Doom3world

btw, the 3rd video is out. It's the lab one where Gorden pushes the computer monitor off the shelf and pisses off the scientest guy, lol. Apparently the scientest guy's name is "Kleiner"

Jeez - tell me about it ! - good d/l speed from there for me. (if you consider 14k a second good )

What's up with this steam beta anyway ?
Why does it need to update itself virtually every time you start it ?

At the moment, all I can say is it sucks, hopefully it'll get better.

BTW - The divx versions of the latest high-res stuff are pretty crap compared with the Gman one. Get hold of the Bink ones if you have the bandwidth.

Well it always updates, because Valve people are working very hard and removing bugs, and figure they should release any fixes right away since they have this network to do it. Second, HL2 will use and be used though Steam for stuff like downloading language files, all multiplayer things, downloading new media, etc. So Valve must have Stream working great with no bugs in order for HL2 to succeed.

Latest video is out - traptown.

http://games.tiscali.cz/clanek/patche2.asp?id=5923

It's the best hi-res one yet !

If you've got the bandwidth, get the Bink version - 190meg.
The divx version is not very good quality - not that much better than the E3 footage.

With Bink, you can see how the game actually looks - pretty damn stunning !

This game is going to rock - anyone believing that the HL2 engine isn't that advanced over the HL1 needs to watch this vid, then play HL1 after.
If that still doesn't convince them, then new glasses and a new brain would be required

Yeah, it's cool and all, but I want the techdemo. To me, that was the coolest part. I also want to see the Gordan stained glass window in high-res

The bit outside with the spinning blade and the falling garbage bins and shit all seems a bit gimmicky, but other than that it's excellent. I love the wheel which goes whizzing past at the end

These videos are great.

From what I've seen so far Half-Life 2 seems focused on physics while Doom 3 is focused on the visual and auditory.

Either way both games are going to rock.

Just imagine you would be right - in that case HL2 would focus on something they didn't develop

But you're definately right about Doom.

I've seen all the Half-Life 2 videos to date. So far the only impressive features of the game are the real time physics and the refraction effect in liquids. And since the physics of Half-Life 2 are present because of the Havok 2 engine it makes you wonder what the fellas at Valve have been working so hard on.

Yep, I ask myself that question, too.

Since the HL2 engine can handle these advanced reflection/refraction shaders for some surfaces, why don´t they do bumpmapping, at least for the character models?? It would be such a tremendous gain in visual quality.

Having seen Doom3, I don´t want to play anything less anymore...

For example, Unreal2 had a really big polygonal detail on both characters and environment, but somehow I always had the impression that the look is "oldfashioned", even though it´s very detailed.

Well, maybe I´m too used to the bumpmapped look now.

Doesn't HL2 support bumpmaps, and per-pixel lighting? I'm fairly certain about the per-pixel lighting, but I can't really tell from the videos if there's bumpmapping or not.

I´m sorry for my unprecise wording. I was referring to per-pixel-lighting in general when I wrote bumpmapping.

The engine can apply pixelshaders/fragmentprograms to surfaces, that´s for sure. So that implies that it COULD do per-pixel-lighting, but only if they have bumpmaps/localmaps/whatever you wanna call them, that means they have to model the characters in highpolyversions and then renderbump and blabla, you guys here know the story better than anyone else.

So if someone with the superhires-bink movies should be able to tell us if the models are per-pixel lighted or not. I´ve seen the E3-movies on a DVD of the german "Gamestar"-magazine, and I´m almost sure they are not perpixel-lighted, but the footage quality isn´t perfect (not nearly as good as it could be with a decent VOB encoding ).

By the way, does it use OpenGL or D3D?

I remeber reading somewhere that the game is bumpmapped but instead of rendering bumpmaps from highpoly models Valve is painting them manually.

Which again leads you to the obvious, if they don't need to construct highpoly models and the physics engine is licenced, what exactly are they working so hard on?

Have you guys really looked at those movies ?

They are damn amazing - and this is just video, it's not the game.

We still haven't seen the high-res footage in the city center, the low res stuff was quite simply jaw-dropping.

What have they been working on ?

Well :-

1. Physics - everyone is saying 'ah, it's just the havok2 engine', seeming to completely forget that it has to be incorporated into the HL2 engine and most likely heavily tweaked and modified.

2. .wav speech syncronisation - which looks very impressive.

3. Highly detailed models with facial muscle animation - sure, they are not Doom3 quality, but they are still very impressive.

4. A flexible shader solution - some of the best water effects etc. I've seen for a while - and what about the camera projection stuff ?

5. Transforming landscape (dunno what this is called technically) - morphing ?

And then, what about the game concepts, storyline etc. and the implementation of that ?

It's not like you get all your tech sorted out and your done - you actually have to create a detailed believable world - probably the most time consuming factor of all.

By the time DoomIII is released, it will have also been in development for 5 years !

All we've seen of half-life so far is a handful of short movies, so how can you judge what they've been up to ?

Simple answer - you can't until you play the game.

That friggin amazing facial animation system with speech synch?

well possible its cause i'm developing games myself that i don't see anything groundbreaking in half life not technically and visually..
compared to doom3 tech their stuff is kinda dull and technically at least one generation behind.. that surely doesn't inculde gameplay which is not judgeable by the videos




from the look of the hl videos and the havok videos i can't see much differences there.. but if it would be that much work to implement the engine they would have coded their own from scratch.. so i would say that that was a more or less easier part




if the general animations would be on paar with the facial stuff that would be amazing.. as for the models itself.. i'm not impressed at all.. after doom3 it look kinda dated..



then you haven't seen much water shader.. from the look of the shader is a kinda simple water shader using precompiled cube maps for the reflection.. this is a kinda cheap way and you find such stuff in alot of games. like morrowind for example or tons of other games.. its not that
complicated.. what would be alot more impressive is if the water would be actualy recealculating the cube map every few frames so you actualy have realtime reflections.. in the docks video you don't see the reflections of the bird on the water for example.. with a little more shaderwork this could be done easily...



wow.. *cough* dragging vertexes around is not really that hard..
the red faction engine was way more impressive with their destrucable
enviroments (not that the game used that feature at all but thats another story)




well we don't know about that yet so no one can judge how it feels to play hl2.. just how it looks

but that doesn't stop anyone to talk about hl2 tech..

Yo!



I'm amazed too, but not by the visuals.
The thing that interests me the most is how much the AI is able to take the physics system into account when making a decision.



Not to critizise anybody, but what could change in the vids when watching them in higher res? What I saw wasn't jaw dropping, but it looked like a damn fun game to play.
No pun intended.



Agreed.

I think the hardest part is to teach the physics to the AI not the integration in the game engine itself from a technically standpoint where it comes to call function somewhere which get's handled by the physics engine.



Still the thing that interests me the most.
Imagine how CS modders will drop out from the scene in an instant when they have to set up 40 bones alone for the face
The amount of people bitching about complexity will by far outweight any "Who is the develope who makes a game with the Doom3 engine" discussion.



I think the characters will come closer to Doom3 characters when you see it ingame, but the whole style is so different from Doom that things look seriously off when you compare them next to each other.

HL2 eventually will suffer more from it's backward compatibility regarding the world geometry rendering to cheer the established hl/cs crowd than from the character models.

On the ohter hand Valve tries to be innovative in other fields (facial expression), so a lightmap based world geometry isn't something I can't live with.



I've seen better water effects two years ago in countless shader demos at very very high speeds, but as an ingame implementation it looks very spiffy. Unfortunately only when you look at the water from the outside - from inside out the view isn't as distrorted as I'd wish (afaik).

That camery projections stuff is neat but I'm very certain you can do this in Doom3 + many other games also.



Displacement mapping.
No big deal nowadays but cool notheless - not meant to diss the engine.
I think it may be harder to redetect the world collision model than to displace the geometry.
I've seen very very cool displacement mapping demos applied to characters - now that would seriously rawk!
Sorry to mention Doom³ agian, but not in a competitive way this time...

Since the Doom³ engine renders the whole scene blank into a buffer (der_ton may want to explain the correct buffer for us:) ) before it calculates the lights it may be too much of a performance hit to perform and redetect displaced geometry. Also vis and collision data may change.

I for example have made an .lwo landscape lately which I want to animate in MAYA in several different ways which could be triggered later.
Different technique, but the outcome should be pretty good (aside from potential texture stretching).
The insane texture resolution of 2048*2048 allows almost unique texturing wich is not only a great thing, it's damn cool in fact. Terrain without a green grass pattern ...
You can also renderbump it for additional detail... blahh - you know.



Looking very much forward to play an interesting story!



Don't lie!
We all are mappers and get told 24/7 by the regular players who couldn't start up a level editor even if it would creep up their ass that making a huge detailed level takes up maximum 3 to 4 hours!



Isn't more like 3 years ?

The rendering tech from John Carmack is a constant evolution and I supect we only see one of many renderes he has coded till he got one sufficient enough to achieve what he did foresee.
Also the first year of after-Q3 dev. was split between Q3:TA dev and The Quest dev which got changed to Doom³.



Because we are the best forum in the world...?



No - the fate of every game get's determined the day it gets announced.
I remember when Doom³ got announced how the whole Shackcrowd was making statements that they would only by it if it has feature a or b and even then it's very questionable. Viewed from this angle it would be much easier for developers to just announce games and feature lists instead of writing software and creating art assets.

Looks like we're getting paid by the word today

well yeah but my post is more agressive than yours :P

Earns you a red Orb and a bonus point