Death Warrant Forums

I have to agree with Alpha on the optimization part. I got about 50fps average and thats not too good . . .

I should average at least 100+ fps, with peaks at 150 or greater. Some maps I can get 300fps-500fps peaks @1920X1200 resolution with everything on the highest settings on my 8800GTX. (this is in single player mode.)

If you want help on the optimization part, I'd be glad to help.

On the plus side, I do like the new look. I would like to see it lit up just a tiny bit more. Those textures are really nice, and not to be able to see them because they are a bit too dark is a shame. Perhaps I need to dig out my Spyder Pro and recalibrate my screen which I haven't done in the past 6 months, I dunno.

DW_Wraith wrote:I have to agree with Alpha on the optimization part. I got about 50fps average and thats not too good . . .

I should average at least 100+ fps, with peaks at 150 or greater. Some maps I can get 300fps-500fps peaks @1920X1200 resolution with everything on the highest settings on my 8800GTX.

If you want help on the optimization part, I'd be glad to help.

On the FPS... I agree it's an issue. That this map chugs is kind of puzzling considering it doesn't seem all that visually intensive. The raw file size perplexes me as well, though I guess having 2 terrains and custom textures and meshes myLeveled can explain most of that...

Pretty much all of the changes I've made up to this point have been geared towards changing the layout for the new link setup so that this map doesn't turn into a huge last node stalemate. If you want to do some optimization work on this one Wraith, feel free, I've still got some work to do on Cosimia so I won't have a chance to do any on this until next week.

In my opinion the anti portals need work. From looking in the editor some of them don't look like they are doing anything other then wasteing cpu time. Also you have 4 terrain actors, 1 floor, 1 ceiling, and 1 for each tunnel. The ceiling doesn't need to be so detailed. Currently its a 256x256 height map, You should at least experiment with the bkcollisonhalfres setting. Id also look into lowering the resolution of the height map and leave it no higher then 128x128. Another option is shutting off collision for it and lower the softz point of the map, but i don't think that would feel right. The passages should be zoned like ahebban.

For game play, Those spinning shields need to go. The core is already half unexposed by being enclosed in that area. With the shield, You might as well leave the core unconnected in the link setup. The map's main gameplay issue was you couldn't get the primary up. Core hits were actually pretty rare. I didn't analyze weapon lockers or vehicle loadout.

No superweapons?

I'll get to work on optimizing it a bit. By Sunday sometime I should have a big part of it done, and will post it.

Sunday 11:40p.m.

I am putting the map in the download section for anyone to see. Its a bit better, but having a ceiling and the fact that the rock static meshes collision is figured on a per poly basis, its never going to be a "fast" map. I am mulling over the possibility of creating some ceiling panel meshes with built in collision models to help alleviate the problem, as well as possibly trying to make some collision models of the rocks, but for me they are hard to do without ending up with nearly as many polys as the origional. I will experiment with stuff and see what happens. The map filesize itself has shrunk 20 MB in size, and the framrates have gone up slightly, but until the mentioned problems are dealt with, it will not be a fast map. I will continue working on it for the next few days ...

http://www.clandw.com/Download/index.ph ... 20project/

Hey Wraith, just checked this one out...

First thing I notice is the map is less than half the size it was before. Whatever you ended up doing on that front definitely improved things a lot. Good job. One of my major reservations about this map before was how large it was and how much of a problem that would pose on redirect, seems like that issue is solved now.

Overall things are looking better for framerates. Only dipped low (~20fps) in the smoky area due to the particle effects there, but that's an issue that I know causes problems for my system.

Some things you might want to consider -- Use the spherical BSP brush to create a collision brush for some of the rocks and then turn off collision on the mesh (vertex edit the collision brush so that it generally fits the shape of the rock).

Replacing the whole upper area of the cave with specifically designed meshes/collision will definitely work ... What I'd probably do, just for the ease of doing so, is reduce the ceiling to 128^2, turn off collision entirely and just use some basic blocking volumes to block things high up.

If you're comfortable doing so creating some custom meshes for the caves instead of using the SM+Terrain approach that I did would be good -- I'm not really comfortable enough regarding modelling to create a mesh to fit existing terrain...

Next version if you finish up what you want to do with antiportals and collision and redo/reinstate the tunnels I think it's feasible to get this running on the server and get some real playtesting going for the time being.

This weekend I will work on the caves as far as replacing the stuff with a complete model as similar to it as I can. Then we can get rid of the terrain used in the caves entirely.

Depending on how much time this will take, I want to tackle the ceiling next, but it probably won't happen this weekend. I hope once the caves are done that the frame rates should improve enough to try it on the server.

OK, I put the latest rev. in the download section. I made the caves and took out the 2 terrains and did some other small things. See what you think, I know I can make a better looking cave, and can work on that and drop it in at a later time. I also added a couple of anti-portals also, but it probably can use more...

http://www.clandw.com/Download/index.ph ... 20project/

Sorry it took alot mlonger than planned, too many distractions...lol

Wraith -

Just went through this map a little bit. With the new ceiling I did see a little increase in FPS but not all that much. One thing I think you should pay attention to is how the ceiling is being lit, now -- In most places the ceiling looks okay but there are some portions where the low poly nature is quite apparent. I've taken some screenshots of the most noticeable ones, which I may put up later on. Basically though, I think the ceiling may stand an increase in vertices to make it look smoother. I also noticed you took out the holes in the ceiling, which is a little disappointing (At least in the middle).

Saw the new caves -- To be honest I prefer the old ones I made using terrain/SMs. The new ones are way too low poly. It's not attractive and they're so low poly that you can almost slide on the slopes. If/since you're concerned about optimization, what I would do is alter the flow of the caves from a zig-zag to a dramatic "S" shape (with cave entrances at either extremity of the "S"). This would prevent any external view from seeing more than a trivial amount of the cave and cut down on the potential render cost that a higher poly cave may have.

I agree about the caves and the ceiling. I'm just trying to figure out why the frames are so low, so I went with really low poly thinking that if that "mesh" is what is solely causing the issue, that it would stand out when changed. Unfortunately everything I've done hasn't done much to speed things up. As I've said before, you can revert to whichever revision you want, I'm totally fine with it and I always learn something from the experience. I did like the caves you created and ceilings. I thought maybe they were the cause of the map running slow compared to the origional, but I was wrong. I have all the revision from "b" to the last one, so if you don't have it handy, I can send you whatever ones you want. If you just want to use the last one you had, its all well and good and I'll happily replace the one in the download section.
If the mover is the main cause for the map being slow, then at least we finally found what part needs to be worked on in addition to learning how much a mover can impact performance on a map.

I just like to play with maps, sometimes I can help, sometimes I can't.... :compress:

EDIT:
I keep forgetting, but I do have one suggestion I would like to see, and that is instead of all the crystals being the same blue color, how about putting some ones that are colored red near the red base? It would add a bit of variety....or some "clear" colored ones instead/also???

DW_Wraith wrote: EDIT:
I keep forgetting, but I do have one suggestion I would like to see, and that is instead of all the crystals being the same blue color, how about putting some ones that are colored red near the red base? It would add a bit of variety....or some "clear" colored ones instead/also???

Yeah - On the Movers thing it's a definite learning experience. I'm kind of disappointed because I like the idea of using movers to create more dynamic environments. I guess the lesson I'll take from this is that movers are fine... As long as they're not constantly moving, or as long as they don't have collision enabled.

What I'd recommend is try going back to version RC1Kf and see how removing the movers around the cores might impact performance. If you gain an extra 10fps as Hornet speculated, then we should probably work off of that version. I would then try taking most of the optimizations you did in version RC1Ke, except for the ceiling SM and the SM caves, and see how that works out.

As for other aesthetic improvements, there's room for it, but mostly I just want this as a playable map. I have contacted BigKid Icarus about it and, while he hasn't gotten back to me in a few weeks, he did promise to do some work on improving this map. I think we should work on getting this map set gameplay wise to give him a good idea on what will work for 32p, and if he wants to do aesthetic improvements or extrapolate from the gameplay basics we set up with this thing, then that's up to him.