Unreal 3.0 and Dynamic Radiosity Lighting

Status
Not open for further replies.
Scali said:
You are insulting here...
But where exactly did I insult?
And just because you don't know who I am doesn't mean I am a no-one or know nothing, etc. Get a grip.
Saying Doom3 is "Carmack's stupid little outdated game" is what I consider to be either grossly uninformed of the business of making games or completely egoistic.

I wouldn't find fault with your words but the addition of the words "stupid" and "little" just makes you look bad in my books (not that you may care, of course).

Doom3 (the game) is not "stupid" and it is most definitely not "little". Doom3 (the engine) may be slightly "outdated" but when you are a programmer that started on an engine 4 years ago and 2 major APIs (and the resulting hardware) ago, there will always be some " technology lag". Ferchristsakes, look at the state of OpenGL then and now.

Your words are insulting to a man who has done much to change not only the way game engines (and not just the rendering part) are made but the way hardware are shaped. If you think Carmack has been given too much credit, perhaps you can do something about it beyond words on public forums. I look forward to your first game (don't take too long now).
 
Reverend said:
Saying Doom3 is "Carmack's stupid little outdated game" is what I consider to be either grossly uninformed of the business of making games or completely egoistic.

Both wrong.
I explained why it was outdated... the rest is me getting annoyed, human nature, you know?

Doom3 (the game) is not "stupid" and it is most definitely not "little". Doom3 (the engine) may be slightly "outdated" but when you are a programmer that started on an engine 4 years ago and 2 major APIs ago, there will always be some " technology lag".

Funny that there are other games that have been released, or will be released soon, that have not been affected by this "technology lag" nearly as much.

Your words are insulting to a man who has done much to change not only the way game engines (and not just the rendering part) are made but the way hardware are shaped.

There is a difference between what was, and what is.
Sure, Wolf3d, Doom, Quake, great... fantastic...
Doom3? Nah. Been done before and better. Heck, I even wrote an engine like that myself back in the GF2 days.
Why would I possibly look up to Carmack still, if even I myself have done the same stuff years ago? If I can do it, anyone can, I'm nothing special, just your average graphics programmer.
And it pisses me off that below-average programmers, or even non-programmers go on forums like these, spread all kinds of nonsense about Doom3, and when people who DO know what they are talking about, because they have coded the same stuff aswell, and moved on from there, explain what it's really like, they just get ridiculed and insulted.

If you think Carmack has been given too much credit, perhaps you can do something about it beyond words on public forums.

There is no cure for stupidity.

I look forward to your first game (don't take too long now).

You don't even know who I am. Perhaps you've been playing my games for a long time now?
That's another stupid thing about human nature. They assume that people they don't know are not capable of anything, and can't possibly be smarter or more experienced than they are.
 
Except that nobody else has yet produced the game that combines the shadowing and the polygon-based bump mapping that Doom 3 has. Sure, we've all seen tech demos for some time that used Doom 3-esque technology, but it's still the first game to do it.
 
I doubt if there ever will be realtime engine without preprocessing. Even raytracing, no matter how unified is it, has to do some preprocessing )data structure for efficient ray queries). There will always be hacks unless there will be some revolution in physics that would allow the construction of massively parallel processors capable of simulating universe.

Doom 3 must do some preprocessing, however not to do lightning but to do efficient occlusion culling.

I wonder if it's still using Quake-style PVS+BSP, or something more dynamic.
 
Chalnoth said:
Except that nobody else has yet produced the game that combines the shadowing and the polygon-based bump mapping that Doom 3 has. Sure, we've all seen tech demos for some time that used Doom 3-esque technology, but it's still the first game to do it.

No, but as I said before, that's just a formality, not something revolutionary.
 
rehcra said:
I doubt if there ever will be realtime engine without preprocessing. Even raytracing, no matter how unified is it, has to do some preprocessing )data structure for efficient ray queries). There will always be hacks unless there will be some revolution in physics that would allow the construction of massively parallel processors capable of simulating universe.

Yes, but that would be unified preprocessing, and it would not alter the lighting system in anyway, it is merely there to accelerate the processing.

Doom 3 must do some preprocessing, however not to do lightning but to do efficient occlusion culling.

No it does actually use preprocessed lightmaps, as stated many times before. And these lightmaps introduce limitations on the lighting and animation system. All been said many times before.
Ofcourse you can just continue to ignore it.
 
A Doom 3-equivalent has been done before, and better? When and where? I don't consider tech-demos to be games. There is a huge difference between producing a game and a demo.
 
Scali said:
And it pisses me off that below-average programmers, or even non-programmers go on forums like these, spread all kinds of nonsense about Doom3, and when people who DO know what they are talking about, because they have coded the same stuff aswell, and moved on from there, explain what it's really like, they just get ridiculed and insulted.
That's not the issue here really - it's the way you post replies and hold discussions. You bite every single time; you let your personal feelings dictate what you post in a technical discussion. If you're a lone voice in a debate, the only way you can ever expect people to follow your side/point of view is via thorough explanations and/or demonstrations - I'd be out of job in an instant if I had to reply to all of my students who disagreed with my explanations with comments such as "I find your ignorance amusing".

I tell you one thing though, if I'd written a Doom3-copy engine several years ago, I'd be flogging it for all it's worth right now. Opinions and technical discussions aside, the D3 engine is clearly the flavour of the month/year and anything as good must surely sell as well.
 
Scali said:
Yes, but that would be unified preprocessing, and it would not alter the lighting system in anyway, it is merely there to accelerate the processing.
When we talk about real-time rendering, "accelerating" (as you call it) is essential so that you don't have to wait ages for rendering to complete.

Here is my point: Ray-tracing is proven (theoretically) to have better scalability that z-buffer rendering. (It's only much harder to do in hardware than z-buffer). But even if we had ray-tracing GPU-s, we would have to resort to some hacks to update the scene hierarchy. That makes me think hacks in games will be there for at least next 50 years. (hacked != unified).

Doom 3 must do some preprocessing, however not to do lightning but to do efficient occlusion culling.
No it does actually use preprocessed lightmaps, as stated many times before.
Can't you read? I wrote that Doom's preprocessing is not about lightning, it's about culling.
 
rehcra said:
Here is my point: Ray-tracing is proven (theoretically) to have better scalability that z-buffer rendering. (It's only much harder to do in hardware than z-buffer). But even if we had ray-tracing GPU-s, we would have to resort to some hacks to update the scene hierarchy. That makes me think hacks in games will be there for at least next 50 years. (hacked != unified).

But that's another point altogether.

Can't you read? I wrote that Doom's preprocessing is not about lightning, it's about culling.

How exactly are lightmaps not about lighting?
 
Cat said:
A Doom 3-equivalent has been done before, and better? When and where? I don't consider tech-demos to be games. There is a huge difference between producing a game and a demo.

I believe we were talking about the rendering system here, and the technology used therein.
In which case, Doom3 isn't very revolutionary at all. Everything is based on concepts that have been in use for years. Heck, the main components of Doom3, namely shadowvolumes and Blinn-Phong per-pixel shading date back to the 70s.
They weren't used in games before for a simple reason: we didn't have the hardware. Obviously that would change eventually, now it did. Revolutionary? I don't think so.
 
Neeyik said:
That's not the issue here really - it's the way you post replies and hold discussions. You bite every single time; you let your personal feelings dictate what you post in a technical discussion. If you're a lone voice in a debate, the only way you can ever expect people to follow your side/point of view is via thorough explanations and/or demonstrations - I'd be out of job in an instant if I had to reply to all of my students who disagreed with my explanations with comments such as "I find your ignorance amusing".

Hey, if you attack me and talk rubbish, I make fun of you, you ask for it.
I have already given thorough explanations many times before, I am sick of it. Everyone who still doesn't get it... well, they will never get it. And therefore they should stay out of the discussion, because it is apparently out of their league... But instead, probably because of some emotional bond with Carmack, they find it necessary to attack me.

I tell you one thing though, if I'd written a Doom3-copy engine several years ago, I'd be flogging it for all it's worth right now. Opinions and technical discussions aside, the D3 engine is clearly the flavour of the month/year and anything as good must surely sell as well.

There is a big difference between writing a simple rendering system like Doom3's, and actually producing a full game with all the AI, physics, and whatever content the game requires.
We were discussing the rendering system, I made a similar one years ago, and I bet a lot of people have. That part doesn't impress me. The content of Doom3 is nice, but that has little to do with Carmack's work.

Also, we have yet to see how well Doom3's engine will actually sell. So far, I only know of Quake 4 using it, and that game is not entirely independent of ID anyway, obviously. I wouldn't be surprised if the Source-engine will be a lot more popular.
 
By your definition of 'revolutionary,' id hasn't ever produced anything worthy of your praise.

Theory is always going to be far ahead of application.
 
Cat said:
By your definition of 'revolutionary,' id hasn't ever produced anything worthy of your praise.

Theory is always going to be far ahead of application.

Not at all. Wolfenstein clearly was a novel approach to rendering, and obviously it created a new game genre: the first-person shooter.
Which is... revolutionary.
Doom may have been more evolutionary though, it improved upon the raycasting method.
Quake pretty much was an application of standard concepts, much like Doom3... for the first time, hardware was fast enough to do it in realtime.

Doom3... I see that as Quake + normalmaps + stencilshadows + hacks.
Anyone who knows Tenebrae knows Doom3 wasn't the first ;)
Ironically, Tenebrae applied Doom3's concepts (just like 3DMark03)... problem was just that Doom3 got delayed.
If anything, the prize for 'first full stencil-shadow game' should go to Tenebrae.
 
Scali said:
There is a big difference between writing a simple rendering system like Doom3's, and actually producing a full game with all the AI, physics, and whatever content the game requires.
We were discussing the rendering system, I made a similar one years ago, and I bet a lot of people have. That part doesn't impress me. The content of Doom3 is nice, but that has little to do with Carmack's work.
So the renderer is the easiest/shortest/smallest part of the total workload in a game development?

Also, we have yet to see how well Doom3's engine will actually sell. So far, I only know of Quake 4 using it, and that game is not entirely independent of ID anyway, obviously. I wouldn't be surprised if the Source-engine will be a lot more popular.
The next AvP game will be using D3 and RTCW2 will almost certainly use it; there was another one too but I can't remember off the top of my head. Source definitely has a far larger confirmed list though.
 
Scali said:
Cat said:
A Doom 3-equivalent has been done before, and better? When and where? I don't consider tech-demos to be games. There is a huge difference between producing a game and a demo.

I believe we were talking about the rendering system here, and the technology used therein.
In which case, Doom3 isn't very revolutionary at all. Everything is based on concepts that have been in use for years. Heck, the main components of Doom3, namely shadowvolumes and Blinn-Phong per-pixel shading date back to the 70s.
They weren't used in games before for a simple reason: we didn't have the hardware. Obviously that would change eventually, now it did. Revolutionary? I don't think so.
It's all a matter semantics and how you define revolutionary. IMHO Doom 3 is a revolutionary game engine; it may not be the opinion of a well-versed engine programmer, but of an avid enthusiast who can understand and interpret the direction/development of real-time rendering in retrospect.

Doom 3 may use hacks/techniques that were around long before, but it was a bold move in that no other game before it used the phong shading model for every pixel in addition to a robust (robust meaning a singular implementation for all but 2 game instances) shadow algorithm for the dynamic shadowing of pixels based on light occlusion data. It is real-time scanline rendering's priliminary attempt (or parody, however you may see it) at dynamic global illumination. In this attempt, inderect lighting was unaccounted for, probably because it would'nt have been dynamic. Therefore Doom 3 represents, to some degree, how accurately rendering can be, at this point in time, when rendering everything dynamically and scalably is paramount.

There may have been a few exceptions made for the flashlight, air vents, and grates, but it seems these were more of an aesthetic decision rather than a decision based on engine limitations. I'm not sure if it would have been possible to make the flashlight a spotlight, but I do know that the lightmap interacts dynamically with the environment and does not break the shadowing scheme. Using projected textures for the ceiling fans seems to have been more of an artistic desicion, being that the engine shadows dynamic objects with pointlights at various other instances in the game. We all know about the grates, perhaps the only (if not, one of the few) in-game situations where stencils did not provide a reasonable solution. Therefore we have one case in which a lightmap is used and one in which shadows were removed (self-shadowing is available in-game atm). I believe the main pipeline would have dealt with both cases without the need for omissions/hacks, but that is only a partially informed opinion.

One can claim that level design was artistically taylored to the rendering engine, which is true to some extent, but the gist of having a unified lighting and shadowing algorithm was there. If Doom 3 would not have made the move, we would be continuing to see games which offered dynamic lightmaps on characters with precomputed static lighting on many models and game environments.
 
Scali if you are so important why do you have a morbid fear of saying who you are? Your arguments rely heavily upon your supposed expertise and opinions, so it is only logical that people might wonder about your experience. Why not show what you actually coded years ago, or tell us who you supposedly are so we can all see how right you are instead of just blathering?
 
I wonder why you are talking about doom's lightning over and over. How does doom differ from other games when it comes to visible surface determination (or occlusion culling)? Is it PVS+BSP?
 
I'm just responding to Scali's post and giving a perspective. Everyone and their mother (I sat mine down with me to see Sweeney's UE 3.0 presentation and Carmack's quakecon interview ;) ) knows the information given by myriad of posts made on this forum, however it is important that we all share our perspectives. There is a programmer's perspective, a laymen's perspective and so on. It is much easier to not post, but if everyone thought that way we'd be in serious trouble.
 
Status
Not open for further replies.
Back
Top