Vis testing Created 12 years ago2012-05-30 23:00:53 UTC by camplo

So I did some testing today and made some discoveries. My goal is to find a way to make vis do what we want without the hassle of hint brushing. Even with the node set to 64 vis isn't as tight as I expect. I have a hard time believing the maxnode 64 literally sets the max node size to 64units else you would never be able to see more than 64 units around a corner. It would be like a 64unit thick 45 degree angle around your basic hallway 90 degree turn. Thats not what it does thus I don't believe. Anyway. I made 3 rooms. Room 1 was the control, room 2 had 32 8x32x24 blocks covered in hint at one side of the room and finally room 3 I copied all 6 brushes of the simple room, pasted them in place and covered them in hint. Results Room1 - w_poly 140 no matter which way you looked and where you were in the room, Room2 - w_poly 156 when looking at the side of the room with the 32 hint brushes and 94 when looking at opposite side, Room3 140, same as room 1.
So I was like oh really room 3.

I went back and added 4 164x64x64, textured, floating world brushes to room1 and then in room3 I added the same 4 floating blocks and put inside of those blocks identical sized hint brushes and that's when the magic started. Room1 w_poly while looking at blocks 170, looking opposite direction 96, room3 159 w_poly and 85 w_poly while looking away from them. Sounds like a good start on an alternative way to hint brushing your map. I tried it on my map to see what it would do. I hid all the entities, selected all the world brushes, duplicated them and covered the duplicates with hint. It performed well, I think, considering I did the whole map in about 5 seconds, parts of the map I should not of seen, where not drawn and my w_poly dropped accordingly.

Am I confused? sure. But I think I got it figured out. Hint blocks do not add faces to the w_poly. Hint brushes do chop adjacent faces which is how we end up with only 16 added faces to room2. Also noteworthy item is the floating blocks in room1 did cause cutting of the adjacent walls. By looks of it the cutting was less than if the blocks had of been touching but non the less, they still caused cutting.

I have yet to try other variables. I am interested to see what will happen if I enter a large number for the maxnode size. It would cut down on vis cutting up the world. If my style of of "hint brushing" can still cut out the unseen parts of the map, I may have a winner.

And the results are in. I turned the maxnode size to 8000. Room1 w_poly lowered to 152 from 160 while looking at blocks and raised from 96 to 109 while looking away from blocks and room3 was identical in performance. Room2 hit 140 looking at the side full of 32 floating hint brushes and 103 while looking away. It was also apparent that no additional cutting took place from hint brushes nor floating world brushes.

My wacky do hint brushing perform best with the lowest node size. Cuts were saved with a high node size but vis performed more accurately with 64 node size. A brush cover in hint adds no faces other than the cuts caused from vis and when placed inside a world brush of identical geometry, makes the same exact cuts the world bush would have made But further influences the accuracy of vis, drawing what you see, and not drawing what you cannot see. In the terms of multiplayer, vis accuracy is very important. Once e_poly hits high enough a w_poly of -2 wont save you. I think my style of hint brushing works well as long as the maxnode size is set to 64.

I added the map I used to conduct the test. Green = room1, blue = room3 and Red = room3. The bsp is compiled at maxnodesize 64 to show the effectiveness of the hint brushes.

http://twhl.info/vault.php?map=5802

If allowed I'd like to organize all the information and add it to the tutorial database.

Dude, don't waste your time, it's 2012...

Bruce , calm. Goldsource will still be used because of how simple it is to make stuff for it. No VMT's, no normal and bump maps , and detail isn't even required. Let him do what he wants.

Bruce isn't saying Goldsource shouldn't be used - he's saying there's no need to go to these extents to optimise it, and he's right. This whole thing is purely academic because it has no actual impact on the player's experience.

Yes, better spend that time on mapping itself...

Actually, there is so little about mapping itself and the philosophy and theory behind it in TWHL..... Hmmm..

lol, you know what. You've said that before. And I want to agree with you.....but whens the last time you've made a map for an online server? One with a high e_ploy count per player. 2 players could mean the difference in about 6-4 fps...ok, I get your point but the technique I just proved IS effective and it takes about 10 seconds to do...so why not? There are situations where hinting or setting the maxnodesize correctly could mean the difference between 100fps and 50-60fps. You might be one of those people who think 25fps is playable. But to the hardcore competitive gamer they want every ounce of fps.

Not to mention, When did you guys give up on goldsource. Your opinions of gold source are way out of line. Why? Because goldsource IS the most used FPS engine online, 2012. Period. CS 1.6/HL and its mods continue to be the MOST PLAYED FPS GAME ONLINE.

If its really "2012" like you say it is. Put your map were mouth is. We have one professional mapper and there's room for more. Come on down to ZOD server learn the zombie mod game play and make a map that not only isn't hinted, but also built vis unfriendly and I will show you a server that is lagging to hell.

ps- the average e_poly per player is about 2000....GL

Very few custom maps are played by pro players.
I would agree with Bruce that it is 2012 and unless you have, i don't know a Pentium in your system its not worth it.

I think it's a pretty neat discovery. I haven't run into issues like this myself, but I don't see much point in disputing it either, especially if there's a group of gamers that would benefit from maps using this technique.

If allowed I'd like to organize all the information and add it to the tutorial database.

Go for it!
Use our tutorial proposal system. Note that you don't need a full tutorial in the proposal stage - just an outline of what you'll be writing. Once we approve it, you'll be able to write/publish the complete thing.

Steam says I have about 60 hours worth of cs time in the last 2 weeks. I have been mapping for several months now, understand how to use r_speed, wireframe and the like, I know what e_poly and w_poly represent, and have been staring at net_graph 3 for the last 12 years...... and when I say I know what I know from experience, not he say she say, steam aint lying.

Thanks Daub for the info.

Congrats

60 hours over 2 weeks? Jesus, dude, that's gotta be dangerous. I've only played 84 hours of Skyrim the entire time the game has been out and I considered myself addicted to that.

Thats with working 40 hours a week. See you not even hardcore gamer. How can you satisfy an audience that you cannot relate to?

40 hours? That's less than 2 days a week.

You can. You go buy a beer and fuck the world!