What about next-gen particles?

Shifty Geezer said:
But as PS3 is already squeezing the available BW with just HDR+AA, is it fair to say it'll likely be at a disadvantage when it comes to masses of particles, and hence those sorts of effects will likely take a backseat?

Could that be a Area were Cell would help lots?
Render the Scene without Particles and transparent Polgons, take the ZBuffer and then let Cell produce an Image of the particles only, finally blending it together on RSX. Aslong you can find a solution which can compute Particles per Tile, it should be a nice fit for SPE`s with their local store.
 
london-boy said:
I think ZOE2 is still the best, craziest, busiest example of particle effects.

You haven't played Otogi 2 then?

BTW, Hi all, I'm new to this forum - Been lurking for a while.

Great forum!
 
london-boy said:
True, true, but i tend to consider everything else so draining on bandwidth already (from AA to HDR to hi-res and textures and just about everything else), that particles are the last in the list of bandwidth-suchers if you know what i mean.
Don't know, it's just a matter of them being "fillrate limited" than "bandwidth limited" though in the end one affects the other.

The point still stands, alpha blending bandwidth, and indeed FP10 HDR bandwidth will be there in Xenos but not in RSX. Fill-rate/framebuffer bandwidth limitations will not affect the feasibility of particle effects on Xenos.

What will save RSX is that it can't do HDR and AA, so the bandwidth saved in leaving out AA will help with particle effects.

Jawed
 
The point still stands, alpha blending bandwidth, and indeed FP10 HDR bandwidth will be there in Xenos but not in RSX. Fill-rate/framebuffer bandwidth limitations will not affect the feasibility of particle effects on Xenos.
Even so, I'd expect a sizeable difference in fillrate demands using voxels as opposed to sprites. I suppose it may be remotely feasible using a huge density field texture for the whole effect as opposed to tiny 3d-textured voxels for each particle. Yeesh, the overdraw. eDRAM or not, you're bound to hit fillrate limits if you're drawing a few tens of thousands of pixels hundreds of times over and over again.
 
Would not a more complex 3D system avoid overdraw? Indeed this is something raycasting might come in very handy. You could even get away with casting rays foronly every 3rd/4th pixel and interpolate, determining cloud density and composite the particles over the converntional render. Maybe.
 
Jawed said:
What will save RSX is that it can't do HDR and AA, so the bandwidth saved in leaving out AA will help with particle effects.
Who says particles must be rendered by RSX though? 8)
 
ShootMyMonkey said:
Even so, I'd expect a sizeable difference in fillrate demands using voxels as opposed to sprites. I suppose it may be remotely feasible using a huge density field texture for the whole effect as opposed to tiny 3d-textured voxels for each particle. Yeesh, the overdraw. eDRAM or not, you're bound to hit fillrate limits if you're drawing a few tens of thousands of pixels hundreds of times over and over again.

You'll be vertex or fragment shader limited before framebuffer-bandwidth/fill-rate limited.

Jawed
 
Fafalada said:
Jawed said:
What will save RSX is that it can't do HDR and AA, so the bandwidth saved in leaving out AA will help with particle effects.
Who says particles must be rendered by RSX though? 8)

I can't wait to see the results of developer's experiments in using Cell to post-process a framebuffer, or perhaps in this case, render particle effects volumetrically with clipping planes for local objects and convert the result to a simple object covered in dynamic textures (i.e. the particle effect), for RSX to render and alpha-blend into the final scene.

It's such early days :( it's going to be fascinating to see what developers do...

Jawed
 
Would not a more complex 3D system avoid overdraw? Indeed this is something raycasting might come in very handy.
Yeah, but you actually need the overdraw when alpha blending and having accumulated transparency (assuming that each individual particle isn't going to be completely opaque). Sure, after the alpha has built up to fully opaque, you don't need to draw anything behind, but even then, alpha blended polys are usually rendered farthest to nearest because the accumulation often comes out just wrong if you don't.
 
I was thinking more cast a ray and reduce it's intensity (or increase for light particles) for each particle it passes through. You would only need traverse until a percentage occlusion, so wouldn't need to draw every single particle back to front. So in say 1000 particles deep cloud of smog, by the time the ray's passed through the 10th particle's volume it's already black so stop there.

Actually that wouldn't work. If you've got a super bright light-source at the end, or in the middle, you'd need to keep tracing until you reached the lightsource (or rather, bright object) to add it's intensity to the traced pixel. So you'd have to keep tracing until you hit an opaque surface. I s'pose if particle data could be efficiently enough described it wouldn't be too hard on the memory to do this, using the z-buffer from RSX to determine length of the ray, creating a half-ray traced renderer.
 
dukmahsik said:
i dunno about arguing the point of cell helping out rsx is good for ps3 for the matter.

Err...what?

Anyway, I'm gonna guess at what you're saying and ask..why is it "bad"?

If Cell can help RSX, and the dev wants to do that..why not?
 
gurgi said:
PS3 can do HDR+AA?
Though Vysez replied, I will caution that we don't know for sure if RSX is limited to SSAA or not. That's G70. But I was really trying to ask was that looking at the bandwidth figures, even if PS3 supported MSAA, it'd be squeezed enough that there wouldn't be room for fancy effects, or would there?

Just trying to understand what advances are and are not being made in graphics, looking at more than just the poly counts and shader lengths.
 
Titanio said:
dukmahsik said:
i dunno about arguing the point of cell helping out rsx is good for ps3 for the matter.

Err...what?

Anyway, I'm gonna guess at what you're saying and ask..why is it "bad"?

If Cell can help RSX, and the dev wants to do that..why not?

why would it need helping? what are the costs of helping?

it's very powerful enough as it is.
 
dukmahsik said:
why would it need helping? what are the costs of helping?

it's very powerful enough as it is.

This is true, but it's not like RSX represents the limit of ambition for graphics ;) I'm sure the more ambitious and enterprising devs will find ways to push things further, visually, by going beyond RSX so to speak.
 
dukmahsik said:
why would it need helping? what are the costs of helping?

it's very powerful enough as it is.

Alpha-Blending and using small Polygons without textures is costly in terms of Bandwith. RSX is certainly good with long Shaders and Textures, but aint got excellent Bandwith.

Each SPE on the other Hand has more Bandwith to their Localstore than RSX... so why not put the Bandwith-Intense tasks (blending) where they fit.
Question is: Can you actually make Particle-Systems operate on tiles fitting into SPE`s LS ?
 
Npl said:
dukmahsik said:
why would it need helping? what are the costs of helping?

it's very powerful enough as it is.

Alpha-Blending and using small Polygons without textures is costly in terms of Bandwith. RSX is certainly good with long Shaders and Textures, but aint got excellent Bandwith.

Each SPE on the other Hand has more Bandwith to their Localstore than RSX... so why not put the Bandwith-Intense tasks (blending) where they fit.
Question is: Can you actually make Particle-Systems operate on tiles fitting into SPE`s LS ?

I thought rsx has a large amount of bandwidth? i am beginning to hear more and more on how rsx will be bandwidth limited using high end effects at one time
 
Back
Top