I've been trying to optimize my recent map, Narrow Extent, for inclusion in the 2v2 FHThrowdown contest. I began noticing that at the beginning of a new round (in split screen), there didn't appear to be any frame rate issues, but at some point after moving around through the map, frame rate issues would begin to appear even though I had checked the same views prior and found no problem. View down main hall On Narrow Extent, I know where the problem areas are for frame rate issues. Issues arise when looking down the long hall on the main level and when looking across the lowest level from either end. What I was doing to try to remove the frame rate issues was placing the lower screen player at one of the two problem areas where the player has the most objects in his view. Then I'd run the top screen player around the map. At first I was pretty happy with the revisions I made. There appeared to be no frame rate problems when both players in split screen had a maximum amount of objects in view. Soon though, after running through different parts of the map for a bit, frame rate issues that weren't there before began to arise. I'd have both players looking at the same exact views as they were at the beginning of the round, but now the frame rate was slowing. Interestingly I've almost never found any frame rate issues in the spiral halls, which are made with over 20 braces each, but I'm convinced that after leaving one of the players in a single spot with a maximum number of objects in view (such as looking down the main hall from one end to the other) and running the other player around the map, particularly through the spirals and bringing that player back to the original spot looking down the hall from the other end, the second player, that ran around, retains memory of the objects he passed by and it actually affects the frame rate. [was that a paragraph length sentence I just wrote?] I've done this experiment several times on the map and it always seems to happen. There are no frame rate issues at the beginning of the round with players at specified locations/views. After running one or both of the players throughout the map and bringing them back to the original locations/views, frame rate issues occur. Sorry for the long, drawn-out explanation, but has anyone else ever noticed this? I'd never read about it in any of the frame rate issue explanations I'd seen before. Could it be possible that characters are retaining map information to improve redraw time when they return to certain areas? If so, it makes the issue of dealing with frame rates a lot more complicated than I initially thought. It may no longer be as simple as ensuring that any one individual view isn't showing too many objects.
One thing i noticed is that placing alot of spawns will cause frame rate issues. typcially i only have frame rate issues in forge world but in customs it is never a problem. but it has alway made me wonder if it has any affect on frame rate outside of forge. ALright im a gurrila time to go get some bannas
I too have found that, in my Throwdown submission the spawn objects and objective items cause frame rate loss but in customs the map is fine.
Forge has more visible objects, naturally worse framerate. That issue you noticed could be very interesting, almost as if the game has an object "Cache" that is causing problems with framerate. I can't really figure out what else could be causing it, and it seems you came to the same conclusion. I'll be sure to look into it, because it could mean total objects period (with less relation to object density than we expected) has a hand in this as well.
@ Deathangle and Paranoia, I forgot to mention that the scenario I was describing was in custom games, not forge. I usually don't worry too much about the frame rate issues in forge when it's rendering all the spawn points and objectives because I know those objects won't be rendered during actual gameplay.
Here's an experiment. Find a corner. Without looking around it, face into the wall. Strafe out of the corner so you can see what's around it. Depending on the map, you should see the objects in the distance briefly flicker or change -- that's the game swapping from the low-LOD models to the high-LOD models, and it's only visible for a fraction of a second. Strafe back behind the corner. Wait a short moment. Strafe out again. Depending on the map, the switch between models may be much less (or not even) noticeable this time. Maybe I'm just imagining that the swap time is shorter, or... Maybe the game is caching things, though I can't imagine why it would. (I also don't know why such a cache would be kept for more than ten seconds, or why it would be allowed to expand seemingly without limit -- both of which seem to be happening given the description that you've given, OP.)
Honestly, this is the first time I've even heard of this type of problem, the most problems I've had with framerate have been with using glass and lights on natural geometry that moves (trees, water...)
i know you said that you hadnt had any issues with the spirals hall but... dude... its 20 decoratives smushed into one space. just as another experiment, why not try changing that to a normal hallway made out of a couple simple pieces. i would think that this would almost guarantee an improvement in performance. is the spiral hallway really that key to the gameplay of your map, or is it just an attention grabbing aesthetic feature. i dont mean to be rude but i am guessing the latter, meaning it is not truly needed.
I'm leaning to what **** said, 20 braces in a small hallway is probably going to end up causing more harm than good.
I already have a revised version, the one I intend using for the 2v2 contest, without the spirals and still run into the same problem. I have the map down under a $5,000 budget even. Regardless of whether the spirals are there or not, there is still the issue of frame rate problems accumulating as more of the map is run through. As I said though, I haven't noticed the frame rate issues while looking down those halls. I'm pretty sure that's because those are the only 20 or so objects seen at one time. If we assume, as we previously have been, that the frame rate issue is purely an object density within current view issue, then the spirals shouldn't hurt anything that bad since they aren't seen from the rest of the map and you don't see the rest of the map from within them. While they are just as aesthetic as they are functional pathways, they do not appear to be the culprits here. Even with them, the original maps budget was just over $7,000, far less than many maps out there that are said to not have frame rate issues. Also, the original map doesn't suffer from frame rate issues outside of split screen mode. I'm now simply trying to get the map to play perfectly in split screen mode.
Perhaps this caching issue only happens in Split-Screen? Maybe it's a way for the engine to deal with rendering two screens at once. Have you tried running similar tests Solo?
I haven't noticed a single issue in solo screen play, especially with the revised map budget down under $5,000. Solo screen even plays fine with the spirals.
Dost thou have any lights, FX, glass, or objects that emit light? If not, which I assume to be true, then if you have some of the original map viewable though windows or holes or what-not, that may affect it as the game has to render the different views of the geometry in the distance. Honestly no clue other than that seeing as my throwdown map has 2k budget left with 3 FX, glass and has no framerate problems splitscreen or otherwise.
My map has no lights or effects and only a few objects with lights (I think I've knocked it down to 4 railings in the entire map). Those objects with lights are at the perimeter and not in view when looking across the map. I hate to be the bearer of bad news, but I tested your map out and it does suffer from a good amount of frame rate dropping in split screen when the players are at the perimeter of he map looking to the opposite end. Regardless of the frame rate issue, I really loved your map. Check out what I had to say in your map's thread.