These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE Information Portal

 
  • Topic is locked indefinitely.
 

New dev blog: DirectX 11 Tessellation Tech Demo at Fanfest 2012

First post
Author
Solhild
Doomheim
#101 - 2012-03-28 05:48:01 UTC
K Kerryngktonn wrote:
As far as I understand the tesselation in this scope is different from that is supported in OpenGL - the latter just provides a method to breakdown the geometry from polygons (mainly concave) into normal triangles that are always convex. The tesselation here is more like the one in 3dsmax where the model is just made smoother by subdividing the mech parts into more and more triangles.

Given this I don't see any apparent increase in the value of assets - apart from some curvy ships where these curves will be smoother, better interpolated. Rectangular designs (most caldari ships) won't win much, I suppose.

This really seems like applying a pretty filter on a photo via Photoshop, assets-wise. And so much pain to implement it? I don't see the reason why everyone is so agitated about this. "Oh, subdivided and smoothed geometry, we want it!". It's like upscaling a raster image - you won't gain new detailed geometry, but just a rough emulation of it.

Apart from that, I run EVE on Debian through Wine and won't really be able to take advantage of the tech in the middle-to-long-term run. But that's another story and is really my own problem.


It looks like the key is the new geometry that tessellation generates from the displacement map - check this pic.

The whole article is here, hope it means more to you than me. I have a GTX 580 and the tessellation demo certainly does the opposite of making something smoother.

Vohlos
Zyrconia
#102 - 2012-03-28 06:30:09 UTC
"Do it!"

No more words needed here ;)
DarkXale
The Scope
Gallente Federation
#103 - 2012-03-28 08:47:34 UTC
Lyron-Baktos wrote:
the other effects in that keynote, the ships shields blocking incoming lasers, titans breaking in half and having better looking fire, better looking damage etc etc, is that possible without Tess?

Those have absolutely nothing to do with Tesselation. Titans are just a model/animation upgrade, and shield 'just' involves adding certain standard effects. Both doable in the current client.
Ciar Meara
PIE Inc.
Khimi Harar
#104 - 2012-03-28 09:10:55 UTC
darmwand wrote:
Just curious: is it safe to assume that this is a DirectX feature and not limited to nVidia cards?

Edit: also, looks very promising, I'm looking forward to watching my nicely redone Ishkur in even more detail!


This is important, many of the people have high end computers but with ati hardware. I personally just buy what at the moment of purchase is the best bang for my buck this time that was ati.

I do hope that basicly everything torifranse and others showed us will come to eve, bounding asteroids, shattering, high details, shields lighting up, damaged ships, fire eating up your insides, plates falling off, picture in picture.

Make it so, but make it so that everybody can enjoy it or if they can't run it, switch it off. But the tech demo shown was run on a not to high end gpu, "the most used in eve today", with what kind of margin though, was not explained.

- [img]http://go-dl1.eve-files.com/media/corp/janus/ceosig.jpg[/img] [yellow]English only please. Zymurgist[/yellow]

Princess Hotbox
Ministry of War
Amarr Empire
#105 - 2012-03-28 09:37:31 UTC
I vote no until you can't think of anything else whatsoever that requires those production resources. There is still too much gameplay to be improved or implemented before worrying about shinier ships; they're already very attractive.

Some things that come first off the top of my head:
Starbases.
Ship Redesigning.
FW Conquest.
Sov and infrastructure.
New ships and variants (textures and models).
Effects.

The list is unending.
But, I would be ok if those sexy damage effects were produced instead. Oops
Steve Ronuken
Fuzzwork Enterprises
Vote Steve Ronuken for CSM
#106 - 2012-03-28 09:47:37 UTC
Princess Hotbox wrote:
I vote no until you can't think of anything else whatsoever that requires those production resources. There is still too much gameplay to be improved or implemented before worrying about shinier ships; they're already very attractive.

Some things that come first off the top of my head:
Starbases.
Ship Redesigning.
FW Conquest.
Sov and infrastructure.
New ships and variants (textures and models).
Effects.

The list is unending.
But, I would be ok if those sexy damage effects were produced instead. Oops



Not all programmers are the same.

The graphics update sounds like deep code infrastructure stuff
The stuff you're listing isn't.

Woo! CSM XI!

Fuzzwork Enterprises

Twitter: @fuzzysteve on Twitter

Bent Barrel
#107 - 2012-03-28 10:27:03 UTC
Tom Bodett wrote:
DirectX 11 support would be great.

What about supporting Multithreaded rendering which is part of DirectX 11?


the python client is single threaded anyway, so no gain ....
Bent Barrel
#108 - 2012-03-28 10:43:33 UTC
hmm ... I am more for an OpenGL layer than adopting DX11 ... but that's because it would remove quite a lot of problems for me (linux/wine here).

anyway this is not yet something that would translate into a gain for EVE (CCP, players etc.). Almost nobody plays totaly zoomed in to see all the bumps on the ship armor. if graphics candy is the goal then proper model allignment is more needed (fitted 75mm rails on the atron yesterday, looked horible !!!, same blasters on ishkur).

hitzone detection and proper reaction (explosions on the surface and not in the middle of the ship, shield deflection effects etc), better debris and wreck generation (ship explodes into parts and not into a generic wreck) and similar come to mind.
CCP Guard
C C P
C C P Alliance
#109 - 2012-03-28 11:39:09 UTC
The video from the demo is up and can be see here!

CCP Guard | EVE Community Developer | @CCP_Guard

Maul555
Xen Investments
#110 - 2012-03-28 12:33:49 UTC  |  Edited by: Maul555
Hellz yes i want you to pursue DX11 features in Eve Online! go go go team super awesome graphics!

out of curiosity though I would like to know if tessellation will be supported on my older ATI DX10.1 card. I seem to remember new fancy-dancy tessellation in hardware was a selling point of the card at the time...
Tesh Sevateem
Cherry Candy Mountain
#111 - 2012-03-28 13:40:36 UTC
I definitely would love to see any and all graphical updates to the EVE Universe in an ever flowing stream of updates.

If we didn't have the Trinity update, I wouldn't be playing right now. I know it's a bit extreme, but graphics matter, just like you don't want your girlfriend to get fat. Or, she's actually not getting fat, it's just all the other girls that get slim.

I'd better stop babbling.
Hamster Too
Golden Fowl
#112 - 2012-03-28 14:30:10 UTC
Well, as long as this nVidia assisted effort does not melt my ATI card I would love to see more detailed in-game models.
Nekopyat
Nee-Co
#113 - 2012-03-28 14:40:20 UTC
Bent Barrel wrote:
hmm ... I am more for an OpenGL layer than adopting DX11 ... but that's because it would remove quite a lot of problems for me (linux/wine here).


Not going to happen.

CCP has a DirectX culture, and it is a rare studio that can transition from one to the other given how religious of an issue DirextX vs OpenGL can be. People generally do not like training from one to the other since they are such different beats and toolchain.. not to mention because of the push over the years there is such a glut of DirectX programmers on the market it doesn't make much business sense to go with a tech that has fewer hires available. All the diploma mill game schools churn out DirectX developers at this point, with OpenGL ones coming from a different type of program.

Plus, schools have forgotten how to teach OOP in C, which hurts OpenGL since people look at the C code and go 'oh noes, not C! C is not OOP and that is the one twue way now! The design patterns look wrongz!'
K Kerryngktonn
General Mechanics Ltd.
#114 - 2012-03-28 15:17:16 UTC
My bad, didn't get deeper into this "other", new tesselation term and mixed it with the older and dumber one from like ten years ago or something. So I take my words back on it. I saw a small cave-in in geometry in the CCP screenshots but attributed it to, well, the rawness of the tesselation itself :)

Then it is surely a nice feature leveraging the usefulness of existing resources. Thought about parallax mapping, but they state in another topic that this won't do much to ships and require an additional mapping. So looks like this is it, we're in for tesselation in future.
xxxak
Perkone
Caldari State
#115 - 2012-03-28 16:43:29 UTC
You MUST make this happen.

Don't wait for losers with PCs from 2003.

WIN WIN WIN.

[u]The nerfs to supercaps will cause more super pilots to join the largest alliances who can properly "support" their deployment, further concentrating firepower/wealth in EVE. The end result will be fewer "fun" fights, and will hurt EVE in the long run.[/u]

John McCreedy
Eve Defence Force
#116 - 2012-03-28 17:30:50 UTC
Undoubtedly it looks amazing but how would it handle outside a tech demo and in a high-lag situation such as the average null sec battle?

13 years and counting. Eve Defence Force is recruiting.

mine mi
Garoun Investment Bank
Gallente Federation
#117 - 2012-03-28 18:52:27 UTC
Vohlos wrote:
"Do it!"

No more words needed here ;)


yes one more when? Smile
Bent Barrel
#118 - 2012-03-28 19:48:51 UTC
Nekopyat wrote:
Bent Barrel wrote:
hmm ... I am more for an OpenGL layer than adopting DX11 ... but that's because it would remove quite a lot of problems for me (linux/wine here).


Not going to happen.

CCP has a DirectX culture, and it is a rare studio that can transition from one to the other given how religious of an issue DirextX vs OpenGL can be. People generally do not like training from one to the other since they are such different beats and toolchain.. not to mention because of the push over the years there is such a glut of DirectX programmers on the market it doesn't make much business sense to go with a tech that has fewer hires available. All the diploma mill game schools churn out DirectX developers at this point, with OpenGL ones coming from a different type of program.

Plus, schools have forgotten how to teach OOP in C, which hurts OpenGL since people look at the C code and go 'oh noes, not C! C is not OOP and that is the one twue way now! The design patterns look wrongz!'


I do not expect it to happen, but one can dream :-)
Quade Warren
Aliastra
Gallente Federation
#119 - 2012-03-28 19:55:11 UTC
I noticed during the tech demo that the asteroid collisions were determined by the shape of the ship, not an invisible bubble. Does tessellation allow for better collision detection and if this was implemented, would players be able to rely on the shape of their ship to determine collisions in the future?

Getting stuck on an acceleration gate when I can visually see there is no obstruction is annoying, but understandable. An improvement on this scale would be, quite frankly, beyond awesome at this point.
Andrea Griffin
#120 - 2012-03-28 20:12:27 UTC
Quade Warren wrote:
Getting stuck on an acceleration gate when I can visually see there is no obstruction is annoying, but understandable. An improvement on this scale would be, quite frankly, beyond awesome at this point.
I'm going to guess no; collision detection with real objects that matter (gates, other ships, etc) are all done server-side, where the physics engine takes care of how things move around. We can't have the client being responsible for this sort of work, because then you'll have people flying through gates and stations and whatnot.