[RandomTiger]

I have added some code to process particles using D3D's point sprite system in order to try and speed things up a bit. The new system only has to send one point to the graphics card for every four the old system has to, it uses a smaller vertex structure and batches the particles as much as is possible. The new system only needs to transfer 20% of the amount of data to the graphics card.

-d3d_particle is the flag to activate this, when its fully tested and bug free I'll make it on by default. Some older cards may not support enough features to take full advantage of this which will cause the code to fall back to the old route when dealing with large particles.

I get a considerably higher frame rate for scenes with large number of particles, tell me how you get on.

http://mysite.freeserve.com/thomaswhittaker/c_code/freespace/fs2_open_r.rar

Dont forget to use the flag

[Lightspeed]

Can't really say about performance but apparently it still mixes up a few things:



The size is way too small, and they don't seem to appear everytime (as if i set a double $Time value).

[Deepblue]

Maybe the particles are now actually able to be adjusted with the radius function.

[Lightspeed]

the radius function does something really different

The $scale function sorts out the size.

It's a bit tricky with the names and nobody seems to know this but:

$Radius:

The "thickness" of the trail where particles may appear - if you set this to 2+ you'll swarm the whole screen with particles -- this is what i used for randomizing my trails

$Scale:

THIS one manages the size of the particle. I don't know in what it measures but when i set it to 20, about half of my screen was filled with the particle.

So none of those should have changed, since they always worked, even if nobody knew how  

[RandomTiger]

Originally posted by Lightspeed
Can't really say about performance but apparently it still mixes up a few things:



The size is way too small, and they don't seem to appear everytime (as if i set a double $Time value).

Thats a non standard effect I take it.
You would have to supply me with a copy of what ever it takes to make that to balance it out.

[Liberator]

It looks like Lightspeeds's new Akheton effect from this thread.

[Lightspeed]

it happens with any particle effect used.

i.e. where i noticed it: my morning star, my kayser, my maxim & my akheton. They're all messed up in similar ways (too small + wrong 'spawn rate)

[RandomTiger]

OK, I'll look into that, I can account for the scale problem but I'm not sure about the spawn rate.

Either way its optional default off so no harm done.
Will post to thread thread when I have an update.

[aldo_14]

There's not a possibility that you had a different fov setting in the command line (for the old exe) and forgot to set it in the modified exe, is there?

[Lightspeed]

There's not.

It's both the same exe, actually
(with 3d3particles enabled, and disabled)

i'm not sure if it's the spawn rate, but it looks like it is. Whatever it may be it needs to be fixed

[aldo_14]

Originally posted by Lightspeed
There's not.

It's both the same exe, actually
(with 3d3particles enabled, and disabled)

i'm not sure if it's the spawn rate, but it looks like it is. Whatever it may be it needs to be fixed

ah

To be fair, I didn't think it would be that...but never hurts to check.

[RandomTiger]

Originally posted by Lightspeed

It's both the same exe, actually
(with 3d3particles enabled, and disabled)

Hold on there.
Could you check what happens with any other fs2_open exe?
Its possible I've introduced a bug into the general particle code in this exe.

I would like to rule that out before doing anything else.

[Nuke]

i accidently found out about the radius flag while tweaking my nova flak cannon. the radius sets the area in which the particles spawn. i wonder if i could make a beam out of it. can we also have a $pattern setting for differend particle spawn meathods, such as random or spiral.

[Lightspeed]

Originally posted by RandomTiger


Hold on there.
Could you check what happens with any other fs2_open exe?
Its possible I've introduced a bug into the general particle code in this exe.

I would like to rule that out before doing anything else.

no, it looks correct (like with the older build) with the flag turned off.

[FreeTerran]

hmm sorry but i can't see anything that should be other with the particles

[Bobboau]

so does this replace draw_bitmap or are you running a big ole batch rndering?

is there any way to do something like draw_rotated_bitmap

[RandomTiger]

Originally posted by Lightspeed

no, it looks correct (like with the older build) with the flag turned off.

So are you saying the spawn rate problem was caused by my new code?

[Bobboau]

I think he's saying your build looks right whaen he doesn't enable pointsprites via the comand line

[Lightspeed]

Originally posted by RandomTiger


So are you saying the spawn rate problem was caused by my new code?

exactly. I turn the flag off and it works fine. I turn it on and whee -- we got the effect you can see on the right: not enough particles and screwed up size. It could be either spawn rate or speed though.

[Trivial Psychic]

I decided to give this build a try, but I got a CTD a split second into the mission.  I was using -htl and the custom flag if you're interested.  I'm thinking that this is either a hardware issue or a driver issue.  Since I'm running a GF2 MX 32MB, my card may not support this feature.  If it is supported, then it may be a driver issue, since I can't use the newest drivers for the GF2 MX.  I tried running the latest drivers for a few months, but it was a serious drain on my system, causing the screen to go black for a few seconds sometimes, occasionally resulting in a lockup.  It also created serious drags in some of my older games, so I reverted to the version I had been using since 2001... much more stable.

In any case, this older version may be resulting in the problem, if this feature IS supported by the GF2 MX.  What this boils down to, is me getting a Radeon 9600 Pro (or XT) 128MB card once my hours at work start to pick up sometime in March or April probably.

Later!