DX11/OpenGL 4.0 support?

[Nintonito]

WOuld be nice, considering how good the engine is. If not in SS3, will SS4 have this?

[DEN [CT]]

What features from DX11/OGL4?

[Rasta]

txaa, tessellation

[Nintonito]

Автор сообщения: DEN CT

What features from DX11/OGL4?

Really up to you guys. Hell I would be content with a well-optimzed renderer. But It would be nice to see things like Tesselation, HBAO, etc.

[kwahoo]

Автор сообщения: DEN CT

What features from DX11/OGL4?

All increasing performance features eg. GL 4.4 multibind, ARB_draw_indirect, ARB_bindless_texture. I don't need thousands of new features, but I'd like 120/144 fps on mid-range PC.

Back in 2001-2002 TFE and TSE were less demanding than every single game with similar graphics.

[Nintonito]

Автор сообщения: kwahoo

Автор сообщения: DEN CT

What features from DX11/OGL4?

All increasing performance features eg. GL 4.4 multibind, ARB_draw_indirect, ARB_bindless_texture. I don't need thousands of new features, but I'd like 120/144 fps on mid-range PC.

Back in 2001-2002 TFE and TSE were less demanding than every single game with similar graphics.

This +1. We need enhanced performance. Show those AAA devs how it's really done.

[Rockhead Gaz]

I actually feel that some of the more recent patches have really done a lot to improve performance. Of course, any further improvements would be more than welcome.

[DEN [CT]]

OK...

Tesselation
We're still not sure, lotta change in content. :( For the time being, parallax occlusion mapping is quite nice, isn't it? ;)

TXAA
Can be done on DX9... for now, we find FXAA as best tradeoff between quality and performance, but TXAA is really really nice, yes.

HBAO
DX9 also. However, it's a bit slow. We'll have another approach for our SSAO which will improve performance and quality at the same time. Just when we find some time to code that in. :-/

ARB_draw_indirect
Doing quite well with "standard" DX9 occlusion culling with 1 frame delay, thank you, but who knows, maybe some day :)

Multi-threaded DX11 device
Probably yes!

Performance
OpenGL is the fastest (from CPU side).

Anything else? Really, I'm curious about your input on this topic, so shoot...! :)

[kwahoo]

Автор сообщения: DEN CT

Anything else? Really, I'm curious about your input on this topic, so shoot...! :)

I wanna see glorious return of Temple of Effects and Temple of Models full of Utah Teapots;) Ok, I'm joking:)

[Nintonito]

Автор сообщения: DEN CT

OK...

Tesselation
We're still not sure, lotta change in content. :( For the time being, parallax occlusion mapping is quite nice, isn't it? ;)

TXAA
Can be done on DX9... for now, we find FXAA as best tradeoff between quality and performance, but TXAA is really really nice, yes.

HBAO
DX9 also. However, it's a bit slow. We'll have another approach for our SSAO which will improve performance and quality at the same time. Just when we find some time to code that in. :-/

ARB_draw_indirect
Doing quite well with "standard" DX9 occlusion culling with 1 frame delay, thank you, but who knows, maybe some day :)

Multi-threaded DX11 device
Probably yes!

Performance
OpenGL is the fastest (from CPU side).

Anything else? Really, I'm curious about your input on this topic, so shoot...! :)

Valid points. Frankly I would think that with some expert coding, you could leverage the new features of GL/DX to improve performance even further. Not only that, but things like HBAO and TXAA are better optimized for DX11 than DX9. Lastly, and this is a serious engine 4.0 comment, if you decide to abandon forward rendering, DX11 makes it much easier to maintain feature set.

[Solais]

Автор сообщения: kwahoo

Автор сообщения: DEN CT

Anything else? Really, I'm curious about your input on this topic, so shoot...! :)

I wanna see glorious return of Temple of Effects and Temple of Models full of Utah Teapots;) Ok, I'm joking:)

Actually, that's not a bad idea. It was always a thing for mappers and modders to reference and learn from.

[Terepin]

Автор сообщения: DEN CT

OK...

Tesselation
We're still not sure, lotta change in content. :( For the time being, parallax occlusion mapping is quite nice, isn't it? ;)

TXAA
Can be done on DX9... for now, we find FXAA as best tradeoff between quality and performance, but TXAA is really really nice, yes.

HBAO
DX9 also. However, it's a bit slow. We'll have another approach for our SSAO which will improve performance and quality at the same time. Just when we find some time to code that in. :-/

ARB_draw_indirect
Doing quite well with "standard" DX9 occlusion culling with 1 frame delay, thank you, but who knows, maybe some day :)

Multi-threaded DX11 device
Probably yes!

Performance
OpenGL is the fastest (from CPU side).

Anything else? Really, I'm curious about your input on this topic, so shoot...! :)

Since SS games are huge open areas, DX11.2 will come in handy. And the best part of it is that it is HW compatible with DX11.

[kwahoo]

Автор сообщения: Terepin

Since SS games are huge open areas, DX11.2 will come in handy.

Are you talking about Tiled Resources (arb_sparse_textures in OGL) aka Megatextures?

Megatextures are cool... when you have 200 artists to work on level. IMO makes no sense or Croteam.

[Nintonito]

Автор сообщения: kwahoo

Автор сообщения: Terepin

Since SS games are huge open areas, DX11.2 will come in handy.

Are you talking about Tiled Resources (arb_sparse_textures in OGL) aka Megatextures?

Megatextures are cool... when you have 200 artists to work on level. IMO makes no sense or Croteam.

Meh if you restructured the assets, it could make for better streaming, which means smoother performance, and increased texture detail in future engines.

[SpotteR]

Hi all
need not greatly overstate the settings not all strong PC

middling PC
Processor: 64-bit 3.0 GHz
Memory: 4GB - 6GB
Graphics: geforce 8800-9600 & radeon HD

remove DX 9-11 rendering & support only cross-platform OpenGL.
OpenGL 3.2 support video: Radeon HD, R; GeForce 8000, 9000, GT, GTX.

sorry my english, this is google translate.

[Terepin]

And you wrote it here because...?