Any chance that maybe eventually, Doomsday Engine could have support for OpenGL 3 or 4? Just a question I've been meaning to ask since the Doomsday project is going so well.
Any chance that maybe eventually, Doomsday Engine could have support for the OpenGL equivalent of DirectX 10 or 11? Just a question I've been meaning to ask since the Doomsday project is going so well.
We are planning to rewrite most of the renderer and indeed it will require a newer version of OpenGL. However, it remains to be seen which version we end up choosing as the minimum: it depends on which rendering techniques we want to use.
We intend to keep the 1.9 renderer around as an alternative as long as possible, too, to support older OpenGL 1.4 systems.
It's astoundingly but sad that such a game/engine that has been ported to numberless platforms, and certainly leads in the number of ports/engines that exist on the native platform (PC), is still missing modern engine with per-pixel lighting and all shader charms (with required media like specular-normal-gloss-height maps for retexturing and skinning), while all of its brothers (Quake 1-3 modern engines, even Wolfenstein 3D (shareware episode) remake with ppl: http://www.moddb.com/games/wolfram ) feature such tech and frofessionally made media.
As a great example of 2.5D engine with ppl i'll mention Duke Nuckem 3D HRP (with Polymer renderer)
I think, that making classic full world PPL (like in Doom 3 or Berserker Quake 2 or WolfRam) is not the best idea for Doom, while the game will look too different, lighting will be too contrast (overbright ond overdark in some areas i mean) and it will also require relighting each level. I believe, that the best way to add "next-gen" look but save classic Doom and jDoom feel is using lighting model simular to Duke Nuckem 3D HRP (Polymer renderer) or Quake 2 XP 1.26.2, combining existing lightmaps with dynamic lighting. This won't require relight, will feature correct "next-gen" lighting in third-party maps and will carefully keep original feel and look af the maps, the way it is meant to be by level-designers.
2 skyjake
Great! Has DE team started rewriting the renderer, any progress to show? Looking forward to...
Per-pixel lighting is absolutely part of our plans for the next-gen renderer, but as we want to move Doomsday forward as a whole rather than focusing solely on the renderer, progress is not that quick on any particular front.
The latest status is that I've written an OpenGL 2 drawing subsystem that is used for the new task bar UI in 1.11. It will eventually be used as the foundation of the next-gen renderer.
As to OpenGL 3+, as I commented before, it ultimately depends on which rendering techniques are needed to implement the vision we have for the renderer. However, we are also mindful of not excluding OpenGL ES compatible devices (iOS/Android), as the mobile platforms are continuing their rise in importance.
Great! It's exactly what I wanted to hear from DE team Can't wait to take a look at the next-gen classic DooM even on mobile devices (personally intereted in iOS port)!
So it is possible that you will only go up to OpenGL 2 and never progress beyond that? It makes sense to keep progressing just to allow for better and better graphics as time goes by.
So it is possible that you will only go up to OpenGL 2 and never progress beyond that? It makes sense to keep progressing just to allow for better and better graphics as time goes by.
The big change in OpenGL 2 was the programmable rendering pipeline that allows shaders to be used. OpenGL 3/4 expand on this, but they don't have as dramatic fundamental changes in the functionality of the API. The biggest challenge for us is to build a new renderer that deals with shaders and uses geometry and texture data in a GPU-friendly manner -- the old 1.9 renderer is fixed-function and draws GL graphics in a mostly "unbuffered" manner, uploading all drawn geometry again and again each frame. That is why Doomsday cannot currently take advantage of all the power of modern GPUs.
Also, just like we don't want to exclude mobile platforms, we neither want to exclude computers that are a couple of years old. OpenGL 2 is currently quite widely supported while the later versions aren't (considering all our supported platforms). But even so, once a renderer is OpenGL 2 compatible it is much easier to expand it with optional support for GL3+ features.
I have a Radeon HD 6870 from 2010 and it is OpenGL 4.1 almost 3 years ago. So I guess mobiles will take longer to have the processing power and hardware to support these abilities. soon we will have to find ways to create practical quantum computers or optic computers and 3D circuits because we can't make them much smaller and faster while keeping them cooler with current silicon transistor tech.
But that is neat. so this means that once OpenGL 2.0 is fully integrated, pretty much all performance issues and slowdowns that we have had with models and light, etc, will be eliminated? This sounds great if this is the main culprit.
Comments
EDIT: Have since corrected mistakes
We intend to keep the 1.9 renderer around as an alternative as long as possible, too, to support older OpenGL 1.4 systems.
As a great example of 2.5D engine with ppl i'll mention Duke Nuckem 3D HRP (with Polymer renderer)
I think, that making classic full world PPL (like in Doom 3 or Berserker Quake 2 or WolfRam) is not the best idea for Doom, while the game will look too different, lighting will be too contrast (overbright ond overdark in some areas i mean) and it will also require relighting each level. I believe, that the best way to add "next-gen" look but save classic Doom and jDoom feel is using lighting model simular to Duke Nuckem 3D HRP (Polymer renderer) or Quake 2 XP 1.26.2, combining existing lightmaps with dynamic lighting. This won't require relight, will feature correct "next-gen" lighting in third-party maps and will carefully keep original feel and look af the maps, the way it is meant to be by level-designers.
2 skyjake
Great! Has DE team started rewriting the renderer, any progress to show? Looking forward to...
The latest status is that I've written an OpenGL 2 drawing subsystem that is used for the new task bar UI in 1.11. It will eventually be used as the foundation of the next-gen renderer.
As to OpenGL 3+, as I commented before, it ultimately depends on which rendering techniques are needed to implement the vision we have for the renderer. However, we are also mindful of not excluding OpenGL ES compatible devices (iOS/Android), as the mobile platforms are continuing their rise in importance.
Also, just like we don't want to exclude mobile platforms, we neither want to exclude computers that are a couple of years old. OpenGL 2 is currently quite widely supported while the later versions aren't (considering all our supported platforms). But even so, once a renderer is OpenGL 2 compatible it is much easier to expand it with optional support for GL3+ features.
But that is neat. so this means that once OpenGL 2.0 is fully integrated, pretty much all performance issues and slowdowns that we have had with models and light, etc, will be eliminated? This sounds great if this is the main culprit.