow for me to do them entirely with CSimpleParticleSystem, if I want to. Truth be told, it will probably be more efficient to use HeatCloud, which I think is faster code. I dunno yet. It does have to be fast code, though.
I've already written up one semi-dynamic flame, and I'm sure I can improve upon it over time. I'm not ready to put any of that on the table yet, though.
The smoke/fire is another project entirely. It may share bitmaps with this, but probably not. These are pretty much custom-built for the larger scales.
However, the SizeGrowth/Mod tags + SpawnDelay should allow for me to do them entirely with CSimpleParticleSystem, if I want to. Truth be told, it will probably be more efficient to use HeatCloud, which I think is faster code. I dunno yet. It does have to be fast code, though.
I've already written up one semi-dynamic flame, and I'm sure I can improve upon it over time. I'm not ready to put any of that on the table yet, though.
Er... I don't think you understand the true extent of the problem. Here, lemme put up a less-flattering screenshot, that shows the extent of the pop-over going on:
As you can see, there are obvious problems here.
While all of the smoke in this shot is using a ColorMap value ending with 1.0 (full additive alpha), it's clear that things aren't working right. For example, the central column should be thickening the effect of the smokepuffs behind them... that's not what happens.
Going the other way with ColorMap will cure the problem, but the cure is worse than the disease:
Here, I'm using alpha values of 0.1 for the smoke. As you can see, it is now adding each alpha/RGB value, resulting in a great deal of pure white, which is not exactly the desired result
As you can see, there are obvious problems here.
While all of the smoke in this shot is using a ColorMap value ending with 1.0 (full additive alpha), it's clear that things aren't working right. For example, the central column should be thickening the effect of the smokepuffs behind them... that's not what happens.
Going the other way with ColorMap will cure the problem, but the cure is worse than the disease:
Here, I'm using alpha values of 0.1 for the smoke. As you can see, it is now adding each alpha/RGB value, resulting in a great deal of pure white, which is not exactly the desired result
Last edited by Argh on 10 Nov 2006, 13:31, edited 1 time in total.
As a last note, before I crash into bed... basically, what's happening here is that each entire system is obviously being assigned a z-buffer level. Each system works fine on its own- it's only when I have multiples of them layered like this that there are serious problems. I've only had this problem with CSimpleParticleSystem. With HeatCloud, the problem was that the alphas were always additive, so I had to keep that in mind when designing the individual bitmap elements, for the first nuke, so that everything didn't wash out... and the smoke just uses a straight alpha, so it wasn't a problem.
You are correct that the systems aren't internally z-sorted, this is for speed reasons and for most purposes it isn't vissible. I'll see if i can add in an option for z ordering for those events that need it though.
Delayspawner spawn a completely new explosion generator, so it will have its new internal i and r.
Delayspawner spawn a completely new explosion generator, so it will have its new internal i and r.
That would be awesome, for the rare occasions when it's actually necessary. Needless to say, that isn't the case for most garden-variety explosions- my previous tests have shown that it's working quite handsomely for smaller-scale stuff with less complexity. However, having that option will make this kind of larger, layered event work, and the extra cost, CPU/GPU-wise will be worth it for sure.
-
[KnoX]ElementalGizmo
- XTA Developer
- Posts: 266
- Joined: 24 Aug 2006, 01:33
Sooooooooooooooooo. I WANT IT, I WANT IT NOW! lmao Awsome yum yums dude. XTA v8 had a few new effect that noruas has been playing about with, but that nuke WOW.
Please, when you get a vertion of this working to your high standards pass it on! Ill be over joyed to stick it in XTA v8.1
Ow and this :
PauloMorfeo:
Unit script and modell is from rattle, the explosion script looks like this
Code:
[properties]
{
emitVector =dir;
gravity =0,0,0;
colorMap=1 1 1 0.01 0 0 0 0.01;
texture =circularthingy;
airdrag = 0.95;
particleLife = 40;
particleLifeSpread = 30;
numParticles = 200;
particleSpeed =0.2;
particleSpeedSpread=0.5;
particleSize = 1;
particleSizeSpread = 1;
emitRot = 90;
emitRotSpread = 0;
directional = 1;
}
What dimentions are we talking about here for a texture for an effect. Ive no idea about creating stuff for this, but would love a vertion of that smoke on the BB n INTI on xta.
Truly cool stuff on this page. There should be a main page about effects for the developers. I found this the other day......
http://taspring.clan-sy.com/phpbb/viewtopic.php?t=8040
Also extreamly cool stuff that i havent seen b4. Is there an EXPERIMENTAL FX page about i havent found yet? Or am i doomed to wondering the forums trying to find cool FX to add here n there ¿
Please, when you get a vertion of this working to your high standards pass it on! Ill be over joyed to stick it in XTA v8.1
Ow and this :
PauloMorfeo:
Unit script and modell is from rattle, the explosion script looks like this
Code:
[properties]
{
emitVector =dir;
gravity =0,0,0;
colorMap=1 1 1 0.01 0 0 0 0.01;
texture =circularthingy;
airdrag = 0.95;
particleLife = 40;
particleLifeSpread = 30;
numParticles = 200;
particleSpeed =0.2;
particleSpeedSpread=0.5;
particleSize = 1;
particleSizeSpread = 1;
emitRot = 90;
emitRotSpread = 0;
directional = 1;
}
What dimentions are we talking about here for a texture for an effect. Ive no idea about creating stuff for this, but would love a vertion of that smoke on the BB n INTI on xta.
Truly cool stuff on this page. There should be a main page about effects for the developers. I found this the other day......
http://taspring.clan-sy.com/phpbb/viewtopic.php?t=8040
Also extreamly cool stuff that i havent seen b4. Is there an EXPERIMENTAL FX page about i havent found yet? Or am i doomed to wondering the forums trying to find cool FX to add here n there ¿
Tested CSphereParticleSpawner. The z-buffer sort works- and you're right, it's quite expensive, CPU-wise, so I will have to be careful.
It doesn't seem to be affected by Gravity, however. Everything else seems to work. I need Gravity to perform the animations, though.
It doesn't seem to be affected by Gravity, however. Everything else seems to work. I need Gravity to perform the animations, though.
Last edited by Argh on 18 Nov 2006, 02:53, edited 1 time in total.
