This is a bit off topic, but, have you guys seen this?! http://mywebpages.comcast.net/chuk/Doom/ Check out the wallpaper. Bump mapping looks great, but that marine looks amazing!!!
What do you think?!
looks good when does it come out?
Looks good, but I think the stills are deceptive. Bump maps look great in stills, but when you can move around the objects you can very quickly start to see the actual geometry of the object. Just wondering how the bumped characters will look in game. One good thing about Doom3, the requirements will probably be high. I can’t force people to upgrade to play my game, so when external forces cause a mass upgrade that’s good… 
Actually, in my opinion, the game looks much better in motion. You don’t have time to pick out the small flaws, and the animation is superb.
I agree it’s about time someone made a game worth upgrading for
– Zeno
it does look much much better.
IMO the first shots from that tokyo video were good, then all the shots afterward eg monster in bathroom looked terrible (esp the shadows).
but now it looks like theyre back on track again (though its prolly only just better textures )
>>I agree it’s about time someone made a game worth upgrading for <<
there is im just sticking in the finishing touches, get ready to KILL EM ALL
The motions and collision testing are quite important too. I wonder if there will be any of those objects that look like they are floating in mid airjust because 1/10th of the object is on top of another object.
I’ve seen some clips (on TV) and the FPS seemed interestingly high. You will definitly need the latest system. I say system and not just graphics card!
V-man
Originally posted by V-man:
[b]The motions and collision testing are quite important too.
I’ve seen some clips (on TV) and the FPS seemed interestingly high. You will definitly need the latest system. I say system and not just graphics card!
[/b]
According to the recent article in PC Gamer, a ragdoll-type physics engine (I’m guessing similar to what Hitman2 and UT2k3 are using) is already in place as well as per-poly hit detection.
Oh, they are currently running w/ an ATI 9700 at 800x600 resolution. I’m sure that’ll improve before release, but it gives an indication of what type of horsepower you’ll need to really get the most out of this game. Hope my system is up to the challenge…
Those really are low poly models. That spindly boss creature looks pretty bad. The geometry reduction using normal maps looks good on big fat characters, but really looks bad with thin splindly ones, in my opinion. I hope these tradeoffs are worth it just to get the shadows…
Originally posted by knackered:
Those really are low poly models. That spindly boss creature looks pretty bad. The geometry reduction using normal maps looks good on big fat characters, but really looks bad with thin splindly ones, in my opinion. I hope these tradeoffs are worth it just to get the shadows…
Better than it running like crap because the model has too many triangles…
Mmm, but not better than using perspective shadow maps, rather than shadow volumes. He could then jack up his poly count to return us to the number of polys our eyes have now become accustomed to in video games.
Beauty is in the eye of the beholder. I think it looks good, showing a screenshot closeup on a double page spread is not typical of the ingame experience.
Aww, cmon. I think the monsters in Doom III have as many if not more tris than those in any current FPS. It just jumps out at you more because of the realism of the lighting and textures.
The poly counts will increase in future games based on this engine. I think it’s probably a trade off they had to make to be able to trace silhouettes, do mesh skinning, physics, and accurate collision detection on mediocre CPU’s.
Another thing to consider is that the poly-count to shadow volume overdraw ratio is not linear. Doubling the poly count would probably only increase the depth complexity a little bit.
– Zeno
> the animation is superb.
Are they finally using actual skinned characters? Better late than never, in that case 
actually taking a closer look at the pictures yes they are very low polygon. i thought there were meant to be 5000 polys a charcter in doom3. these look like quake3 characters ie around 1000.
though i suppose 1000tris with decent lighting + textures looks better than 10000 with crappy textures + lighting. actually i dont suppose, i know ive tested it. very similar to why sometimes a raytaced demo at 320x240 looks better than a normal 3d accelerated scene at 1024x768
Judging by their silhouettes they’re using less polys than quake3 models.
Obviously this is because of the need for shadow volumes - but I’ve been playing around with perspective shadow maps, and they seem to be a better solution than volumes. Maybe in doom4.
I don’t think they will go over 1500 polys…
From my personal experience with shadow volumes they are very heavy on the cpu and you can’t put the models on the card like ut2k3 does…
Also if you do multiple passes with one poly well you get n*1500 polys then
Besides on the polycount question I saw some interview where they didn’t actually answer it but only said “the bumpmapping will make up for this” so I won’t be that high I guess 
On the better/worse looking in motion stuff, well it will look better if they find some way to reduce the “shadow popping” on polygons otherwice it can lead to annoying artefacts…
Charles
Using vertex shader shadow volumes would greatly reduce the cpu burden. The card still has to deal with the silhouette polys of course but you also need to render your model for every light when using perspective shadow maps. Of course the silhouette has more polys than the backface culled model but I still think that vertex shader shadow volumes aren’t that bad.
The models have between 2000-6000 polygons. Interview with JC ( from post-E3 ),
http://www.beyond3d.com/interviews/carmackdoom3/index.php
While creating the shadow volume geometry is a lot of work for the CPU, it quite easy to optimize ( especially for a portal based engine ).
Hum,
(…)
While creating the shadow volume geometry is a lot of work for the CPU, it quite easy to optimize ( especially for a portal based engine ).
(…)
btw and how to optimize ? i’ve been searching stuff about this but haven’t found anything usefull.
tks
quasar
What does cpu intensiveness of shadow volumes have to do with portal based visibility? Unless if you mean that you can easily cull away nonseen shadows which is about the best optimization i think. What can be better than not doing it in the first place? i too am curious on optimizing shadow volume cpu work.