Request for advice -- numerous signs
by John Doppler Schiff · in Artist Corner · 08/01/2006 (10:10 pm) · 2 replies
Hi folks,
I'm working on a level that contains a large, multi-story shopping mall: approximately 96 stores, each with its own signs and posters. I'm using TGE + TLK 1.4.
The signs are the sticky issue at the moment. I'm weighing three approaches:
1. Use a DTS with variable skins.
2. Use TLK's projected decals.
3. Embed the signs in the DIF for the mission.
Of the three approaches, the third is the easiest, but the least flexible. I expect that the signs will change periodically over time, so the variable skins or the decals look like the way to go. I'm concerned about the impact on frame rates with over 100 decals or separately skinned transparent DTS objects however.
What's your take on it? Is there another approach I've overlooked?
Your opinions are greatly appreciated!
-- JohnDopp
I'm working on a level that contains a large, multi-story shopping mall: approximately 96 stores, each with its own signs and posters. I'm using TGE + TLK 1.4.
The signs are the sticky issue at the moment. I'm weighing three approaches:
1. Use a DTS with variable skins.
2. Use TLK's projected decals.
3. Embed the signs in the DIF for the mission.
Of the three approaches, the third is the easiest, but the least flexible. I expect that the signs will change periodically over time, so the variable skins or the decals look like the way to go. I'm concerned about the impact on frame rates with over 100 decals or separately skinned transparent DTS objects however.
What's your take on it? Is there another approach I've overlooked?
Your opinions are greatly appreciated!
-- JohnDopp
#2
It's funny that you should mention off-world colonies! The game I'm working on takes place aboard a space station, and this was intended to be one of the few large open spaces on the ship. The other shopping centers are crowded bazaars and hectic marketplaces, and they use portals liberally, but this wide-open space just doesn't offer many opportunities for sectioning off areas like that.
In retrospect, I may have to move parts of this mall indoors to take advantage of better zoning.
With regards to the ray-casting, I suspect that would bring my target audience's systems to a screeching halt. The geometry of the level isn't obscenely complex, but it's involved enough to make a low-end CPU sweat some silicon.
08/02/2006 (6:39 pm)
Thanks for the feedback! I'm currently working with the DTS approach, and it looks like a winner. I decided to loop through an array populated with the poster and sign info (transform, sign texture, sign type). As I add each sign to the level, it's skinned and positioned. Nice and flexible, easily updated, no strain on the brain. I'll have a test ready tonight, and we'll see how the frame rate stands up to all this mishegas.It's funny that you should mention off-world colonies! The game I'm working on takes place aboard a space station, and this was intended to be one of the few large open spaces on the ship. The other shopping centers are crowded bazaars and hectic marketplaces, and they use portals liberally, but this wide-open space just doesn't offer many opportunities for sectioning off areas like that.
In retrospect, I may have to move parts of this mall indoors to take advantage of better zoning.
With regards to the ray-casting, I suspect that would bring my target audience's systems to a screeching halt. The geometry of the level isn't obscenely complex, but it's involved enough to make a low-end CPU sweat some silicon.
Associate Orion Elenzil
Real Life Plus
You might just give the DTS approach a shot and see how the performance turns out.
It might be possible to quickly mock-up a sample.
fwiw,
the project i'm working on has 350+ dts objects in the mission, and we get decent-tho-not-phenomenal framerates. most of them share common textures, tho.
also, Zones are your friend. if the shopping mall has an "open" design it's hard to do much Zoning, but if it's um, say, like a series of tunnels connecting discrete small areas, maybe like you might find in a dungeon or some sort of off-world mining colony or something, then wow! Let's hear it for zoning! I think more city planning commisions should pass more laws requiring that real-world environment be designed for more efficient Zones and Portals. ;)
you might also look into doing some visibility culling. for example by casting rays to the corners of each poster. if the bottleneck is transferring texture data to the graphics card, this might be worth it. if the bottleneck is CPU, then obviously it's maybe not such a hot idea. ray-casting for visibility isn't ideal and can lead to Popping of posters in some cases, but if the environmental geometry ain't too complex it can be pretty close to good enough.