What about next-gen particles?

PC-Engine said:
:?: You got all of that from watching that crappy quality video? The only inkling of rain I saw was at the beginning with the dragon's eye, but it didn't look impressive at all...


Sounds like you're assuming a lot of stuff from just looking at the pictures...

Err, no. I watched the high-res trailer (in the folder linked to above). How would I get any of that, sans perhaps the density of the rain, from still pictures? The video is crappy quality, but you can see all that stuff in it. The eye stuff is obvious, you can see the "pitter-patter" of the rain off the stone when the dragon does is first "roooar" looking up toward the camera (kind of hard to describe this effect...basically whatever stone looks like when it's raining heavily on it ;)), and the heavy rain in the air is..obvious.

Maybe we've different eyes for this kind of thing, but I guess I'll say that it looks very impressive. I'd like direct-feed media to clarify that 100%, certainly, but the despite its quality, the video doesn't exactly suggest that the rain "didn't look impressive at all", IMO :)
 
the video doesn't exactly suggest that the rain "didn't look impressive at all", IMO

Well my comment was about the rain on the eyelid which didn't look impressive at all. The rest of the stuff that you could actually see has already been done in games like MGS2 and RE4.
 
PC-Engine said:
Well my comment was about the rain on the eyelid

Did you see the rain roll off the eyeball itself? It's reminscint of windshield effects in some driving games, but better quality and not so flat.

PC-Engine said:
The rest of the stuff that you could actually see has already been done in games like MGS2 and RE4.

The rain in those games looks great, particularly in the latter. Maybe the general level of rendering in Lair is tricking me, but I'm not sure if they're quite up to the same standard from what I can make out. I look forward to better media..

Though if we're reaching a point of diminishing returns - as far as this particular type of particle system is concerned, at least - it suggests some things about the importance of this debate at least as far as rendering goes. Behaviour and simulation is another issue.
 
london-boy said:
mj77 said:
Hi everybody i am new here. I figured I'd post this here. It's a high res of the Lair trailer.

http://upsilandre.free.fr/videos/lair_072105_qthigh.mov

Flippin heck, that looks amazing. If that's all realtime, i'm happy with the next gen, considering this will be an early game.

What part if it did you think looked amazing? The rain? The modeling and the number of dragons is what one would expect nextgen.
 
PC-Engine said:
What part if it did you think looked amazing? The rain? The modeling and the number of dragons is what one would expect nextgen.

One might expect a lot from nextgen, but what one would appear to be getting currently with many "nextgen" games might not meet expectations. So when a game comes along that does look like what you'd expect a nextgen game to look like, it may be cause for some small celebration ;)
 
Titanio said:
PC-Engine said:
What part if it did you think looked amazing? The rain? The modeling and the number of dragons is what one would expect nextgen.

One might expect a lot from nextgen, but what one would appear to be getting currently with many "nextgen" games might not meet expectations. So when a game comes along that does look like what you'd expect a nextgen game to look like, it may be cause for some small celebration ;)

Sure, but is it because of the rain? ;)

We've seen the hundreds of characters onscreen from nextgen games already and it's getting pretty old. Adding a layer of rain isn't going to make it more impressive that is unless the rain moves with the wind currents based on physics. ;)
 
PC-Engine said:
We've seen the hundreds of characters onscreen from nextgen games already and it's getting pretty old.

If you compare the quality of the models typically used with this technique vs those in Lair, there is a difference IMO ;) But admittedly we do not know how many Lair will be dealing with simultaneously.

Simply, though, I am impressed by it because of the models used (their detail and quality is not typical of next-gen games I've seen thusfar) - the muscle system has a neat effect - the lighting looks very good and is giving everything a nice "prerendered" look (and I've seen many express scepticism at its realtime nature because of that), and yeah, the atmospheric/weather effects look really good to me. Just my opinion, you know..you're welcome to yours, I doubt I'd question you as persistently on it.

Anyway, getting this thread back on track - since we're talking about particle effects - and a question for Barbarian or Npl or Faf etc. re. the theory of putting alpha blending on Cell - might using additive blending or modulate blending negate the need to sort the particles? Could save a lot of time..
 
If you compare the quality of the models typically used with this technique vs those in Lair, there is a difference IMO

The one dragon looks good, but when they show a bunch of them onscreen you lose all the detail and the others are so far in the distance there's no detail at all...

As to nextgen particle effects, show me rain that moves with the wind currents a real physics based rain to model interactions.
 
There are a lot of things currently done this gen. However, they cannot be done nearly as well as the next-gen systems. So 100 models on screen has been done before. 100 models on screen of cg-quality has not. Particle rain has been done before. Convincing particle rain has not. And so on. It's about quality, not quantity. Lair looked like the coolest thing shown btw. I'd like to see direct-feed video of...ANYTHING from that PS3 presser. So much quality is lost with these camcorder vids, no matter what resolution they encode at. PEACE.
 
So 100 models on screen has been done before. 100 models on screen of cg-quality has not.

And that Lair trailer doesn't show 100 cg-quality models either. Oh and when I said done before I'm talking about games like HS and 99 nights not current gen games.
 
Titanio said:
Anyway, getting this thread back on track - since we're talking about particle effects - and a question for Barbarian or Npl or Faf etc. re. the theory of putting alpha blending on Cell - might using additive blending or modulate blending negate the need to sort the particles? Could save a lot of time..

I initially had thought about smoke, order aint got a meaning if all particles have the same color and no extend, you could go as far and just counting how many particles fall in a (sub-)pixel and then do the blending( or more advanced processing ) in one step using a Pixelshader. I would only use the SPE approach if the Psrticles are reasonable dense and small.

both additive and multiplicative blending (thats modulate, right?) are sort-independant, the Question is: does it look "real" enough, especially if you have lots of overdraw? I guess not.
 
Titanio said:
Anyway, getting this thread back on track - since we're talking about particle effects - and a question for Barbarian or Npl or Faf etc. re. the theory of putting alpha blending on Cell - might using additive blending or modulate blending negate the need to sort the particles? Could save a lot of time..
One thing I was contemplating was raycasting particle volumes - this is in my half waking state mind you so I haven't given efficiency much thought yet. But you could avoid a great deal of overdraw with certain particle types, and there's basically nothing to sort.
(speaking of - rendering this kind of stuff is easy to tile - and you don't necesserily have to render these effects at full screen resolution either).

Resolving sorting between different substances is a whole other issue though (eg. smoke with fire) - perhaps depth based blending could be used for this but I honestly have no clue how good or bad that would look right now.
 
Npl said:
I initially had thought about smoke, order aint got a meaning if all particles have the same color and no extend, you could go as far and just counting how many particles fall in a (sub-)pixel and then do the blending( or more advanced processing ) in one step using a Pixelshader. I would only use the SPE approach if the Psrticles are reasonable dense and small.

Yeah, I figured this would be a good approach for small dense systems. I think up to a certain point, keeping it on the GPU would be fine..

Npl said:
both additive and multiplicative blending (thats modulate, right?)

Yeah, that's modulate. It'd certainly be cheaper, but I'm not too familiar with the visual side effects. I'm guessing for some systems it work better than others..some not at all, perhaps.

Fafalada said:
Resolving sorting between different substances is a whole other issue though (eg. smoke with fire) - perhaps depth based blending could be used for this but I honestly have no clue how good or bad that would look right now.

Yeah, I've only been considering systems of particles of the same type till now, hehe. Throwing another type into the mix makes things interesting. I've not heard of depth-based blending, but I'll have to look into it..

It's an interesting idea, for sure..generating particles, transforming and rendering all within the CPU is a compelling end goal, if it can work. My biggest worry is the rasterisation, although that could be parallelised obviously, and in a number of different ways (but then you need more SPEs). Staggering updates over frames could be interesting too, if rasterisation performance is problematic..

Thanks to both of you for your feedback!
 
Additive and modulate blends are order independent and there are some effects that can use them quite well. The problem of course is that with additive you saturate fast to white, and with modulate you saturate fast to black.
It becomes quite apparent when you have let's say, a smoke trail, and you look down along the path, you get this very unrealistic 'glowing' in the middle.
Some games sort per particle system, and as long as each system is only using additive or modulate, and doesn't overlap (much) with another system, you're fine.
When using full blend modes, it becomes more problematic. For things like clouds, you can have static sorting, and keep the systam ordered top to bottom (since we assume you'll always look at the clouds from below).
So obviously there are special cases, and some games don't even bother. In a fast exposion effect lets say, most people wouldn't notice a sorting discrepancy.
 
So obviously there are special cases, and some games don't even bother. In a fast exposion effect lets say, most people wouldn't notice a sorting discrepancy

Actually I never bothered with sorting, except for boids (where you need some structure to find your k-neares neighbours).

Alphablening was finde, too. Even for smoke and water effects. Regarding you just have to use dark textures and low alpha values. For water those white appearing here and there can be interpreted as caustic/lens and even helps the visual appearence.

fire.jpg

water.jpg
 
Back
Top