directX , Clear buffer problem

hi

i want to use stencil buffer to render a mirror effect.

1 - render mirror surface
2 - render mirror just in stencil
3 - render reflection of object using stencil
4 - render objects

but when i try to clear buffers

device->Clear(0 , 0 , D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER | D3DCLEAR_STENCIL, 0x000000 , 1.0f , 0);
device->BeginScene();

it seems that TARGET isnt cleard when is add "D3DCLEAR_STENCIL" !!
even if id dont set : Stencil buffer active, by clearing it , TARGET wont clear !
i'm really confeused !! (
please HELP ME !!

I think you need to put the clear after the begin scene.

yakul
it wont solve the problem

I googled for the documentation about D3DCLEAR_STENCIL, and the first hit was a very similar problem: http://www.gamedev.net/community/for...opic_id=199232

But that thread isn't very helpful. The documentation tells me this

This method fails if you specify the D3DCLEAR_ZBUFFER or D3DCLEAR_STENCIL flags when the render target does not have an attached depth buffer. Similarly, if you specify the D3DCLEAR_STENCIL flag when the depth-buffer format does not contain stencil buffer information, this method fails.

Is that correct? You can/could/should try this too: turn the direct3d debug version on in the control panel, and watch your debug output (with dbmon.exe for example)

i think problem comes frome: D3DPRESENT_PARAMETER !
Stencil format is : D3DFMT_D24FS8, but if it failes
format is changed to : D3DFMT_D16,
does "D3DFMT_D16" mean not to use stencil ?
i use this type of code in another project too. its works fine, but in this one..!

and on more thing : i have installed DirectX SDK but there is no directx icon in my control panel...!

Originally Posted by Akinak

i think problem comes frome: D3DPRESENT_PARAMETER !
Stencil format is : D3DFMT_D24FS8, but if it failes
format is changed to : D3DFMT_D16,
does "D3DFMT_D16" mean not to use stencil ?

Quite so. It means "only use a 16 bit depth format". You should try to fallback to D3DFMT_D24S8, most videocards support that mode (as opposed to the 24 bit floating point z-buffer that D3DFMT_D24FS8 is).