Annoying R_speeds thing... Created 21 years ago2003-12-19 18:54:45 UTC by Marlborometal

Let?s say I have almost completed an enormous map for my mod and then, when I compile it without Rad and with a fast Vis I get HUGE R_speeds in some places like a non much detailed hallway and some parts of a room. It annoys me ?cos I never had this problem before so, what?s going on?. I?m trying to find the command to see what the engine renders in those points but I can?t remember or find it... Anyway, I have more detailed parts and get slower R_speeds, why?.
Thanks in advance.

Well, the more wpolys you have the slower it takes ...

Run full vis, if you do not have any errors, it should not take more than 15-20min even for the complex maps

Some usefull Commands:

gl_wireframe 2
r_speeds 1
r_draworder 1 (software only)
r_drawflat 1 (software only)

the last one is handy for seeng how your brushes have been split

Also, try experimenting with hint brushes, and some laylouts in your maps to reduce r_speeds

that's all i can think of right now..

Compiling right now Kol. I hope it helps ?cos I worked hard on this map :)... I?m also having strange "Leaf portal saw into leaf" errors in parts that there weren?t... Can it be I?m pushing hammer???. I sorted out the leaf portals saw into leaf errors but more and more appear, it?s strange.

Hmm... Usually these ace caused by complex brushes cutting into the adjasent walls... like pipes etc...
turn those into func_walls

Strange, when I compile the map with a cordon bound in the area where the error ain?t supposed to be it works fine... If I also hide and not compile the newest part it runs too... What could it be?.

The map as a whole

Hmm..

Make sure you got rid of all the brush errors etc..
then if you're still getting problems, try splitting your map into seperate maps, read my levelchange tutorial
^t3h pimpage

I sorted it... Deleted the part that caused the problems and pasted it again :P. It?s strange, but works. I can?t split my map due to some common things in it but I would if I could (by the way, I read your tuturial when came out and liked it ).
Oh, and I did a fast VIS (3 mins) and a Full VIS (more than an hour!!!!, I simply got tired of it, now I?m starting again). With a Fast VIS and the r_wireframe 2 turned on i see the engine is rendering the WHOLE map. Is this natural???!!!. I got r_speeds of 5 fps which is pretty bad so, with Fast vis it seems that it?s like not having VIS at all... Or is my map what gone wrong???.

Uuuh.. That doesen't look very healthy

Upload the map, so i can look at your laylout and vis blockers...

You should really be compiling maps with full vis for final compiles, since fast and normal usually leave errors in the visibility, which leads to some mishaps in the lighting.

A full VIS made it work almost properly... The layout is tricky since there are large spaces to see in front of the player, I mean, there are world brushes in between but these are broken by doors and they don?t act as vis blockers. The final result is that thew engine does not always render what it has too. Sometimes renders too much. But now works. Only 1:30 hours of VIS :(.

heh, only 2 peole in this thread...

Upload it.. or give it to me and i'll tell you what to fix.. i'm feeling charitable today.. tiz the sezon!

I think that if there?s a thing to fix is, evidently, the size of the map plus the size of a determined section that represents a huge, enormous cavers. It has loads of void space in the middle that takes ages to VIS. What I do is, given that this particular place works well, just hide it and don?t render it in the testing but, well, when I do a fast vis the place is fully rendered if you look at it (in the center of the map) and then comes the 5 fps.
I was wondering the following... Consider this:

-----------|  NOWHERE-VOID    |--------------------------
                |                           |
                |                           |
                 ---------------------

HULL 1 HALLWAY HULL2

---------------------

|                            |
|    VOID                 |

------------------ -------------------------

O.k... If I put some HINT blocks in the VOID parts, will it prevent some parts of the Hull2 to be visible from Hull1 and vice-versa?.
Thanks Kol.

PS: I?d upload it but since it?s near its completion I?d feel a bit silly about being unable to complete it myself.
PSS: Yes, we?re only 2 here.

It is getting a bit wierd indeed... Its like a ghost thread that only 2 people can see....

Or you can think of it as beeng really special and actually knowing what's going down here

Place hint brushes along lines of sight, as diagonals, and make then 'cut' or 'overlap' a bit into the walls to make them work

I can see it!

Yay gods!

Oh well..

Hello 7th. Welcome to the ghost thread.
I think the "R_speeds" word in the topic?s name is making people afraid of seeing it... So, placing Hint textures may greatly increase the performance of my maps but, still, I can?t understand fully what?s to be done... Let?s see, I take a look at the hallway in top view and place HINT textures "closing" the area the player is not supposed to see in the next room. It?s that?.

well basically, if you place them along lines of sight...

I really should do an example

You should... Example maps vault, maybe??.

Yep.. in exactly 48 hours

Once i get on a machine with hammer

I?ll wait anxiously.

It'll be your x-mass present

Oh!!!!. I nearly forgot that you get the presents on Christmas!. Spaniards get it on January, 7th :(.

What's that for?

I am assuming you have read R_speeds the tutorial?
I haven't replied because Kol is doing really well answering the questions
But for what it's worth, large caverns with lot's of textured brushes and lot's of space will increase compile times. Null anything the player can not see, but is actually in the visible part of the map. (Like if you noclipped around inside.... I am talking about the brushes you would see then, not the external brushes)
Also, when running r_speeds 1, you will only get a true representation when you do a full CSG,BSP,VIS and RAD compile. Failure to use all the compile tools will NOT give you a true indication of the speeds.
As for Hint Brushes... I tried to write a tutorial, but kept getting results that were not consistant and didn't follow any pattern. So rather than write about something that I couldn't figure out but have had a little success with, I decided not to do it at all. Hint brushes can increase the wpoly count if you are not careful.

I have found that the best way to manage speed is to build the map with them in mind in the first place. The addition of a few enlarged textures, a few func_walls and a few solid brushes to block the visible line of sight works wonders. Plan before you map ! Don't forget that the engine will think you can see over walls, so walls to block visibility in caverns are not always good.

Also, for those who tell you that a 13 Hour compile time is acceptable... they are full of it. There is no reason that a map should take more than an hour on a PII 400Mhz with 256Mb of RAM...... Actually that isn't quite true as there are some designer specific maps, but it is a good general rule. Case in point was an RMF that I tried to compile recently, 5 hours later it was still going. A little fixing reduce the compile to 15Min's....
Moral of the story.... If it is taking a long time, have a look at the rmf from the engines point of view and try to see why.

it doesnt take more than a minute on my comp:)

All good stuff

Null texture Warning: Wait till you get the HOM's OMG!!!!

Actually, if you really put your effort into it, you can compile really complex looking maps in munutes (excluding RAD)

My Vis times never reach even a half an hour on a PII 350mhz
granted, i have 320+MB Ram but...

So something?s evidently wrong with my map. 1:30 hours of VIS only :(. I don?t dare to do a RAD to it still.

Put it in the Vault, I'll have a look tommorow. If you remind me by posting here saying you have done so.

Thanks for your offering Alex, but as I told Kol, I?m about finishing it and it would be very sad for me to give up at this point. :).

Do I have to texture with SKIP non-visible parts of brush entities?.

Arghh!!!!. Help meeee!!!!. This is getting worse!!!. I?ve put some skip textures and now it seems that I have TONS (yes, tons) of Mixed Face Contents errors. I know what causes this error (mixing water and sky or water and some other texture and so on) so... What am I doing wrong?. Should I use the Skip texture???. Heeeeeelp!!!.

No,
Don't use skip unless you are making Hint Brushes.
Some people use NULL to texture all the non visible parts of brushes, but I have not proven that it actually reduces the wpoly count.
And as Kol mentioned, it can very easily turn into a Hall of Mirrors problem.

You really need to look at the map again, but this time from the engines perspective. What textured brushes does it think it can see. I know this is a damn site harder than it sounds, but it get's easier the more you do it.

I really do suggest that you let someone have a look at it. I will if you want. No need to throw it away.

O.k, o.k... :). We?re here to learn and to credit everyone :D. I?ll mail you my map.

I was going to post it but... It?s 6,4 mb of RMF!!!!!!!!!!!!. What do you guys think about this????. Do this size responds to the map?s compiling time?. (Still there are no point entities in the map, just the architecture and brush entities... I?m afraid of my own map).

WHAT I FIND USEFULL AGAINST HIGH RSPEED IN OUTSIDE LEVES IS MAKING A GROUND TEXTURE BY A SCALE OF 3 LOOSK MUCH BETTER

Taking it into account. Keep it coming :D.

rmf's zip really well...

and

NULL textures DO reduce wpoly counts,
I went form 850 in a corridor to 580 by using null and streching textures (Great, will do some testing) Andy

Another thing,
make sure that the v. of the compile tools you're using suppourts 'null texture stripping' and you have a NULL texture in one of your wads otherwise, its worthless.

my ver3 map compile in 1hour... if you can help me to reduce that number that will be great

btw kol you siad you woudl try to find some leeks in my tfc map ...did you find them??

O.k Kol but, what are NULL textures?. Where are them?. How do I use them?. Tell me about them, please, I only have those SKIP and HINT textures in my wad account. I?ll try to zip that. Merry Christmas.

The null texture you can make yourself if you need it..

just make sure the name of the texture is

Null

Basically, place null textures anywhere incide the map, but on faces where the player will never be able to see them,
on top of a really high ledge or something.

DieH@rd: It depend's on what's taking the hour.. some of my maps take up to 3/5 hrs...

Um.. the null texture gets stripped from the bsp, you can't see it
hence the reduction in r_speeds ...

The thing is that I did it and did not work. I named it "null". Should I name it "NULL" or "Null"???. I mean, maybe this thing is case sensitive. I?m also using zhlt 2.5.3. Any additional commands?.

If your using ZHLT 253, make sure you upgrade to ZHLT 253 custom build 1.7 as the Null texture is included, look in the ZHLT folder. the zhlt.wad is there and it has Null....

And yeah, I know it's actually MHLT, but let's not confuse things.

well I never had any r speed probs at all? its wiered some of my maps are very big! and have alot of stuff in them? but then I gotta watch out for that annoying paper clip digital pest that keeps springing up every few minutes! its evil incarnate i tell you, any hoo i might try some stuff with the null texture any ideas?

Right, bit late comming, but:
The null texture:
Its INCONCLUSIVE

The null texture indeed, works.

But, and this is a big but, it doesen't seem to make any big difference in the example maps, however, in larger maps, it makes an enormous difference. I won't include those larger ones as examples for obvious reasons...

It also has the effect of reducing bsp size, the first map, is 200KB (No null textures) the second one is 180KB (with null texture stripping)

This is a very small difference, but if you apply this to a larger map, this means you can reduce the bsp size by up to 20%

The wPoly reduction is a mere 30 to 20, very small, but this was with full vis.
Most people run normal vis, which means the reduction will be a noticable one.
However, the fps rate went up by 1.5 frames on average...

So you see, if you don't use it, you're not missing much, but if you do use the null texture, then be aware that it only makes a difference in large quantities.

(i'll upload those example maps later on)

And loque: Check your spec, if your comp is less than a month or so old, you can run every map with no problems....

And loque: Check your spec, if your comp is less than a month or so old, you can run every map with no problems....

yep its about 3 months old though not as powerful as my step dads comp who bought his a year ago

Ok.. there's your ans...

Anywho:
NULL Example maps

All your null texturings

More questions than answers i think, but make your own conclusions

Well Kol and Andy, I thank you very much for your help, as always. I also thank other people but mostly both of you, Kol for dedication and Andy for cleariness... I understood how they work and how do I use them. I also understood how to set some VIS blockers to reduce the polycount and also started to manage those tricky Hint textures. Thank you very much guys :).

wow, thanks