Crackdown: shader simulates raytracing Tech

Is'nt Lair on PS3 churning out at 10 texture layer's?

 

Yeah, you can see lots of slowdown in this recent vid, and there probably isn't much hope of the release being better as about mid way though the video the developer showing the game off claims "we are completely locked at 30fps at all times."

 

Is that the best barometer though? Being a video from X06 (recent?), its a full 5 months before the actual release date. Did you stream or it DL it? (i'm DL'ing the big one now...)

 

According to BlimBlim who has been playing a 1month old debug build:

 

I downloaded it, and yeah it's an older video but I'm pretty sure that is the lead developer doing the talking and if he calls that "completely locked at 30fps" then I can't rightly hope they have been doing much to improve the framerate issues since then.

 

That video appears to be co-op. Is it possible we could see slowdown when playing cooperatively, but not when playing the game by ourselves?

Because the frame rate in latest videos looks very smooth to me.

http://media.xbox360.ign.com/media/773/773714/vids_1.html


Anyways, we will get to see how the frame rate is in 5 days with the demo.

 

The first part of the video is single player, later in the video they switch to coop, framerate issues appear in both game modes.

 

I should note that 14 is actual material-specific -- it's also the maximum we can support because we need to set aside texture arrays for shadow maps. And yeah, we're basically hitting PC+Xenon+PS3. Well, I can't say that's a common thing. I think the biggest complexity stuff is the transitional components where we're constructing little bits of geometry that effectively blend between two materials and alleviate hard edges in terrain. So the number of pixels filled for these complex materials isn't so bad.

Major characters probably get materials with some 10 layers or so... again, mainly because of all sorts of complex blends between layers at different scales. We also tend to be able to save a lot of space and bandwidth with these types of things since a main diffuse map could be 1024x1024, but then have a detail map which uses the same UVs scaled many times over, but be a 128x128 map and the effective image resolution is still equivalent to using 1 4096x4096 texture.

Well, to generate those maps you need something to be modeled and textured in the first place. You just don't need to draw the actual geometry or process any of it at runtime. And the thing is if they've got all that geometry on the outside, then it is kind of a load to render real geometry for interiors considering how many windows there may be visible from the outside.

 

www.gametrailers.com

search for crackdown, and then watch ''trailer 2''.

There is no slowdown in that trailer, and it looks superb.

 

What, were you thinking that framerate issues were a feature they would show off in a trailer?

 

http://www.1up.com/do/newsStory?cId=3156556

Check the crackdown feature, seems theyve really tightened up the framerate.

 

I'm in. This game looks stupid fun!

 

Ummm... trailers are captured at fixed framerates irrespective of whether the game actually can run at that framerate or not. All the routines are given a fixed timestep of 1/30th of a second, so it looks like the game is running at 30 fps when you actually play the video. or whatever framerate you want to encode the video at. Sometimes they might play with you and step at faster framerates and blend images to make it look like there's extra motion blur. This isn't new. Pretty much every game trailer you've ever seen or ever will see with realtime footage does this.

Things that aren't really "trailers" per se, like realtime demonstrations which were filmed on camera are a different story...

 

Kinda a strange comment from the woman about only shooting people and the agility and racing aspects almost to be a seperate game.

I think she kinda missed the appeal of the RPG elements, personally, I would want my agent to be maxed out in nearly every category, explosives, driving, agility, as such I'll probably spend 60% of the game just leveling up, which would make those activities the bulk of the game for me.

I dunno, I think the true character development elements offer much more variety than missions in GTA, which the only variation on killing people were the 'pizza delivery' missions. Granted GTA had a handful of cool flying missions, but other than that very repetitive. Basically, trussed up fetch quests.

The idea of developing my own character, acquiring new abilities, and doing things like scaling the largest building in town, or racing across rooftops parkour style is much more true variety than we've seen before, imo.

It seems the one thign they missed the boat on are some really nice and original boss battles, with the powers available, they shoudl've been able to think up some really off the wall, original boss battles, which it doesn't seem like they have.

 

When I heard about the character advancement (which seems to be limited, since it isn't strictly an rpg) I actually started looking forward to playing the game (the demo at least), previously I had no interest at all. If I like the demo I might end up picking it up... but the rpg-ish aspects are what have me interested at all, even... gta with rpg like aspects sounds kind of fun to me!

If I end up liking the game and getting it, I have a feeling I'll be spending most of the time maxing out my character too.

 

Where the hell did this guy learn how to count?

 

It's for mass user understanding as there can be disambiguation between billion and trillion. It's not uncommon to see that sort of wording in astronomy books in describing light years or radii in more down-to-earth scales.

 

The hell in school.[/Spock]

 

I think they employed the Instant Radiosity algorithm to calculate the global illumination and then store the result in the lightmap. The algorithm actually is more like a splash version of Photonmap, and can be implemented efficiently on GPU. It's true Quake 3 and Half Life 2 already used precomputed radiosity map (or albedo map), but this tech is very fast and easy to use. We used this in our game too.

I'm really interested. What the "unusual way" are they using?

Well, if you are familiar with the Relief Mapping based algorithms (relief map, parallax occlusion map, cone map, steep map, ...), all of them are running a micro-raytracer on each pixel to detect the actual surface point defined by a height map (see the document of ATI's Toy Shop demo). But, these algos have "granularity" artifact, which is very annoying at some view angles and can not be MSAAed. So, the term "the next generation beyond parallax mapping" maybe refer to something improved on this?

Deferred lighting has a few major deficiencies like cannot cooperated with MSAA on the current gen hardware. Maybe that's the reason why they outlining everything?

 

Xenos doesn't have the current gen problem (actually, "last-gen", since G80 has the same capability as Xenos, being a D3D10 GPU) - it can resolve individual MSAA samples, whereas last gen hardware can't.

I guess that the "outlines on everything" effect is achieved by accessing individual MSAA samples at poly edges and determining if there's a surface discontinuity there (according to Z).

Jawed