I am wondering if it can perform as well as cegs, is is able to use the cegs already created or do you effectively have to write all new cegs? What is it? How much performance impact can it have and what are its limitations? This is just an open discussion on strength and merits.
Please some of you please do not turn this into some pissing match. I just wanted to learn a bit more as I have heard it mentioned a few times.
Can LUPS truely replace CEGs?
Moderator: Moderators
- Evil4Zerggin
- Posts: 557
- Joined: 16 May 2007, 06:34
Re: Can LUPS truely replace CEGs?
AFAIK LUPS cannot read standard CEG files, although it does duplicate much of the functionality of CEGs. I remember jK saying it was faster than comparable CEGs, which doesn't surprise me--CEG and COB are interpreted languages (I think that's the terminology?), implemented in the most straightforward way that one could expect; there's probably some GLSL magickery that makes it faster too.
Re: Can LUPS truely replace CEGs?
There is a SimpleParticles class, which emulates to 100% the behaviour of the ceg one (just the direction tag is missing). But because the used airdrag equation couldn't be optimized for shaders (some airdrag values are calculated on the CPU) I wrote a SimpleParticles2 class which use exponential functions instead, and so runs to 100% on the GPU.
Back to topic, there are still some classes missing like groundflashes and raising spheres (spikes are possible).
A problem are particles trails, cos the performance decreases dramatically with the number of fx (not particles!). But ribbons are doable (rockets etc.).
Biggest problem is that there is still no uniform interface for weaponFX in LUPS, atm each fx is implemented by its own script.
So my todo list for LUPS is:
* render all unsynced fx in the same pass as synced fx (saves double work, i.e. in distortion pass)
* write an ingame dev IDE (atm you already can edit fxs via /luaui reload, but that's still too complicated imo)
* write uniform synced (weapon) FX interface
* implement trepan's VBO userdata object and make particle trails possible
Back to topic, there are still some classes missing like groundflashes and raising spheres (spikes are possible).
A problem are particles trails, cos the performance decreases dramatically with the number of fx (not particles!). But ribbons are doable (rockets etc.).
Biggest problem is that there is still no uniform interface for weaponFX in LUPS, atm each fx is implemented by its own script.
So my todo list for LUPS is:
* render all unsynced fx in the same pass as synced fx (saves double work, i.e. in distortion pass)
* write an ingame dev IDE (atm you already can edit fxs via /luaui reload, but that's still too complicated imo)
* write uniform synced (weapon) FX interface
* implement trepan's VBO userdata object and make particle trails possible
Re: Can LUPS truely replace CEGs?
If there's ever a standardized way to write a CEG-like system in the context of LUPS, I'd be more than willing to write a lot of particle system examples and do a lot of testing. The lack of a standard way to set up the parameters, etc., and the lack of documentation has kept me from getting to this.
Re: Can LUPS truely replace CEGs?
Even without the VBO interface, you could do a lot (and faster
than CEGs), by using vertex textures (as demo'ed by the volcano
widget). The downside is that the technique only works on newer
video cards (so an alternative would have to be provided anyways)
than CEGs), by using vertex textures (as demo'ed by the volcano
widget). The downside is that the technique only works on newer
video cards (so an alternative would have to be provided anyways)
Re: Can LUPS truely replace CEGs?
Meh, I'd use this in a heartbeat, all things considered, if it gave me the ability to throw a lot more particles around, did stuff like groundbounce, etc., like the volcano demo did. That was a pretty amazing tech demo. Maybe I should go read the source again, now that I have half a clue, if it's still available.
Re: Can LUPS truely replace CEGs?
using vertex textures has many more drawbacks
Re: Can LUPS truely replace CEGs?
such as?