News PGR3: Textured, not stirred

No, no. I like hearing about PGR3! It does look mighty pretty. If I get a 360, I'll be happy to get my ass handed to me as a hat by you. I'm used to GT, so it'll be a hard switch.

Personally, though, I'm still going to wait on both consoles to see which one is better for me.

 

Video cards are still fillrate bound, so, no, you can't have huge poly counts when adding a PPU, because a videocard can only render so many triangles before it starts to affect the performance. Texture memory would still have to be huge because we don't want to see all the buildings textured the same, now do we?

Also, I don't see why you would want everything to be built out with actual polys instead of high res textures. I don't care if a wall is a texture or poly soup, as long as it looks like a wall, and makes the correct sound when I shoot it or run into it. Adding more polys means more unneccessary collision checks, more polys to check to for lighting and more polys to render.

In terms of deformable terrain the main issue isn't being able to render the worlds, it is the AI and the design of the game. If you can shoot down any wall you want how would the game designer be able to predict where you might go? He would then need to account for every case and it would be a nightmare to change any aspect of that design. I would still like to see deformable environments used more, so that a room looks like I had a gunfight in it when I am done.

In relation to the topic, the screenshots are amazing. I don't know why everyone is very "meh" about PGR3. The videos of this game are underwear ruining, screenshots just don't do them justice. It is a racing game, it should have a lot of good driving mechanics, good looking / well laid out environments, and AI that doesn't cheat, that is it.

 

Respect quickresponse!! You actually registered to reply to this thread Woot!!

I agree with you partially, but then when come up to the textured wall from side, it looks like weird looking blurry thing.

 

Thanks. I agree that the flat wall, no matter how good the bumpmapping will still look flat if you are looking down the wall. That is one of the things that I can accept because in the context of the game that doens't take me out of the experience. Chances are if you have time to stop and look down a wall then the game isn't very appealing or is overly appealing. I think that made sense.

 

First of all, Welcome to the forums!!!! Always glad to have a new bit-techer among us!

Second of all, I agree with you minorly on some points, but very little as a whole. The entire idea of "Fill rate" is largely due to the fact that a single unique texture is "skinned" across the polys (which requires lining up, etc) then modified through pixel processing/post processing. It sounds like you understand a bit of what's going on under the hood, but you're missing a few big points.

Pixel transformation is a huge part of the fillrate process. Because the graphics card is running its own "fake" light, approximated by algorithms. Replace these with a raytraced light source from a PPU, you just reduced the amount of calcuations by half or more that need to be done on the GPU die. Let a single chip handle the falloff equations, you have even less work. In fact, the most work will then be done by the culling and geometry section.

As for textures, you wouldn't need the same amount of memory. You're not loading 1024x1024 in anymore, you're loading a much smaller amount (we'll say 30x30 for a BS target). Enough to get the actual *TEXTURE* of the object. This can be tiled (as opposed to stretched) over the object in accordance with its wireframe. Meaning that there only needs to be one copy in memory, referenced many times on each screenshot. Big difference from what you're talking about. The memory required would be enough to hold the references and to hold a completed frame for its AA.

The eventual perfection would be wireframing a brick down to every little chip in its surface, every dimple and rise in its texture, then painting it with a flat color (w/a nod to diffuseness, specular highlight, etc), "brick." If broken randomly, every splinter would retain its specific design, rather than the tex "skinning" incorrectly. The wireframe would allow perfect shadows from a raytraced lightsource, across every dimple.

The difference this would make in PGR3 is minimal, I grant. But that continuing evolution is one that should be strived for. So that if you crash your car, your car crumples based on the impact on the detailed wireframe (physics calculated at time), and shatters the glass on the first couple buildings you hit.

Once you get to games like Half-Life, it would make a huge difference. But for now, the textures on walls look "painted on," from the electrical outlets to the crown moulding. Imagine being able to shoot the plate off an electrical outlet, or pulling the chair-rail off with a grenade blast, or breaking the *frame* of a window out with your crowbar to open it that much more...

So, that's kind of where I'm coming from. As I said, I didn't say it wasn't pretty. I only said that it is not technologically amazing in any way. It's just more of the same, the process refined a little bit further. I'd like to see the process redefined to go towards more realism instead of more "faking" reality.

 

*dozes off into dreams about games in 2010*



*wakes up*

...Who would pay those guys for making that detailed wireframes though? Would the games take 10 years in making?

Hey but I just got an idea thinking about that brick of yours. You could use that brick in alsmost any game. Thus game companies could create libraries of various high detailed objects and then just use them to build environments to various games. However that kind of library would take damn long and expensive to create. Who would fund it? Amazing investments should be made. Also another big problem arises. Whenether someone tries to unify something in computing world mess comes out (DVD+, DVD- etc.) Companies would probably want to... Another idea hits Cheap Mod Wannabe's head... Wait wait wait. Why the F*** would somene make wireframes like that? That is too hard/timeconsuming. We could just create small particles.... let's say small sand like particles. We make let's say a square wireframe and then just fill it with those small particles. Some coding... not too difficult. We call that a cell technology and we make millions... we make small longer pieces and fill a low-Q wireframe of a tree with them facing the end if branches... awesome! Code some interactions between each particles and then save objects into libraries. One company makes just objects while game/architecture/design and other companies buy certain object (or order specific objects) and then use those objects to create inateractive-virtualenvironments..... =) So when do we start Da Dego?

*run's off to the store to buy Patenting for Noobs book*

 

You know, you may have just hit on something, Cheap! In each object you "sandbox" (because you're filling a frame with particles) you could specify a level of coherence. That would make things function when their frames are ruptured more like liquids, powders, or splintering solids, all defined by your coherence factor.

We're dangerous, cheap. If we keep this stuff up, we'll come up with some great ideas to never be technologically implemented. Because companies are too cheap, and we're waaaay too lazy. But still, I think there might be some viable ideas here!

 

Woot woot imagine: BitTech Systems present to you.... The next generation graphics/physics engine: CellTech


11:40 pm CMW thinks: I need some to rest a bit from that planning. No more than
5mins and then I'll be back sketching...

11:45 pm *goes to the fridge, gets a beer and cheese*

12:05 am *falls asleep watching the last samurai on a couch*

 

I skimmed through the post here and find it quite sad that people are so hyped about this ... yes it looks nice, it has superb lighting and that is what makes the difference...but... that texture IS a photo (Marlin downtown resource CD www.marlinstudios.com/products/st5/st5.htm )... as are everything else in the game... it's photoreal because they ARE using Photos of the locations needed. That in all honesty does not take a huge amount of skill to do or produce.

In fact just by looking at the sample given on this site you can tell..the detail in one part of the decor up at teh top of the phot starts to face each other (lenz warp) ... these are not hand painted in anyway.. yes they will be edited to reduce seams etc. But at the end of the day it's still just a high quality 1024 photo slapped onto the side of a poly..and that does not impress me one bit as far as skill or technical ability goes.

What impresses me is there lighting, it's nice.

K.

 

Yeah I agree kaylon, but I think the point is that now technology has come so far as to give us the massive amounts of memory and processing power needed to use 1024x1024 photos in games. We could still use photos in today's games but when shrunk to 128x128 the detail is lost so it looks better sometimes to have an artist draw it.

To me texture quality makes a big difference for how the game looks. I hate it when you move up close to something or zoom in with your binoculars and all you see is a blur. High res photos should reduce that hopefully.

But yeah even so, for me I don't think that's the most important thing. Lighting certainly plays a big part second probably to the designers' attention to detail in architecture and models.