I Need A Better Program For Collision Hulls
Fortuna_Wolf
Join Date: 2003-02-02 Member: 13033Members
Join Date: 2003-02-02 Member: 13033Members
<div class="IPBDescription">Look at the screenie.</div> <img src='http://www.truffula.net/~fortuna/Graphics/ns_fw100000.JPG' border='0' alt='user posted image'>
I can understand some slight issues with the way the collision hulls are calculated, but this is abosfsckinglutely ridiculous. I can stand 10 feet off the edge of the stairs and still float on its collision hull. On the stairs that are straight on the X and Y it doesn't happen, but half of the steps are angled at 45 degree angles and this happens on all of them.
HELP!
I've spent all day working on this one hive trying to come up with something resembling the concept. First it wasn't vising correctly and I was still getting polycounts of 900 staring at a wall, and now I'm trying something else and the collision hulls are messed up
I can understand some slight issues with the way the collision hulls are calculated, but this is abosfsckinglutely ridiculous. I can stand 10 feet off the edge of the stairs and still float on its collision hull. On the stairs that are straight on the X and Y it doesn't happen, but half of the steps are angled at 45 degree angles and this happens on all of them.
HELP!
I've spent all day working on this one hive trying to come up with something resembling the concept. First it wasn't vising correctly and I was still getting polycounts of 900 staring at a wall, and now I'm trying something else and the collision hulls are messed up
Comments
HELP!
I've spent all day working on this one hive trying to come up with something resembling the concept. First it wasn't vising correctly and I was still getting polycounts of 900 staring at a wall, and now I'm trying something else and the collision hulls are messed up <!--QuoteEnd--></td></tr></table><span class='postcolor'><!--QuoteEEnd-->
What version of the tools are you currently using? Looks like you've got a case of the first clipping bug that my stuff is supposed to combat (described at <a href='http://www.xp-cagey.com/articles/clip_hull' target='_blank'>http://www.xp-cagey.com/articles/clip_hull</a>).
If you're using 1.7p10, what cliptype setting are you using?
EDIT: removed img from quote, added link to article.
And thanks for that article.
I'm busy combatting another problem right now... I have two weldables. two doors, a changetarget and a relay. One weldable sets the relay which triggers a door. the other weldable triggers the changetarget which changes the relay's target to the other door. right now though it seems that the changetarget isn't working at all... I'll probably have this fixed by the time you respond though.
And thanks for that article.
I'm busy combatting another problem right now... I have two weldables. two doors, a changetarget and a relay. One weldable sets the relay which triggers a door. the other weldable triggers the changetarget which changes the relay's target to the other door. right now though it seems that the changetarget isn't working at all... I'll probably have this fixed by the time you respond though. <!--QuoteEnd--> </td></tr></table><span class='postcolor'> <!--QuoteEEnd-->
-cliptype is a command line argument for CSG, you should probably start with "-cliptype precise" for combating that specific problem. The default value uses the same code that traditional compilers use and doesn't fix either bug--it's set that way for backward compatability.
come on, use the proper hull file in the ns folder under something like ns_hul.txt (it has 3 lines of info in it)
amckern
come on, use the proper hull file in the ns folder under something like ns_hul.txt (it has 3 lines of info in it)
amckern <!--QuoteEnd--></td></tr></table><span class='postcolor'><!--QuoteEEnd-->
Wrong answer. See above.
EDIT: Having the correct hull file won't fix the invisible ledges that are displayed in that image unless you change clip generation methods. Try it yourself.
While doing a test of a room for the map I'm working on, I found several unusual clipping errors. In one part of the map, there was a 'block' of clipping, so that if you got ontop of it you could walk around on it. I went and put in a CC - to make a screenie like Fortuna - and the entire level became non-solid! Except for the 'block' of clipping, which was still there exactly to same.
I read through this thread, and double checked my settings. I've been using -cliptype precise and the latest version ( 1.7p10 ) tools. On a hunch I went and removed that, so it compiled on whatever the default cliptype is. Bingo - The clipping worked perfectly ( There are other problems but that's beside the point ).
I'm going to leave cliptype precise off for now, but from reading about what the precise setting does I'd want to use it rather than not. Any ideas? I'm going to play around a bit and see If I can find the problem.
Edit:
I went and made logfiles of each of the 2 cliptypes, there is one difference. On cliptype precise during hlcsg I recieve the folowing errors: <!--c1--></span><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1-->0 brushes (totalling 0 sides) discarded from clipping hulls
CreateBrush:
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 0
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 2
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 32
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 34
(7.58 seconds)
<!--c2--></td></tr></table><span class='postcolor'><!--ec2-->
Without cliptype precise there were no errors. Is the cliptype creating errors or merely looking more closely and only reporting them?
While doing a test of a room for the map I'm working on, I found several unusual clipping errors. In one part of the map, there was a 'block' of clipping, so that if you got ontop of it you could walk around on it. I went and put in a CC - to make a screenie like Fortuna - and the entire level became non-solid! Except for the 'block' of clipping, which was still there exactly to same.
I read through this thread, and double checked my settings. I've been using -cliptype precise and the latest version ( 1.7p10 ) tools. On a hunch I went and removed that, so it compiled on whatever the default cliptype is. Bingo - The clipping worked perfectly ( There are other problems but that's beside the point ).
I'm going to leave cliptype precise off for now, but from reading about what the precise setting does I'd want to use it rather than not. Any ideas? I'm going to play around a bit and see If I can find the problem.
Edit:
I went and made logfiles of each of the 2 cliptypes, there is one difference. On cliptype precise during hlcsg I recieve the folowing errors: <!--c1--></span><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1-->0 brushes (totalling 0 sides) discarded from clipping hulls
CreateBrush:
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 0
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 2
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 32
Warning: Illegal Brush (edge without opposite face): Entity 0, Brush 34
(7.58 seconds)
<!--c2--></td></tr></table><span class='postcolor'><!--ec2-->
Without cliptype precise there were no errors. Is the cliptype creating errors or merely looking more closely and only reporting them? <!--QuoteEnd--> </td></tr></table><span class='postcolor'> <!--QuoteEEnd-->
Thanks for that error listing -- I think I have the problem narrowed down to the code responsible for determining the direction that bevel edges should face; the error you're getting supports that (it is most likely a result of a bevel face pointing 180 degrees from the correct direction, which is going to result in an illegal brush). The webbed tools thread goes into some technical detail about my current approach for a solution... I'll hopefully have something you can try by the end of the week.
Could you attach a screenshot of the brushes the error lists taken in the editor, preferably selected? I'd like to see what brush shape is causing the problem.
<img src='http://www.geocities.com/shadowics/brush.JPG' border='0' alt='user posted image'>
<a href='http://www.geocities.com/shadowics/brush.htm' target='_blank'> - Click here if it doesn't work - </a>
Edit - I did some more testing - The 'block' of clip I've been referring to is connected to the brush beneath it that you would normally walk on top of, sort of like it's the clip for that brush, only many times larger than it should be, remove the brush and the clip vanishes. However, the other clipping errors remain. Also the floor of the large cube I put around it is not entirely non-solid, only part or parts of it are. I'm not sure how I could find what parts except for walking around and falling through a lot. This is confusing though, as there aren't any other brushs near the floor except the sides of the large cube.
Edit - *image link*
<img src='http://www.geocities.com/shadowics/brush.JPG' border='0' alt='user posted image'>
<a href='http://www.geocities.com/shadowics/brush.JPG' target='_blank'> - Click here if it doesn't work - </a>
Edit - I did some more testing - The 'block' of clip I've been referring to is connected to the brush beneath it that you would normally walk on top of, sort of like it's the clip for that brush, only many times larger than it should be, remove the brush and the clip vanishes. However, the other clipping errors remain. Also the floor of the large cube I put around it is not entirely non-solid, only part or parts of it are. I'm not sure how I could find what parts except for walking around and falling through a lot. This is confusing though, as there aren't any other brushs near the floor except the sides of the large cube. <!--QuoteEnd--> </td></tr></table><span class='postcolor'> <!--QuoteEEnd-->
After the brushes are copied with expanded dimensions for use in the clip hull, the tools attempt to use CSG to create a single list of faces that will be used to create the final clip hull. If a brush has an illegal shape (possibly due to the bug in precise), it will cause unpredictable behavior when the hull faces are being generated. This is probably why a simple floor brush appears to have holes in the clipping hull-a bad automatically generated brush is being substracted from the floor, causing the hole.
I did have another thought, so just to make sure, I exported the map to a .map and opened it up in Hammer. The map compiled fine again with cliptype precise, same as in Quark, except that in game the clipping errors had 'moved'. There where still non-solids and invisible walls but either they were in different positions or new ones had been created. I ran the .map in Quark again and the positions matched what I had gotten before in Hammer - it seems changing the file type affected the errors.
P.S. - Edited the image in the previous post - hopefully you will be able to see it now if you couldn't before.