wined3d: Improve render target to texture blits.

This commit is contained in:
Stefan Dösinger 2007-02-12 19:18:46 +01:00 committed by Alexandre Julliard
parent f22d37e4dd
commit e534340450
1 changed files with 276 additions and 43 deletions

View File

@ -2191,6 +2191,240 @@ static HRESULT WINAPI IWineD3DSurfaceImpl_Flip(IWineD3DSurface *iface, IWineD3DS
return IWineD3DDevice_Present(D3D, NULL, NULL, 0, NULL); return IWineD3DDevice_Present(D3D, NULL, NULL, 0, NULL);
} }
/* Does a direct frame buffer -> texture copy. Stretching is done
* with single pixel copy calls
*/
static inline void fb_copy_to_texture_direct(IWineD3DSurfaceImpl *This, IWineD3DSurface *SrcSurface, IWineD3DSwapChainImpl *swapchain, WINED3DRECT *srect, WINED3DRECT *drect, BOOL upsidedown) {
IWineD3DDeviceImpl *myDevice = This->resource.wineD3DDevice;
float xrel, yrel;
UINT row;
BOOL warned = FALSE; /* deliberately not static */
IWineD3DSurfaceImpl *Src = (IWineD3DSurfaceImpl *) SrcSurface;
ENTER_GL();
ActivateContext(myDevice, SrcSurface, CTXUSAGE_BLIT);
/* Bind the target texture */
glBindTexture(GL_TEXTURE_2D, This->glDescription.textureName);
checkGLcall("glBindTexture");
if(swapchain->backBuffer && SrcSurface == swapchain->backBuffer[0]) {
glReadBuffer(GL_BACK);
} else {
glReadBuffer(GL_FRONT);
}
checkGLcall("glReadBuffer");
xrel = (float) (srect->x2 - srect->x1) / (float) (drect->x2 - drect->x1);
yrel = (float) (srect->y2 - srect->y1) / (float) (drect->y2 - drect->y1);
if( (xrel - 1.0 < -eps) || (xrel - 1.0 > eps)) {
FIXME("Doing a pixel by pixel copy from the framebuffer to a texture, expect major performance issues\n");
}
if(upsidedown &&
!((xrel - 1.0 < -eps) || (xrel - 1.0 > eps)) &&
!((yrel - 1.0 < -eps) || (yrel - 1.0 > eps))) {
/* Upside down copy without stretching is nice, one glCopyTexSubImage call will do */
glCopyTexSubImage2D(This->glDescription.target,
This->glDescription.level,
drect->x1, drect->y1, /* xoffset, yoffset */
srect->x1, Src->currentDesc.Height - srect->y2,
drect->x2 - drect->x1, drect->y2 - drect->y1);
} else {
UINT yoffset = Src->currentDesc.Height - srect->y1 + drect->y1 - 1;
/* I have to process this row by row to swap the image,
* otherwise it would be upside down, so streching in y direction
* doesn't cost extra time
*
* However, streching in x direction can be avoided if not necessary
*/
for(row = drect->y1; row < drect->y2; row++) {
if( (xrel - 1.0 < -eps) || (xrel - 1.0 > eps)) {
/* Well, that stuff works, but it's very slow.
* find a better way instead
*/
UINT col;
if(!warned) {
warned = TRUE;
FIXME("Doing a pixel by pixel render target -> texture copy, expect performance issues\n");
}
for(col = drect->x1; col < drect->x2; col++) {
glCopyTexSubImage2D(This->glDescription.target,
This->glDescription.level,
drect->x1 + col, row, /* xoffset, yoffset */
srect->x1 + col * xrel, yoffset - (int) (row * yrel),
1, 1);
}
} else {
glCopyTexSubImage2D(This->glDescription.target,
This->glDescription.level,
drect->x1, row, /* xoffset, yoffset */
srect->x1, yoffset - (int) (row * yrel),
drect->x2-drect->x1, 1);
}
}
}
vcheckGLcall("glCopyTexSubImage2D");
LEAVE_GL();
}
/* Uses the hardware to stretch and flip the image */
static inline void fb_copy_to_texture_hwstretch(IWineD3DSurfaceImpl *This, IWineD3DSurface *SrcSurface, IWineD3DSwapChainImpl *swapchain, WINED3DRECT *srect, WINED3DRECT *drect, BOOL upsidedown) {
GLuint backup, src;
IWineD3DDeviceImpl *myDevice = This->resource.wineD3DDevice;
IWineD3DSurfaceImpl *Src = (IWineD3DSurfaceImpl *) SrcSurface;
float left, right, top, bottom; /* Texture coordinates */
UINT fbwidth = Src->currentDesc.Width;
UINT fbheight = Src->currentDesc.Height;
TRACE("Using hwstretch blit\n");
/* Activate the Proper context for reading from the source surface, set it up for blitting */
ENTER_GL();
ActivateContext(myDevice, SrcSurface, CTXUSAGE_BLIT);
/* Backup the back buffer and copy the source buffer into a texture to draw an upside down stretched quad. If
* we are reading from the back buffer, the backup can be used as source texture
*/
glGenTextures(1, &backup);
checkGLcall("glGenTextures(1, &backup)");
glBindTexture(GL_TEXTURE_2D, backup);
checkGLcall("glBindTexture(GL_TEXTURE_2D, backup)");
glReadBuffer(GL_BACK);
checkGLcall("glReadBuffer(GL_BACK)");
/* TODO: Only back up the part that will be overwritten */
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, Src->pow2Width, Src->pow2Height, 0,
GL_RGBA, GL_UNSIGNED_BYTE, NULL);
checkGLcall("glTexImage2D");
glCopyTexSubImage2D(GL_TEXTURE_2D, 0,
0, 0 /* read offsets */,
0, 0,
fbwidth,
fbheight);
checkGLcall("glCopyTexSubImage2D");
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
checkGLcall("glTexParameteri");
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
checkGLcall("glTexParameteri");
if((IWineD3DSurface *) This == swapchain->backBuffer[0] || TRUE) {
src = backup;
} else {
glReadBuffer(GL_FRONT);
checkGLcall("glReadBuffer(GL_FRONT)");
glGenTextures(1, &src);
checkGLcall("glGenTextures(1, &src)");
glBindTexture(GL_TEXTURE_2D, src);
checkGLcall("glBindTexture(GL_TEXTURE_2D, src)");
/* TODO: Only copy the part that will be read. Use srect->x1, srect->y2 as origin, but with the width watch
* out for power of 2 sizes
*/
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, Src->pow2Width, Src->pow2Height, 0,
GL_RGBA, GL_UNSIGNED_BYTE, NULL);
checkGLcall("glTexImage2D");
glCopyTexSubImage2D(GL_TEXTURE_2D, 0,
0, 0 /* read offsets */,
0, 0,
fbwidth,
fbheight);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
checkGLcall("glTexParameteri");
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
checkGLcall("glTexParameteri");
glReadBuffer(GL_BACK);
checkGLcall("glReadBuffer(GL_BACK)");
}
checkGLcall("glEnd and previous");
left = (float) srect->x1 / (float) Src->pow2Width;
right = (float) srect->x2 / (float) Src->pow2Width;
if(upsidedown) {
top = (float) (Src->currentDesc.Height - srect->y1) / (float) Src->pow2Height;
bottom = (float) (Src->currentDesc.Height - srect->y2) / (float) Src->pow2Height;
} else {
top = (float) (Src->currentDesc.Height - srect->y2) / (float) Src->pow2Height;
bottom = (float) (Src->currentDesc.Height - srect->y1) / (float) Src->pow2Height;
}
/* draw the source texture stretched and upside down. The correct surface is bound already */
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP);
glBegin(GL_QUADS);
/* bottom left */
glTexCoord2f(left, bottom);
glVertex2i(0, fbheight);
/* top left */
glTexCoord2f(left, top);
glVertex2i(0, fbheight - drect->y2 - drect->y1);
/* top right */
glTexCoord2f(right, top);
glVertex2i(drect->x2 - drect->x1, fbheight - drect->y2 - drect->y1);
/* bottom right */
glTexCoord2f(right, bottom);
glVertex2i(drect->x2 - drect->x1, fbheight);
glEnd();
checkGLcall("glEnd and previous");
/* Now read the stretched and upside down image into the destination texture */
glBindTexture(This->glDescription.target, This->glDescription.textureName);
checkGLcall("glBindTexture");
glCopyTexSubImage2D(This->glDescription.target,
0,
drect->x1, drect->y1, /* xoffset, yoffset */
0, 0, /* We blitted the image to the origin */
drect->x2 - drect->x1, drect->y2 - drect->y1);
checkGLcall("glCopyTexSubImage2D");
/* Write the back buffer backup back */
glBindTexture(GL_TEXTURE_2D, backup);
checkGLcall("glBindTexture(GL_TEXTURE_2D, backup)");
glBegin(GL_QUADS);
/* top left */
glTexCoord2f(0.0, (float) fbheight / (float) Src->pow2Height);
glVertex2i(0, 0);
/* bottom left */
glTexCoord2f(0.0, 0.0);
glVertex2i(0, fbheight);
/* bottom right */
glTexCoord2f((float) fbwidth / (float) Src->pow2Width, 0.0);
glVertex2i(fbwidth, Src->currentDesc.Height);
/* top right */
glTexCoord2f((float) fbwidth / (float) Src->pow2Width, (float) fbheight / (float) Src->pow2Height);
glVertex2i(fbwidth, 0);
glEnd();
/* Cleanup */
if(src != backup) {
glDeleteTextures(1, &src);
checkGLcall("glDeleteTextures(1, &src)");
}
glDeleteTextures(1, &backup);
checkGLcall("glDeleteTextures(1, &backup)");
LEAVE_GL();
}
/* Not called from the VTable */ /* Not called from the VTable */
static HRESULT IWineD3DSurfaceImpl_BltOverride(IWineD3DSurfaceImpl *This, RECT *DestRect, IWineD3DSurface *SrcSurface, RECT *SrcRect, DWORD Flags, DDBLTFX *DDBltFx) { static HRESULT IWineD3DSurfaceImpl_BltOverride(IWineD3DSurfaceImpl *This, RECT *DestRect, IWineD3DSurface *SrcSurface, RECT *SrcRect, DWORD Flags, DDBLTFX *DDBltFx) {
WINED3DRECT rect; WINED3DRECT rect;
@ -2432,73 +2666,72 @@ static HRESULT IWineD3DSurfaceImpl_BltOverride(IWineD3DSurfaceImpl *This, RECT *
(swapchain->backBuffer && SrcSurface == swapchain->backBuffer[0]) ) { (swapchain->backBuffer && SrcSurface == swapchain->backBuffer[0]) ) {
if( ( (IWineD3DSurface *) This != swapchain->frontBuffer) && if( ( (IWineD3DSurface *) This != swapchain->frontBuffer) &&
( swapchain->backBuffer && (IWineD3DSurface *) This != swapchain->backBuffer[0]) ) { ( swapchain->backBuffer && (IWineD3DSurface *) This != swapchain->backBuffer[0]) ) {
UINT row;
WINED3DRECT srect; WINED3DRECT srect;
float xrel, yrel; BOOL upsideDown, stretchx;
TRACE("Blt from rendertarget to texture\n"); TRACE("Blt from rendertarget to texture\n");
/* Call preload for the surface to make sure it isn't dirty */ /* Call preload for the surface to make sure it isn't dirty */
IWineD3DSurface_PreLoad((IWineD3DSurface *) This); IWineD3DSurface_PreLoad((IWineD3DSurface *) This);
/* Make sure that the top pixel is always above the bottom pixel, and keep a seperate upside down flag
* glCopyTexSubImage is a bit picky about the parameters we pass to it
*/
if(SrcRect) { if(SrcRect) {
if(SrcRect->top < SrcRect->bottom) {
srect.y1 = SrcRect->top;
srect.y2 = SrcRect->bottom;
upsideDown = FALSE;
} else {
srect.y1 = SrcRect->bottom;
srect.y2 = SrcRect->top;
upsideDown = TRUE;
}
srect.x1 = SrcRect->left; srect.x1 = SrcRect->left;
srect.y1 = SrcRect->top;
srect.x2 = SrcRect->right; srect.x2 = SrcRect->right;
srect.y2 = SrcRect->bottom;
} else { } else {
srect.x1 = 0; srect.x1 = 0;
srect.y1 = 0; srect.y1 = 0;
srect.x2 = Src->currentDesc.Width; srect.x2 = Src->currentDesc.Width;
srect.y2 = Src->currentDesc.Height; srect.y2 = Src->currentDesc.Height;
upsideDown = FALSE;
}
if(rect.x1 > rect.x2) {
UINT tmp = rect.x2;
rect.x2 = rect.x1;
rect.x1 = tmp;
upsideDown = !upsideDown;
} }
ENTER_GL(); if(rect.x2 - rect.x1 != srect.x2 - srect.x1) {
stretchx = TRUE;
/* Bind the target texture */
glBindTexture(GL_TEXTURE_2D, This->glDescription.textureName);
checkGLcall("glBindTexture");
if(swapchain->backBuffer && SrcSurface == swapchain->backBuffer[0]) {
glReadBuffer(GL_BACK);
} else { } else {
glReadBuffer(GL_FRONT); stretchx = FALSE;
} }
checkGLcall("glReadBuffer");
xrel = (float) (srect.x2 - srect.x1) / (float) (rect.x2 - rect.x1); /* Blt is a pretty powerful call, while glCopyTexSubImage2D is not. glCopyTexSubImage cannot
yrel = (float) (srect.y2 - srect.y1) / (float) (rect.y2 - rect.y1); * flip the image nor scale it. If GL_EXT_framebuffer_blit is available it can be used(hopefully,
* not implemented by now). Otherwise:
/* I have to process this row by row to swap the image,
* otherwise it would be upside down, so streching in y direction
* doesn't cost extra time
* *
* However, streching in x direction can be avoided if not necessary * -> If the app asks for a unscaled, upside down copy, just perform one glCopyTexSubImage2D call
* -> If the app wants a image width an unscaled width, copy it line per line
* -> If the app wants a image that is scaled on the x axis, and the destination rectangle is smaller
* than the frame buffer, draw an upside down scaled image onto the fb, read it back and restore the
* back buffer. This is slower than reading line per line, thus not used for flipping
* -> If the app wants a scaled image with a dest rect that is bigger than the fb, it has to be copied
* pixel by pixel
*/ */
for(row = rect.y1; row < rect.y2; row++) { if(FALSE /* GL_SUPPORT(EXT_FRAMEBUFFER_BLIT) */) {
if( (xrel - 1.0 < -eps) || (xrel - 1.0 > eps)) { TRACE("Using GL_EXT_framebuffer_blit for copying\n");
/* Well, that stuff works, but it's very slow. } else if((!stretchx) || rect.x2 - rect.x1 > Src->currentDesc.Width ||
* find a better way instead rect.y2 - rect.y1 > Src->currentDesc.Height) {
*/ TRACE("No stretching in x direction, using direct framebuffer -> texture copy\n");
UINT col; fb_copy_to_texture_direct(This, SrcSurface, swapchain, &srect, &rect, upsideDown);
for(col = rect.x1; col < rect.x2; col++) { } else {
glCopyTexSubImage2D(GL_TEXTURE_2D, TRACE("Using hardware stretching to flip / stretch the texture\n");
0, /* level */ fb_copy_to_texture_hwstretch(This, SrcSurface, swapchain, &srect, &rect, upsideDown);
rect.x1 + col, This->currentDesc.Height - row - 1, /* xoffset, yoffset */
srect.x1 + col * xrel, Src->currentDesc.Height - srect.y2 + row * yrel,
1, 1);
}
} else {
glCopyTexSubImage2D(GL_TEXTURE_2D,
0, /* level */
rect.x1, rect.y2 + rect.y1 - row - 1, /* xoffset, yoffset */
srect.x1, row - rect.y1,
rect.x2-rect.x1, 1);
}
} }
vcheckGLcall("glCopyTexSubImage2D");
LEAVE_GL();
if(!(This->Flags & SFLAG_DONOTFREE)) { if(!(This->Flags & SFLAG_DONOTFREE)) {
HeapFree(GetProcessHeap(), 0, This->resource.allocatedMemory); HeapFree(GetProcessHeap(), 0, This->resource.allocatedMemory);
This->resource.allocatedMemory = NULL; This->resource.allocatedMemory = NULL;