vis question

If you're looking for mapping help or you reckon you're a mapping guru, post your questions / solutions here

Moderator: Moderators

Post Reply
[-[usf]-] commando
Sergeant Major
Posts: 103
Joined: Sun Dec 15, 2002 4:09 pm
Location: australia
Contact:

vis question

Post by [-[usf]-] commando »

just wondering on vis
when u add the 2 vis boxes and join em
do they take up much vis data?
or do u do vis blocks in small blocks only?
Image
User avatar
diego
Captain
Posts: 245
Joined: Sun Dec 21, 2003 10:47 pm

Post by diego »

I'm not sure how much data it takes up, but I only used vis_leafgroups for large areas. My map is kind of Doughnut shaped, but you CAN go into the center buildings. So I used large blocks to make sure if you were on the 1 side of the map, you couldn't see buildings on the opposite side.

I think I spent two months trying to get it right so that I had no buildings or parts popping on and off. I had to combine it with a Blocksize of 128 in my compile to get it to fill in the holes.

I've got probably 2 dozen boxes in my map. I have to check, but I think my vis size is only around 600kb.
Diego
User avatar
tltrude
Chuck Norris
Posts: 4774
Joined: Sun Jul 07, 2002 4:03 am
Location: Oklahoma, USA
Contact:

vis

Post by tltrude »

Yes, it adds to the vis compile time. But, if it helps preformance, it is well worth it.
Tom Trude,

Image
User avatar
Jack Ruby
General
Posts: 1443
Joined: Fri Feb 07, 2003 12:20 am
Location: london

Post by Jack Ruby »

Tom, I need your advice.

My farplane is say 3000, I make leafgroups that are say 3100 in both X and Y dimensions, I cover the entire map with these leafgroups, they all butt up next to eachother perfectly to enclose everything.

I then compile with vis_derived 1 to kill the vis stage as supposedly I already did all the work.

What are the negatives of doing the vis like this ?

Is it better to just make specific leafgroups in areas prone to low fps and run a normal vis compile ?

Thanks in advance.

Jack
Philip II of Macedon sent a message to Sparta:
"If I win this war, you will be slaves forever."
The Spartan ephors sent back a one word reply: "If".
blue60007
General
Posts: 1247
Joined: Sun Mar 07, 2004 11:44 pm

Post by blue60007 »

Jack, I did this on a map for BCDA (Execpt I did 1024 blocks) and basically all of the map was covered. I only set up (did the targeting) in blocks that the player could get to, so the compiler was still doing some calculating. I had to do this because of a bad VIS design and with hardly anything visible I was getting 30's for FPS, once I did this I saw some areas jump from 30's to 60's or 70's (the inside areas). For the outside areas with 40's to 60's jumped to the 60's (note some areas wern't affected). This map is about half inside and half outside, so this wasn't a very effecient method inside but it got the job done. I would suggest doing this only on mainly open levels. Basically I used 1024 blocks instead of 3500 (with a farplane of 3400) was the fact that I had some large buildings that would be effective in blocking visibilty. I had greater accuracy this way. I hope this helps a bit...


In a nutshell, if your whole map is suffering I would set the whole map up with blocks and do the targeting in all accessible areas. If only certain areas are bad, set up leafgroups everywhere but only target the ones in areas that are suffering. There is no need to redo something the compiler has already done a good job of. You'll have to see if this messes up the already good areas up any. And to get back the original question, each leafgroup doesn't take up much data, but unless you, say put it to the top and bottom of the skybox you'll be spliting up portals a lot more, resulting in more VIS data.

Be sure to put the leafgroups (that are targeting) all the way to the top and bottom of the skybox, if and only if this won't mean putting the leafgroups into another playable floor above or below, if you doy ou won't be able to do a seperate VIS job up there. Doing so will prevent extra splits in areas you normally wouldn't be able to access.
Image
User avatar
tltrude
Chuck Norris
Posts: 4774
Joined: Sun Jul 07, 2002 4:03 am
Location: Oklahoma, USA
Contact:

man vis

Post by tltrude »

If you take all the man vis out and compile normally, you will see that normal vis does do some good (structural walls). Using tris 2, and walking around, faces will pop in and out of the render. You can also see the spots the the normal vis could not handle, and those spots are where man vis can do the most good.

So, don't just use one thing or the other--use everything you can find. Of course, compile time and better preformance are always going to be factors on opposite sides of the scale.
Tom Trude,

Image
User avatar
diego
Captain
Posts: 245
Joined: Sun Dec 21, 2003 10:47 pm

Post by diego »

I stumbled across this Tut for blocksize at planetwolfenstein, today. I've seen man-vis tuts here and on quake 3 sites before. They've all been a big help, But for some reason, this tutorial really explained the blocksize aspect to me clearer than anything else I've read.

And now I have a much better idea how to apply my vis_leafgroups in the future.


http://www.planetwolfenstein.com/tramde ... ksize.html
Diego
Bjarne BZR
Site Admin
Posts: 3298
Joined: Wed Feb 05, 2003 2:04 pm
Location: Sweden
Contact:

Post by Bjarne BZR »

A nice find Diego.
Admin .MAP Forums
Image
Head above heels.
Post Reply