What games have exploited consoles' graphics the most?

[Fafalada]

Anyway, PS2 lends itself to such engines due to its powerful vector processors, but GCs T&L can't handle flexible polygon meshes, it would probably be a huge hit on the CPU to have it do that by itself.

Commonly, LOD refers to having multiple resolution models that you then swap as needed, and that hardly takes any great computation power (except from the artist that must create mode models ).
Flipper also has a flexible display list format that naturally lends itself to this kind of use.

you don't have a (per-pixel) normal map in the equation, so you don't have a pixel-level granularity (or near so) of the lighting on the surface side.

Well you might as well use a different term then, because I sure as heck have seen evaluation of complex equations per pixel without ever using normal map inputs.
Or are you arguing that inputs or even just equation elements define the frequency of the evaluation?

 

[archie4oz]

Commonly, LOD refers to having multiple resolution models that you then swap as needed, and that hardly takes any great computation power (except from the artist that must create mode models ).
Flipper also has a flexible display list format that naturally lends itself to this kind of use.

Yeah, I'm rather surprised how many games are still fussing with computing LOD at run-time these days when it's much easier to whack the artists on the head to provide you with your model instances (hell many create them anyways when building the base model, all you have to do is re-channel those efforts).

 

[Panajev2001a]

Isn't it better to use Subdivision surfaces and NURBS and do dynamic LOD in real-time ?

 

[marconelly!]

In any Quake game, did you ever look at the walls/floor when you shot a rocket down a hallway?

I have about 1 hour of playtime in all three quakes combined, so I'm really not an authority on the game. Honestly, I just assumed that was vertex lighting for rocket.

In any case, if the lighting in games I mentioned should not be called per-pixel, so shouldn't lighting in PC/XB Silent Hill 2. Furthermore, I'd like to hear what would be the proper name for such lighting.

 

[zurich]

In any case, if the lighting in games I mentioned should not be called per-pixel, so shouldn't lighting in PC/XB Silent Hill 2. Furthermore, I'd like to hear what would be the proper name for such lighting.

Well i'm no programmer, but the lightmaps seen in Q1/2/3 when firing a rocket down the hall are a hell of a lot different (and more primitive) than the flashlight in XB SH2/Halo.

 

[marconelly!]

Another game that uses simillar flashlight is the Ghost Hunter - it has the same look as the XB/PC SH2 flashlight or MGS2 flashlight.

Halo used that too, but it also had bumpmapped surfaces that could get lit, so I guess it's out of question.

 

[darkblu]

Well you might as well use a different term then, because I sure as heck have seen evaluation of complex equations per pixel without ever using normal map inputs.

normal maps here are just a representation means, of course their presence is not mandatory to the 'per-pixelness' of the illumination.

now, in my dictionary per-pixel illumination has two components: pixel-level granularity when calculating the incident vector (i.e. on the light side) and pixel-level granularity when calculating the reflection vector (i.e. on the surface side). light maps can provide only the former, unless you somehow come up with light-maps pre-calculated such as to take into account the surface roughness characteristics, but that'd be rather impractical, IMO.

Or are you arguing that inputs or even just equation elements define the frequency of the evaluation?

the frequency of the evaluation is set. but the input/parameters delimit the maximal representable output frequencies. as in the example at hand: with normal maps you get per-pixel granularity for the reflection, as opposed to surface-interpolated normals which are incapable of representing reflection characteristics such as material roughness (unless of course micro polies are used, but i believe it's beyond our topic)

ok, i realize all we've been discussing here is to a great extent a matter of semantics and definitions. but to get back to the crux of this discussion, light mapping has been know and used in game titles since the very first days of voodoo graphics on the pc scene. no need to point it out 9 years later.

 

[darkblu]

In any case, if the lighting in games I mentioned should not be called per-pixel, so shouldn't lighting in PC/XB Silent Hill 2. Furthermore, I'd like to hear what would be the proper name for such lighting.

lightmaps

 

[marconelly!]

I see what you mean, but woudn't those be 'dynamic lightmaps' (as opposed to 'static lightmaps') or something like that, then? Furthermore, what if the surface doesn't need a normal map by it's design? If it's meant to be perfectly smooth, without any pixel-size bumps. Does that, by your definition, mean that it's impossible to have 'per pixel lighting' on such surface?

 

[darkblu]

I see what you mean, but woudn't those be 'dynamic lightmaps' (as opposed to 'static lightmaps') or something like that, then?

well, there could be devised such a distinction. in general, the more realistic lightmaps are dynamic at least in the aspect that their projection on the surface is caclulated dynamically. AAMOF properly dynamically-projected lightmaps can look very realistic, in contrast to, say, the very rudimentary use of flat-on 'hotspots' on a wall.

Furthermore, what if the surface doesn't need a normal map by it's design? If it's meant to be perfectly smooth, without any pixel-size bumps. Does that, by your definition, mean that it's impossible to have 'per pixel lighting' on such surface?

a perfectly-smooth surface is a special case of the general definition of a surface of roughness 'n'. in the Phong illumination model micro roughness is handled as the diffuse factor of the equation, and such a micro roughness can often be approximated on a macro level. but then comes specularity *and* pixel-level roughness, for which you need some form of control over the reflection vector. now, if a given lighting technique can handle only the special case of a perfectly-smooth surface, or one which exhibits only micro-roughness, then a question about its per-pixel correctness arises. one that everybody can answer for him/herself 8)

 

[marconelly!]

Well, all I'm saying is that by that definition, games that go after certain visual styles (say Zelda WW or Rez) that simply don't need any normal maps usage - those games couldn't be said to have per pixel lighting no matter what they do. That just seems kinda restrictive and even illogical,

 

[London Geezer]

it's all about definitions...... as thingy said, it's up to u to call it whatever u like given the circumstances...

 

[darkblu]

Well, all I'm saying is that by that definition, games that go after certain visual styles (say Zelda WW or Rez) that simply don't need any normal maps usage - those games couldn't be said to have per pixel lighting no matter what they do. That just seems kinda restrictive and even illogical,

/note/ as i already mentioned, normal maps are just a means of representation, pixel-level granularity reflections are not inherently-bound to normal maps. /eon/

the per-pixel _lighting_ we've been discussing with you so far is a means to achieve photo-realism. now, i don't recall either wind waker or rez to have had strong attempts at that, do you? they may have employed other various per-pixel _shading_ techniques, but not realisitc per-pixel lighting. see the difference?

 

[marconelly!]

the per-pixel _lighting_ we've been discussing with you so far is a means to achieve photo-realism.

So that kind of lighting is now only reserved for 'photorealistic' looking games? :\ Toon-shaded games cannot be said to have per pixel lighting, even if they don't use vertex lighting and the lighting is pixel-correct? What kind of lighting should I say they have, then? Who is the authority making these naming conventions, btw?

As I've said, the whole ordeal of naming the *lighting type* by the type of *surface material* that light is cast onto seems kinda illogical to me.

But I guess to get back on topic:

but to get back to the crux of this discussion, light mapping has been know and used in game titles since the very first days of voodoo graphics on the pc scene. no need to point it out 9 years later.

Well, if you'd rather not point the type of light used in those games, be my guest, but still very few games have that neat looking flashlight effect, so why not point it out?

 

[darkblu]

the per-pixel _lighting_ we've been discussing with you so far is a means to achieve photo-realism.

So that kind of lighting is now only reserved for 'photorealistic' looking games? :\ Toon-shaded games cannot be said to have per pixel lighting, even if they don't use vertex lighting and the lighting is pixel-correct?

what makes them 'pixel-correct'? give me your defininition of 'pixel-correct' (i've already given you mine).

What kind of lighting should I say they have, then? Who is the authority making these naming conventions, btw?

nobody in particular, they arise from the field of the art. btw, i can recommend some very good literature to you, if you wish.

As I've said, the whole ordeal of naming the *lighting type* by the type of *surface material* that light is cast onto seems kinda illogical to me.

you're misinterpreting what i said. it doesn't come from 'the type of surface material'. surface roughness is a general propery of _any_ surface.

i hope the fact that illumination is a phenomenon produced by the interaction of light radiation and a surface (i.e. its properties, incl. roughness) does seem logical to you, no?

Well, if you'd rather not point the type of light used in those games, be my guest, but still very few games have that neat looking flashlight effect, so why not point it out?

what's so hard in calling it 'projected light maps'? in many people's minds that's a well-established technique, and they would get your point right away.

 

[Ug Lee]

btw, i can recommend some very good literature to you, if you wish.

I'll take you up on those recommendations. 8)

 

[marconelly!]

what's so hard in calling it 'projected light maps'? in many people's minds that's a well-established technique, and they would get your point right away.

Isn't that actually 'dynamically projected light maps'? In that case, what's stopping me to use 'dynamically projected lightmaps with normal-mapped surfaces'? It just sounds clunky.

 

[Panajev2001a]

Complex... a waste of time... Just go with micro-polygons and be happy...

Seriously, I understand more and more why Sony is rumored to be experimenting with a mciro-pokygon based rednering paradigm with PlayStation 3... it simplifies a lot of things, it is more organized, logical and elegant than the mess of per-pixel, sub-pixel, per-polygon mix we have now... also the rumours of the PlayStation 3 CPU being a parallel processing monster and with tons of bandwidth to spare kind of help supporting those rumours ( PlayStation 2's approach was in that kind of direction already... very fast and flexible T&L engine and a streamlined rasterizer VERY fast at rendering small triangles in LARGE quantities... GSCube followed on the micro-polygon approach )...

 

[darkblu]

I'll take you up on those recommendations. 8)

"advanced animation and rendering techniques" father and son Watt
http://www.amazon.com/exec/obidos/t..._1_1/002-6738532-5479249?v=glance&s=books

this book gives very elaborate accounts on most of the techniques used for photo-realistic (and not only) rendition circa the early/mid '90s. the lighting section is particularly comprehensive. surprisingly, it also has a good intro on general 3d theory (although one may wonder what its place is in such a book). the overall style of the authors is very readable, despite the complexity of the topics.