splinter cell 3 pc demo....thoughts...?

Some SM 1.1 and 3.0 comparison shots

http://www.shacknews.com/ja.zz?id=9596795

Needless to say, the lighting in the 3.0 shots looks much better.

 

why do the rock faces change? Parallax mapping?

 

Looks like it. . . Pretty nice.

 

Thank you for this post.

/me crosses SC3 from "looking forward" list.

 

At how much of a FPS loss? Is it worth that? Thats up to the end user.



I think it does look better, not sure so about "much better". Just looks brighter, as if someone upped the brightness.

edit, took the links to the pics out, they were removed.

 

It's definitely more than just brighter. . . Even ignoring the difference in shader versions, there is another very significant difference.

 

Due to confidentiality, I can't talk much but let's just say that next-gen consoles play a role in the decision. We can discuss about possible NVIDIA influence all we want (and I'll never comment on it) but perhaps it may be a good idea for some of you guys to ask a few developers what their thoughts are about SM2.0, SM 3.0, SM 4.0 and how all three models relate to game-design decisions taken/will be taken while considering the next-gen consoles. For example -- will SM 4.0 gain strong acceptance when we have SM 3.0 now ...?

[edit] PS. You can all try asking Dany Lepage and see what he can tell you that you can post publicly!

 

Counts easier and faster implementation as a technical reason?

With SM3 they were able to build large shaders with all options the need and activate for each object the right options with some flags. You can't do this with SM2 because the length of a shader is limited at a low level. SM2.X cards permitted the necessary length but missing the static branching feature. Without this feature every instruction in a shader will be execute even if you don't need it for the current object. That leads to a situation were you can't use the same concept that you allready use for a SM3 path for a SM2/SM2.X too. If you want to use SM2/SM2.X you need a more complex concept.

 

And the PS/VS 1.1 ??

US

 

Well, Epic was able to overcome these problems in UE3, right? It was Ubi's decision, and probably not a wise one.

And Rev, SC3 does not even come on next-gen consoles, right? And UE3 does...

 

Maybe he meant that the engine would be used for a SC on next-gen consoles (Xbox 2 launch title?)...
Still, it doesn't really excuse the complete lack of SM2.X support.

Working on next-gen product and therefore including SM3.0 is fine and all, but forget, on purpose, to support a vast numbers of cards out there because of it strikes me as an odd decision to make.

Anyway, I'm just saying, because after SC, SCT and after having played this demo (And to a certain extent MGS3), I now consider myself SCed out.

 

The drama queen are out already.

SC and SCT were all Dx8 and the game was really good. Plus their main market was the XBox anyway.

I think people should see the sm3.0 path as a bonus since the content is already aimed at current consoles. (Same was true for Riddick.). So why complain ?

 

And the PS/VS1.1 Path???

 

This is a simple case of Ati versus Nvidia.
I, for myself, am surprised that it didn't create more noise than that.

I didn't read Rage3D, though. Maybe the streets are already on fire...

 

Dany Lepage, Mon Oct 11, 200

B3D : There are a number of noticeable gradients (or "banding" if you like) on the wet per-pixel lit stone next to the stairs (presumably from the light above the door at the top of the stairs). Is this due to the screenshot taken using the SM1.1 renderer? Will this artifact be visible using a SM2.x or SM3.0 renderer? I assume this has to do with precision.

Yes, the gradients are likely due (didn't test but considering the pattern direction, seems like it) to the limited precision of SM 1.1. The other main source of "banding/gradients" is the compression on the textures and most specifically the normal map. Non compressed normal mapped are significantly higher quality than normal maps using DXT1/3 compression. That said, while this creates problem with shader assets trying to represent new/shiny/polished materials, it does work very well for organic noisy shaders like the one in this screenshot. Internal computations at 24/32 bits are solving this problem.

Errors in my x800 xt pe

 

Your x800 is not computing the data at 24/32 bits. This is in sm2.0, and it's using 1.1.

 

I know it...

 

Ah, ok. I thought you were wondering why this was happening.

 

Just played it.

I hate supposed action games which turn out to be jump-and-run right at the start. I also hate all games which have that crappy Tomb Raider kind of camera movement. Also, the controls suck so badly, it's unbelivable.

Nice gfx, but I certainly won't be bying this.

 

The SM1.1 Path use multipass shading. As Example the stone with the life boat you can see at the beginning use 6 shaderpasses.

You can use this concept for a SM2 version too but this would make nothing better. If you use 6 SM2 shader instead of 6 SM1.1 it will only run slower on Geforce FX cards but don't give you any visual improvements. The reason for this is that the passes are still store the intermediate results in the 8 bit framebuffer and use alphablending to connect everything together.

You can try to reduce the number of passes but it is only save to build on SM2 pass from two SM 1.1 pass. If you want to connect more than two passes together you need a system with at least a little intelligence.

Anyway this concept that is used for the XBox too is not the solution that give users with SM2 cards the SM3 effects. The only way is to port the SM3 shaders to SM2 shaders with a different concept.