[chaphack]

Alright, i think i recall why D3 shadows feel so much cooler than SH3! Shall i say D3 shadows are more..."dynamic"(!?!). You know those shadows dont just stay in boring predetermind places(ala SH3, off stationary chairs/cupboards etc), they have "!motion!" .

They are somewhat like SC's, but more reactive to objects/lights around ... at least that is what i recall right now...

[marconelly!]

Everything in SH3 throws, receives and has self shadows, except for some small objects in some scenes. It's not just static geometry, but also moving characters, including your own. I think that in your quick-to-dismiss-anything-on-ps2 state of mind you haven't even given that game a proper look. D3 uses normal maps which give it's geometry an extra boost - it appears as if every little nut and crain is properly lit, although it's just a texture detail. That gives it that more detailed and smooth shaded look that I think you are talking about.

As for SH3 camera, I don't know why you need a developer to tell you to press R2 and have camera move around with you in those larger open spaces in the game, just like it would in any other 3rd and 1st person game.

[_phil_]

Alright, i think i recall why D3 shadows feel so much cooler than SH3! Shall i say D3 shadows are more..."dynamic"(!?!). You know those shadows dont just stay in boring predetermind places(ala SH3, off stationary chairs/cupboards etc), they have "!motion!" .

SH3 provides moving shadows every time you walk in it with flashlight in hand (that is about 80% of time )

And the mater of coolness between the SH3 shadows and D3 shadows is opinion as i think to the contrary that blured SH3 shadows blends and feels in every way better .

[Simon F]

Originally Posted by Thowllly

No. It is not. I'm sorry to say, but you obviously don't have a clue how the lightning in Doom3 works.

The old way to draw shadows in games is to draw everything lit, then darken the things suposed to be in shadow. That is a hack.

Actually I think you should say "an old way".
IIRC what you describe is a suggested way of implementing shadow maps (not volumes). AFAICS, with volumes it's pretty much just as easy to add light as it is to remove it.

[chaphack]

Hmm, maybe it is because there are more sources of automated light/shadow motion in D3 demo, while SH3 depends soley on where you shine you torch...maybe that why D3 seems more impressive to me. You know, the shadows/lights are always in motion...so cooool.. 8)

[Simon F]

Originally Posted by MrWibble

Stencil shadows typically look sharp because they're a pixel-level effect. Each pixel can only be in shadow, or not in shadow.

In order to get "proper" soft shadows, the most accurate technique is to do a whole bunch of different stencil volumes projected from a selection of points inside the volume of the light-source (soft shadows are typically caused because light-sources are not single points, but light emited from a volume of space). You accumulate the result and blend between the lit and non-lit values based on how many volumes overlap.

That's exceedingly expensive even for a small number of samples, and you probably need quite a few to avoid banding effects.

At Siggraph (and Graphics Hardware 2003 ) there was a paper (+update for optimisation) that presented a way of doing soft shadows with the stencil buffer which didn't need the super-sampled light. It produced really good results and ran in realtime. Needed the latest Pixel Shaders though so, as far as consoles go, you'd only get it on the XBox.

[Simon F]

Originally Posted by PC-Engine

Actually DC and its modifier volumes or infinite planes technology were practically born to do stencil shadowing.

Indeed!!!

[MrWibble]

Originally Posted by chaphack

Ok, back to the shadows thingie...

Since we have a fair share of developers in this topic. Have to ask, just because one type of game. SH3(semi-fixed cam with loadings between small areas) did somewhat of shadows, is it fair to use that as a basis, that another type of game(maybe a full 3D fps with dynamic shadows) will achieve similarly satisfactory results, for all its intent?

Fixed camera is somewhat irrelevant, as what makes the shadows move around is either moving objects, or a moving light-source. In SH3 the lightsource is a torch in the player characters hand, and the shadows are plenty dynamic, even for the static background objects.

The upcoming online game "Hardware" has stencil shadows on everything in the game, they're unified into the main lighting model, and it all runs easily at 60fps. Dunno when it's out, but it's been on the online beta trial for a while. There was a talk at this years GDCE from a couple of Sony folk about how they optimised it.

It's no D3 (wtf do you expect?) but it certainly shows that the PS2 is more than capable of doing shadows. In actual fact, it's better at doing stencils than the XBox (IMO). Stencil shadows require large amounts of untextured fill-rate.... which is one of the few areas where the PS2 is faster than any of the competition.

I was speaking to a guy I know at another dev place a few weeks ago, who said that the entire PS2 rendering pipe (scene-geometry, stencils and all) ran faster than the XBox took just to do the stencils...

D3 ran like a pig on a fairly meaty P-IV machine with a high-end (at the time) GeForce. A machine which would certainly toast an XBox and therefore (according to most people here) would also toast a PS2. It's utterly unreasonable to expect anything like that level of engine on any of the current consoles. XBox can probably have a stab at it, but I pretty much guarantee it'll be heavily compromised.

[chaphack]

Thats a nice one from mrwibble. yup, i guess thats why D3 shadows impressed me more, D3 has cooler placed moving light sources compared to a boring old torch.

What do you think about Halo2? I mean its no D3 on full settings, but it isnt bad on mid-low settings. IIRC, HL2 Xbox will run at 30fps on mid settings.

Some pics of the elusive Hardware? game.






hmm...do those shadows go craaazzzzy with moving lightsources too? The rest of the graphics are poooootastic though..... :?

[cthellis42]

Well keep in mind it's heavily compromised on the PC as well. ^_^ I mean, if they're going to go down to initial GF's and Radeons and ~1ghz procs.

Best advice seems to be "get it for PC". At least you'll be able to push the best from the game whenever you get the equipment to.

[Panajev2001a]

Originally Posted by Simon F

Indeed!!!

I refuse to believe it until someone from IMG tech. doesn't help Dan Potter to add modifier volumes and DOT3 blending to KallistiOS and KGL.

:P

[Fafalada]

Originally Posted by MrWibble

The other problem (and for me it's a more serious one) is that you get odd "halo" effects around the shadows. The problem is that the stencil shadow is essentially in 2D once you've drawn it. If you just blur that, you get a nice soft edge to the shadows, but not all of them should look like that.

Well I've played around with this a fair bit, and the halo effect isn't really much of a problem when you tweak a bit (the filter used makes a difference too). With long draw distance small shadows will be more prone to aliasing though (as sampling error relative to shadow size grows). I've considered masking too but it didn't seem worth it :P
Either way, I much prefer the look you get over hard edge - basically a stepping stone until we can start using the better approximations like the one Simon mentioned.

Originally Posted by Chap

Since we have a fair share of developers in this topic. Have to ask, just because one type of game. SH3(semi-fixed cam with loadings between small areas) did somewhat of shadows, is it fair to use that as a basis, that another type of game(maybe a full 3D fps with dynamic shadows) will achieve similarly satisfactory results, for all its intent?

Of course not, SH3 is made entirely out of FMV, and characters are animated sprites. Shadows look blocky because of movie compression, the entire game is one big pre-rendered clip.

Anyway, certain talk about PS2 rendering these things is based on actual code we wrote and bloody benchmarks some of us ran you know. In game measurements even...

Oh btw, to give you more food to troll, the screens of the game you posted don't even selfshadow, so not only look terrible pooh poohy they are inferior too yeah. And where's the shiney shine?

[notAFanB]

Does anyone know if SH3 handles overlapping shadows correctly? From what I can see D3 seems to handle them correctly by the methods Thwolly described

don't know, probably the engine 'could' do so but the title in question (SH3) doesn't do so in any noticable way.

Simon F wrote:
PC-Engine wrote:
Actually DC and its modifier volumes or infinite planes technology were practically born to do stencil shadowing.

Indeed!!!

it is a pity that the only 'real' alternative to IMR's (is that the correct term) that was workable is not a big player in PC space.

yup, i guess thats why D3 shadows impressed me more, D3 has cooler placed moving light sources compared to a boring old torch.

apart from JC's implementation on his engine they are very similar.

tencil shadows on everything in the game, they're unified into the main lighting model, and it all runs easily at 60fps

any idea how much geometry they are pushing in game?

Anyway, certain talk about PS2 rendering these things is based on actual code we wrote and bloody benchmarks some of us ran you know. In game measurements even...

what is the stencil performence like btw just curious?

[Paul]

haha! I know what are you trying to drive at. But it is not going to work. PS2 poor IQ IS of comparable concern, it IS the worst amongst next gen systems. The truth is for all to see. Sadly Xbox 30fps is not of a major concern(paul hits head on Da wAll! ). If it is, then PS2/GC/DC all are the 30fpsDOOMED! Simple as that.

FPS and IQ are opinion, not fact.

I love how you try to downplay Xbox's high end 30fps problem, because it IS a problem for many people who love high frame rates just like PS2 IQ is a problem to you because you personally love clean IQ.

Fact: A majority of high end Xbox games run at 30fps and to many this IS a problem. The end.

Funny, you say Xbox 30fps isn't a problem, but if anyone says PS2 IQ is not a problem you'll jump on them. A strange bias indeed. Oh and no, most ps2/gc high end games run at 60hz FYI.

[notAFanB]

Fact: A majority of high end Xbox games run at 30fps and to many this IS a problem. The end.

to many this is acceptable in many titles.

Fact: Ps2 IQ is THE WORST amongst the 4 systems

nor is it as bad as you imply in your posts among some titles.

Since we have a fair share of developers in this topic. Have to ask, just because one type of game. SH3(semi-fixed cam with loadings between small areas) did somewhat of shadows, is it fair to use that as a basis, that another type of game(maybe a full 3D fps with dynamic shadows) will achieve similarly satisfactory results, for all its intent?

your overgeneralising shadow implementation from a technical pov. would it make any difference if stencilling runs at acceptable rates (still don't have exact figues for this) on PS2 or does that earn more distain?

[randycat99]

Speaking of SH3 shadowing and lighting, anybody recall what the resultant shadowing and lighting from the flash of gunfire created? It never occurred to me to check that out while I had the game under rental.

For the record, I thought the game looked impressive and can only fathom what could be visually achieved with that engine under other gaming themes...

[randycat99]

DP

Uhhh, how can you have a page 7 of 6 pages? Does page 7 really exist or not? Something funky with these forums definitely going on...

[DeathKnight]

I'm talking about that many high end Xbox games run at 30fps.

In most cases the developer chooses to push the graphics to the point where a locked 30fps is the only feasible choice (fluctuating framerates between 30 and 60 are not good). This can happen on any of the systems. Obviously the Xbox has a higher threshold overall than the other systems (the height depending largely on the developer as well). Ninja Gaiden is absolutely beautiful and that runs at a rock-solid 60fps. RSC2 is arguably one of the best looking racers ever and that runs at a rock-solid 60fps.

The other two systems are no less bandwidth limited than the Xbox. They're just bandwidth limited in different ways. All three systems also function differently in their way of handling the graphics pipeline. Both the PS2 and Cube use huge caches to offset smaller system memory size and smaller bandwidth to that memory. The Xbox doesn't rely on a huge cache because it doesn't have the smaller system memoy size and smaller bandwidth to that memory. It still has an insanely fast cache array (either quad or triple of vertex and pixel) with roughly 32GB/sec of bandwidth (that's enough to feed the entire pipeline full speed ahead).

I really don't think it's a problem (developers don't seem to think it's really a problem either). You're trying to make it a problem as a last defense sort of thing. That don't float.... Sure, the Xbox isn't without limitations just like the other systems, but it's surely a lot less limited than them.

[Vince]

Originally Posted by Fafalada

Of course not, SH3 is made entirely out of FMV, and characters are animated sprites. Shadows look blocky because of movie compression, the entire game is one big pre-rendered clip.

Anyway, certain talk about PS2 rendering these things is based on actual code we wrote and bloody benchmarks some of us ran you know. In game measurements even...

Oh btw, to give you more food to troll, the screens of the game you posted don't even selfshadow, so not only look terrible pooh poohy they are inferior too yeah. And where's the shiney shine?

HAHA! This is perhaps the greatest utterance that mankind has yet achieved. Faf, you never cease to amaze me.

[randycat99]

Originally Posted by DeathKnight

In most cases the developer chooses to push the graphics to the point where a locked 30fps is the only feasible choice (fluctuating framerates between 30 and 60 are not good).

That strikes me as a bit self-contradictory. If it fluctuates between 30 and 60, then that means at its worse it will look as "jittery" as something that is running 30 all the time. So nothing was really lost. If you are running at "jittery 30" all of the time, that would seem to be the worst case situation (the Xbox situation, if you will ). "Fluctuations" are highly overrated, IMO. It's when they dip below 30 or more for more than a few frames is when people should be concerned, and that can happen to any system if the load is too great- locked or not.

[DeathKnight]

Going by your way of thinking randycat then why don't developers lock games at odd rates between 30 and 60?

It's better to have a signal sync'd with a TV's refresh rate, just like a monitor. Anything over causes tearing and/or anomalies. Analog televisions with an interlaced signal refresh at 30Hz. Progressive scan televisions can refresh at 60Hz. Developers tend to stick to one or the other for obvious reasons.

[randycat99]

Originally Posted by DeathKnight

Going by your way of thinking randycat then why don't developers lock games at odd rates between 30 and 60?

They don't need to. Pick the obvious rates. That hardly serves as proof of anything.

Anything over causes tearing and/or anomalies.

That's arguable. Stuff like that has more to do with how the frame buffers are implemented.

Analog televisions with an interlaced signal refresh at 30Hz.

Actually they "refresh" at 60. If you are concerned with how often an entire frame is refreshed, then that would be zero, because they simply don't work that way.

Progressive scan televisions can refresh at 60Hz. Developers tend to stick to one or the other for obvious reasons.

P-scan televisions show an entire frame at 60 Hz. So we have 60 Hz refresh in either scenario.

...but all this is an unnecessary departure from the topic. Just concede that it isn't as simple as you had originally asserted.

[DeathKnight]

Originally Posted by randycat99

They don't need to. Pick the obvious rates. That hardly serves as proof of anything.

Simply a question posed to your awkward way of thinking. You implied that fluctuations between 30 and 60 aren't a problem. Therefore I responded with the question of why don't developers lock refreshrates in between 30 and 60 (barring it's feasible) if it's not a problem?

That's arguable. Stuff like that has more to do with how the frame buffers are implemented.

It's not arguable. It'll behave just like the monitor you're looking at. Try running games at an fps that exceeds your refresh rate. You'll get tearing. Why do you think there's a little thing called v-sync?

Actually they "refresh" at 60. If you are concerned with how often an entire frame is refreshed, then that would be zero, because they simply don't work that way.

No, analog televisions refresh at 30Hz. They follow the NTSC standard of a 30Hz refresh rate (it's actually 29.97Hz from the source, but the TV refreshes at 30Hz).

P-scan televisions show an entire frame at 60 Hz. So we have 60 Hz refresh in either scenario.

Not either scenario. Only with progressive scan. A 60Hz progressive signal will be sampled down into a 30Hz interlaced signal for analog televisions (in the case of the consoles this is in the video encoding chip).

...but all this is an unnecessary departure from the topic. Just concede that it isn't as simple as you had originally asserted.

The only thing to concede here is that you're spewing a bunch of bullcrap.

[randycat99]

Originally Posted by DeathKnight

It's not arguable. It'll behave just like the monitor you're looking at. Try running games at an fps that exceeds your refresh rate. You'll get tearing.

This will be contingent on the findings below.

No, analog televisions refresh at 30Hz. They follow the NTSC standard of a 30Hz refresh rate (it's actually 29.97Hz from the source, but the TV refreshes at 30Hz).

A full frame occurs effectively at 30 Hz (but that is a pretty abstract association given the nature of interlaced operation). Each field comes in at 60, however. A field on an interlaced display would constitute a "refresh".

So that leaves both types of TV's with an effective 60 Hz refresh in their own way. So your above comment doesn't really apply since the console isn't outputing a rate that exceeds either setup.

In fact, tearing can occur when the console is rendering reduced framerates due to load, if the frame buffers are setup in such a manner. I've seen it happen on A early PS2 game (it wasn't even hitting 30 fps, I'd estimate). I haven't seen tearing in any game since.

[DeathKnight]

Bah. You're right. Sorry for being so harsh to begin with. It's a bit late (almost 2 in the morning here) and I'm getting edgy Would appear I can't think straight either. No hard feelings eh?