Weird artifacts when looking at a certain point

Crashy

20-03-2006 12:52:03

Hello, I get strange artifacts when I look at a certain point on the terrain. This point is not determined. I've this result since I updated to 1.2 RC1(non sdk). I used the sources avaible on your site, for the SDK, until cvs works.

Some pics(a part of the terrain is stretched and pop, that is really really bad looking).

http://crashy.cartman.free.fr/wip/Screen_1.jpg
http://crashy.cartman.free.fr/wip/Screen_2.jpg
In this one a tile of the terrain is not visible, I suppose it is this tile that is stretched and create the artifacts
http://crashy.cartman.free.fr/wip/Screen_4.jpg

As you can see a "bad" part of the terrain is coming on the "good" part. I've this result only in this zone.

tuan kuranes

20-03-2006 13:10:12

Never seen that !
check if you using same map you an reproduce that using pslm2 demo.

if yes, Can you mail me the terrain and config file ?

ps: check very latest SDK (just uploaded now) for updates on shadowreceiver updated code.

Crashy

20-03-2006 13:21:05

Mmhhh I cannot launch the paginglandscape sample I've compiled(crashes with nothing on the log, the project file must have errors).


I tried your very last precompiled demo, but I get this

-----------------------------------
Details:
-----------------------------------
Error #: -2005529767
Function: D3D9GpuProgram::loadFromSource
Description: Cannot assemble D3D9 shader DecompressVertexShadowReceiver/VP Errors:
(39): error X5326: Read of uninitialized components(*) in r2: *x/0 *y/1 *z/2 *w/3
(40): error X5326: Read of uninitialized components(*) in r2: *x/0 *y/1 *z/2 *w/3
(41): error X5326: Read of uninitialized components(*) in r2: *x/0 *y/1 *z/2 *w/3
(42): error X5326: Read of uninitialized components(*) in r2: *x/0 *y/1 *z/2 *w/3

tuan kuranes

20-03-2006 13:27:24

Mmhhh I cannot launch the paginglandscape sample I've compiled(crashes with nothing on the log, the project file must have errors).


Which compiler is that ?

I tried your very last precompiled demo, but I get this
oups... forgot to replace a code line in shadowreceivers cg.
updated in Source SDK, soon in Binary.

Crashy

20-03-2006 21:43:11

Mmhh no, everything works good with the precompiled demo. I still cannot launch the demo I compiled myself. (I use Visual 7.1).
(recompiled ogre, and updated the plugin). :?

You use only overlay system, no cegui?
I tried to debug but I'd som big problems with debug resources path :lol:


About the first problem: no problems without vertex compression. I tried with your shaders and material, but still the same problem.

The point is approximatively -2400,Y, 0 ... The result is only when I look at it from the west of the map(IE -3000,Y,3000).

I'm quite sure it doesn't come from my sample code: the chooseSceneManager method is the same as yours. And I just load the terrain with


mCamera->getSceneManager()->setOption( "CurrentMap", &nextMapL );


edit:I cannot reproduce the bug with grand canyon(even with my textures), and my app.

tuan kuranes

21-03-2006 09:24:32

You use only overlay system, no cegui?
No, not recently.

cannot reproduce the bug with grand canyon(even with my textures), and my app.
That's weird.
What's shadow mode at start (before rendering)

Crashy

21-03-2006 12:23:34

No shadow mode.

But I resolved it, I don't know how....I just took the grand canyon config file, and I replaced each parameters by those of my own config file...and it works now... :shock:

Mysteries of programming :roll: