SkyBox lighting / Map preformance Created 17 years ago2007-01-06 18:35:13 UTC by Raa

Hi,

I'm currently working on a map and am in the last beta stages. Before I release this as final I was wondering if I could get some others thoughts. It a map for CS1.6.

First question:
Does the height of the skybox walls/ceiling determine or change any lighting characteristics when using an light_environment?

Second question:
Does the height of the skybox walls/ceiling effect map r_speed or performance in terms of user FPS?

No to both, but you wouldn't want to make it overly high for no reason at all, because I think it may increase compile time. Not sure about that though.

First question:
Does the height of the skybox walls/ceiling determine or change any lighting characteristics when using an light_environment?

No

Second question:
Does the height of the skybox walls/ceiling effect map r_speed or performance in terms of user FPS?

Erm.. i was under the impression it does.

Answer to second question:

It increases compile times since it creates more portals.
With more portals, the chance of a portal seeing another portal when it shouldn't becomes higher, meaning: yes, it can have effects in the r_speeds and performance.

If I might add, bigger areas always means more faces. In this case, making the skybox taller doesn't really add any new faces. I think it only becomes an issue when it comes to portal visibility.

you can have a skybox thats big, with say 800 r's, and make the skybox twice as big and have 1000 r's without adding any detail.

I fail to see why...Note, the only thing that we would be changing is the height of the skybox.

The SKY texture references six images, so it's basically a box that follows you around the map. There are no polygons being rendered in these areas where the skybox is, just images, so how could it have any effect on the r_speeds whatsoever? This is the same reason your r_speeds won't drop if you upscale the sky textures, it's because in the game THERE'S NOTHING THERE!

At least, that's what I've come to learn from every tutorial I've read.

I think Raa's question was more oriented towards knowing if making the skybox taller (as seen in my example pic) would have any effect on the r_speeds of a regular map.
If the added area is playable (which was not stated in the question), then the lines of sight are longer and might have more wpolies, as Orpheus said.
Either way, compile times will increase.

Thanks for everyones input. Yeah I was basically just trying to create the best possible experience for the players in the map. There is a huge relationship between server preformance, client FPS, bullet reg, and HitBoxes.

I guess I could post the map so I could get more specific advice on optimization, but I'm just not sure there is anything else that could be
done without changing the map dramatically. I've been mapping for 6years or so, and have done the whole decompile thing which has taught me a lot.

On a side note/question, From what I've read high detailed RAD settings should just increase compile times. However, has anyone else noticed that the Better the RAD lighting quality, the worse the map preformance?

anyway if anyone is interested in giving it a looksy please let me know.

Well, after opening up my test map to try and prove my theory, I found I was entirely correct.

I first took my basic box map, and put two blocks in the center of it, one with a target texture stretched across it so I could stand against my marked corner and look directly at it, so as to assure me that I had the same line of sight during every test. My map ended up looking like Kasperg's first diagram, but with the two floating boxes I mentioned.

On the first compile, I ended up with 271 w_polys. On the second compile, I raised the skybox very, very far up. I got the same w_poly count. Third time around, I placed several boxes and a cylinder with the SKY texture on them, and raised the hight of the skybox even more. In the fourth and final test, I raised the view distance of the map to 7000, just to make sure I was not being limited by it somehow, and there was no change in r_speeds at all.

Thank you, Orpheus, for using my correct title of "master srry".

And raa, I should probably point out that although having a bigger or more complex skybox does not amount to higher r_speeds (unless you having a big skybox actually allows you to have more brushes/models in view), I have heard that it decreases the FPS, however I'm not sure exactly how substantial of a thing this is anymore. Given newer video cards, I doubt you'd even even notice it.

P.S.

Orph, you have now earned the "You're Winner" trophy from me.

What a pointless discussion. Everyone was right some way or another.
In any case, in a fully playable map, making the skybox taller can mean more wpolies in terms of what the engine sees, even though it will visually look the same to the player.

You mean this parameter?

This assumes that you noclip to the top, and look down upon the map.

It's the only one that I could find, and it doesn't describe a realistic situation. When will you be noclipping to the top of any skybox? I've done that before, and you're right, it increases r_speeds the farther away you get (to a certain extent, anyway). I still don't see how this is applicable.

Given your example, the difference would be so slight, that it would be beyond notice.

Tell me what example you would like to see, and I'll make it.

What a pointless discussion.

You're absolutely right, but I just can't stand to be proven wrong without any evidence that ever I was. If I correctly understand what Orph is saying (which I admit is unlikely), then I'd have to disagree with it.

It did show the limits on where srry will stoop to name calling after all.

I never called you any name, you cunt.

To continue my post I was going to edit:

Anyway, this thread should probably just be closed with one of those smart-ass pictures admins seem to like so much, since it's really going nowhere. Orph is unwilling to prove his point, and I'm unwilling to admit mine is wrong, unless someone disproves me.

And to start my new post:

given the parameters I stated.

You said I right there. I not he! And I was conforming to those parameters!

I'm not sure it needs to stop "going anywhere" though.. I mean its only a day or so old. I think it would be beneficial for people to just answer back with what they've experienced in their map making, rather then PROOFS. We all know HL1 engine mapping is buggy, often giving different results when it should not.

Its a healthy worth while discussion topic I think. If done respectively, when is learning about performance optimization not healthy?

for example,

From what I've read high detailed RAD settings should just increase compile times. However, has anyone else noticed that the Better the RAD lighting quality, the worse the map preformance?

To answer your question about RAD quality versus performance, no, that shouldn't have any effect. The lightmap scales don't change, there's just more time put into generating them. I've never seen any difference. Of course, a fullbright map runs faster, though often the difference is pretty much insignificant.

Did you notice any difference then? How much milliseconds per frame difference?

Just a wild guess on your theory: did you have any entities that spanned a wide area? See, entities get rendered when their bounding box lies partially or completely within a visible leaf. Raising the sky might've caused some of such entities to fall within the current leaf (since the leaf was enlarged), hence upping the polygon count. Just a wild guess.

Do you still have the map, Orph? You could check with gl_wireframe 2 to see where those 200 poly's were coming from. That might clear up some confusion.

I thought for sure "raising the skybox" or having more visible sky brushes made the map brighter, but after a quick test map, found it NOT to be true.

Heh, I just tested that, and thought "WTF was Rowleybob thinking?"

The brightness is controlled by how high it is in the light_environment entity, not like texture lights.

I could have sworn on my dm_lars map -big skybox- raising or lowering it made a big difference. The brightness values were very low though--like 1--, so maybe that had something to do with it (I'll tell you in a second when this second compile finishes ;)).

edit:Nope! Absolutely no difference.

Just wondering, but did the r_speeds increase? This would be a "real" situation where it seems like it would show up if it made any difference.

Checking.

edit: Nothing appreciable--5 wpolys or so higher

Jeez, how could you even notice?

If you Increase the hight of your ceiling it WILL increase your r_speeds, Compile time and Frame rate. But not to much.

Not sure about the other question