Sorry your browser is not supported!

You are using an outdated browser that does not support modern web technologies, in order to use this site please update to a new browser.

Browsers supported include Chrome, FireFox, Safari, Opera, Internet Explorer 10+ or Microsoft Edge.

Bug Reports / some issues of collision

Author
Message
DidRocks
20
Years of Service
User Offline
Joined: 28th May 2003
Location: France
Posted: 16th Aug 2003 01:43
Hardware Info:
PC Speed: 1.5Ghz+
OS: Windows XP Pro
Video Card: Geforce4 TI 4600 - RAM: 128Meg

I have some issues with collisions with my .3ds with dbpro and this patch whereas i have absolutely no problem in dbc

(Ianm you should recognize this code ^^)
some screenshot:
- in dbc


- in dbpro

Here the code (with, in remark, some changes for dbc)

i tried also to convert my .3ds in .x, but i have the same issue, unfortunatly... Here is the object:
http://ff7tss.free.fr/DidRocks/temp/FFVIICol.3DS

2 changes beetween dbc and dbpro (which are repeated in the code)

rem for dbc, remove this command and replace by: set camera fov 3.14/3.0725
set camera fov 73.74

and

rem why y=7 with dbpro whereas my objet is 6 height and work in dbc with y=6?
y=7

IanM
Retired Moderator
21
Years of Service
User Offline
Joined: 11th Sep 2002
Location: In my moon base
Posted: 18th Aug 2003 03:16
The object collision is way out. Position the camera at 7,60,0 and point it at 7,0,0 and this really shows.

Not being a modeller, I can't go any further with this, coz I don't have the software ... unless someone wants to convert the object to .x format and either try it with the code or post it to me to try myself ... pretty please
DidRocks
20
Years of Service
User Offline
Joined: 28th May 2003
Location: France
Posted: 19th Aug 2003 05:57
Here it is:
http://ff7tss.free.fr/DidRocks/temp/ffviicol.x

I converted the .3ds with the sdk directX 9.
(there is an other difference: the plain which constitutes the ground isn't at the same place, can you see it in my two screenshots?
Thx IanM for your time you'll look at this!

Best Regards
Ian T
21
Years of Service
User Offline
Joined: 12th Sep 2002
Location: Around
Posted: 19th Aug 2003 06:05
Wow, that's nasty stuff...

--Mouse

Famous (Avatarless) Fighting Furball
Rob K
Retired Moderator
21
Years of Service
User Offline
Joined: 10th Sep 2002
Location: Surrey, United Kingdom
Posted: 19th Aug 2003 17:33 Edited at: 19th Aug 2003 17:34
I get the same results here. I'm pretty sure that this is a bug as it were - although it is probably more a case of "feature needs improving".

I think that in general DBP could really use a better collision system and poly to poly collision would be very useful as well.

LeeBamber
TGC Lead Developer
24
Years of Service
User Offline
Joined: 21st Jan 2000
Location: England
Posted: 26th Jun 2004 01:25
As stated, feature in need of improvement. Who knows, perhaps a collision-week would be a good idea. The source modification is the best DBPro can do at present, aside from the new intersect raycast options and the improved rotated box auto collision.

"We are the knights who say...eki eki eki fatang loopzoing, zanziga....ni"
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 29th Dec 2004 14:23 Edited at: 30th Dec 2004 16:46
I think this is the same bug. Box to polygon collision is off by half the objects (box collision object) size when moving in the +X direction (tested with different size cubes and complex levels all the same result) Here is code showing it. Arrow keys move around. Press any mouse button for close up of collision.



Seems to be perfect in the other 3 directions.

And Sphere to polygon collision is off by half the objects (sphere collision object) size when moving in the +X direction and 18.75% off in the +Z direction (tested with different size cubes and complex levels all the same result) Here is code showing it. Arrow keys move around. Press any mouse button for close up of collision.



The other 2 directions seem dead on. Also with Sphere to polygon and Boxes to polygon, when the object is halfway through the poly it shows no collision in all directions. You can use those same codes to see it.

[edit] The y axis has the exact same errors as x with box to polygon and sphere to polygon.

Sphere to polygon


Box to polygon


[edit2] Sphere to sphere collision is really close in all 3 axi.

[edit3] Just found another error. This is a very serious bug. If you can't get the collision checking working right at least get it so it gives ar error message or doesn't crash. For some reason complex polygon objects touching (like loading a level broken into several objects) when you call collision checking commands on them crashes the program. It took me a while to replicate this one as it works on most of my levels but not all. Here is a code showing it working (with cube objects) and in the next code it crashes (with spheres). comment out the 2 lines calling the object collision() commands and it no longer crashes. Another reason it was hard to find as I don't always check my level objects for collision (usually just movable objects).

Working code but the collision on the x-, x+, y-, and y+ directions is half the objects width off @ center to center collision (coners are very dodgy).,



Crashing code as soon as the spheres touch it crashes,



Zone Chicken
20
Years of Service
User Offline
Joined: 25th Jan 2004
Location: `~-..-~`~-..-~`
Posted: 2nd Jan 2005 11:18 Edited at: 2nd Jan 2005 11:22
Here's one about as simple as it get's, loaded a box as a area, with two spheres, and a box inside, turn on collision, set it to polygons, and then autocam collision. You can hit the area box, and the box inside the area fine with the camera, but when you hit a sphere the program just close's. Same with .x sphere's loaded as well.

Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 2nd Jan 2005 12:54
Yes this collision system is bordering on useless. I have so many work arounds for alot of that that the program grinds to a halt.

Pricey
21
Years of Service
User Offline
Joined: 22nd Feb 2003
Location:
Posted: 2nd Jan 2005 22:19
Quote: "Yes this collision system is bordering on useless"


Hear, hear

:: 1.8 gHZ AMD Duron :: 256 MB Ram :: 32 MB SiS Graphics ::

:: Current Project: Sea Bound ::
blanky
19
Years of Service
User Offline
Joined: 3rd Aug 2004
Location: ./
Posted: 28th Jan 2005 06:27
*Cuddles DBC disc*

Avatar - white cat in a snowstorm. Look closely.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 28th Jan 2005 12:56
Yes I am buying DBC soon. Even though it is slower. It is more stable and can run on older systems. That way I will have DBP and DBC.

Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 29th Apr 2005 15:33
This was fixed in the 5.8 sdk version but is broken worse than it was in the 5.8 public version. Just run the examples I posted. The collisions don't even register in some directions

OSX Using Happy Dude
20
Years of Service
User Offline
Joined: 21st Aug 2003
Location: At home
Posted: 29th Apr 2005 17:04
Mike knows about it and an update to the update of the SDK should be out (hopefully) soon.

Visit http://homepages.nildram.co.uk/~nickk
Calm down dear! Its only an election...
Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 29th Apr 2005 18:14
The collision problem with the SDK is not related to this. That was down to some internal casting.

Will check this one later.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 25th May 2005 05:44
Will this be looked at for DBP 5.9? NGC is all busted up and that leaves just this for alot of people.

Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 1st Oct 2005 07:07
Just a bump for the bug fixing. This really need fixing again.

Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 1st Oct 2005 17:31
The examples that crashed previous versions no longer cause a crash in 5.9.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 1st Oct 2005 21:42 Edited at: 1st Oct 2005 22:32
Yeah the crashing is fixed in 5.8, the error that still exists is the collisions not registering in most directions. I will make a new example showing it and post.

[edit1] Here is an example of sphere to polygon and box to polygon. Collisions are perfect in the +x and +y direction but do not register at all in -x and -y edges.



[edit2] This shows the +z collision edge has perfect collision while the -z edge is off by 1/2 the objects size.



Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 1st Oct 2005 22:46
That seems to be the ones left messed up. Others seem fixed.

Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 1st Oct 2005 22:55
Those 2 are still a problem. Thanks for the code examples. Will see what can be done.
Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 6th Oct 2005 21:40
Here's an update on the state of things in the next beta of upgrade 5.9 -

* Lost In Thought, the examples you posted all now work correctly
* Zone Chicken, your example no longer crashes

* Sphere / polygon collision is corrected
* Box / polygon collision is corrected ( you cannot use rotated boxes yet )

I have tested sphere / polygon and box / polygon in lots of examples with high polygon shapes and it's turned out well.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 6th Oct 2005 22:48
Quote: "you cannot use rotated boxes yet"


Does that mean that you cannot rotate your object or the collision zone just won't rotate with it?

Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 6th Oct 2005 22:50
Yes thats right - the object that uses box collision can be rotated but the collision zone is based on the unrotated box.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 6th Oct 2005 22:54
That will work quite nice for what I am doing Rotatable box collision would be nice but not that critical for me ATM.

Mike Johnson
TGC Developer
21
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 6th Oct 2005 22:55
Was going to release today but wanted to do more testing to be certain things are right. Done lots of tests and I'm happy with the way things are working so hopefully you will be as well As soon as a few other bugs are dealt with tomorrow the next beta will be out and gives everyone chance to take a look over the weekend.
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 2nd Dec 2005 21:06
Polygon collision is still busted in 5.9 final. I am assuming the new collision system will be in the next update, whenever it comes out?

Paul Johnston
TGC Developer
21
Years of Service
User Offline
Joined: 16th Nov 2002
Location: United Kingdom
Posted: 2nd Dec 2005 21:29
Collision is slated for the next update, but I wouldn't like to guess at a date
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 2nd Dec 2005 21:50
I'll send you an email sparky.

LeeBamber
TGC Lead Developer
24
Years of Service
User Offline
Joined: 21st Jan 2000
Location: England
Posted: 22nd Mar 2006 16:31
Please update this collision subject soon with specific bug reports on U6 beta issues, as progress has been made in this area. This rather long debate on the collision system is closed down to keep things recent.

"Small, smart, and running around the legs of dinosaurs to find enough food to survive, bedroom programmers aren't extinct after all "

Login to post a reply

Server time is: 2024-05-08 22:55:08
Your offset time is: 2024-05-08 22:55:08