Dennis,
I’ve tried a range of things: only specifying GL_UNPACK_ROW_LENGTH for
the
YCbCr image, setting GL_UNPACK_ROW_LENGTH to 0 or the image’s pixel
with
for the other formats, setting GL_UNPACK_CLIENT_STORAGE_APPLE to
GL_FALSE,
and probably a few other things as well. In all cases, only the upper
half
of any image with RGB, RGBA or grayscale pixel formats is show,
stretched
over the whole quad. If I remove all calls to glPixelStorei() from the
application, this happens to all image formats.
Here's a brief rundown of the code I have now:
SetTextureData( ... )
{
glBindTexture( GL_TEXTURE_RECTANGLE_EXT, gltex );
glTextureRangeAPPLE( GL_TEXTURE_RECTANGLE_EXT, size, data );
glTexParameteri( GL_TEXTURE_RECTANGLE_EXT,
GL_TEXTURE_STORAGE_HINT_APPLE,
GL_STORAGE_SHARED_APPLE );
glTexParameterf( GL_TEXTURE_RECTANGLE_EXT, GL_TEXTURE_PRIORITY, 0.0f
);
glTexParameteri( GL_TEXTURE_RECTANGLE_EXT, GL_TEXTURE_MIN_FILTER,
GL_LINEAR );
glTexParameteri( GL_TEXTURE_RECTANGLE_EXT, GL_TEXTURE_MAG_FILTER,
GL_LINEAR );
glTexParameteri( GL_TEXTURE_RECTANGLE_EXT, GL_TEXTURE_WRAP_S,
GL_CLAMP_TO_EDGE );
glTexParameteri( GL_TEXTURE_RECTANGLE_EXT, GL_TEXTURE_WRAP_T,
GL_CLAMP_TO_EDGE );
if( type == RGB )
{
glPixelStorei( GL_UNPACK_CLIENT_STORAGE_APPLE, GL_FALSE );
glPixelStorei( GL_UNPACK_ROW_LENGTH, 0 );
glTexImage2D( GL_TEXTURE_RECTANGLE_EXT, 0, GL_RGBA, GetWidth(),
GetHeight(), 0, GL_RGB, GL_UNSIGNED_BYTE, data );
}
else if( type == RGBA )
{
glPixelStorei( GL_UNPACK_CLIENT_STORAGE_APPLE, GL_FALSE );
glPixelStorei( GL_UNPACK_ROW_LENGTH, 0 );
glTexImage2D( GL_TEXTURE_RECTANGLE_EXT, 0, GL_RGBA, GetWidth(),
GetHeight(), 0, GL_RGBA, GL_UNSIGNED_BYTE, data );
}
else if( type == YCrCb )
{
glPixelStorei( GL_UNPACK_CLIENT_STORAGE_APPLE, GL_TRUE );
glPixelStorei( GL_UNPACK_ROW_LENGTH, GetWidth() * 2);
glTexImage2D( GL_TEXTURE_RECTANGLE_EXT, 0, GL_RGBA, GetWidth(),
GetHeight(), 0, GL_YCBCR_422_APPLE, GL_UNSIGNED_SHORT_8_8_REV_APPLE,
data );
}
else if( type == GRAYSCALE )
{
glPixelStorei( GL_UNPACK_CLIENT_STORAGE_APPLE, GL_FALSE );
glPixelStorei( GL_UNPACK_ROW_LENGTH, 0 );
glTexImage2D( GL_TEXTURE_RECTANGLE_EXT, 0, GL_LUMINANCE,
GetWidth(),
GetHeight(), 0, GL_LUMINANCE, GL_UNSIGNED_BYTE, data );
}
}
void draw()
{
glBindTexture( GL_TEXTURE_RECTANGLE_EXT, tex );
glBegin(GL_QUADS);
glTexCoord2f( 0, 0 ); glVertex3f( 0, h, 0.1 * i );
glTexCoord2f( w, 0 ); glVertex3f( w, h, 0.1 * i );
glTexCoord2f( w, h ); glVertex3f( w, 0, 0.1 * i );
glTexCoord2f( 0, h ); glVertex3f( 0, 0, 0.1 * i );
glEnd();
}
I'm really quite stumped now. I'm reasonably sure that the code that
generates the RGB data from the YCbCr input is correct (it's from
another
program, and runs fine there), and messing with the texture coordinates
shows that only half the image is being loaded into the texture.
I figure I've either not set something up, or I'm doing something
incorrectly. But I have no idea which. Thanks for the replies, and I
hope
somebody can point me in the right direction...
- Simon Wilson
On 21/4/05 7:56 AM, "Dennis Piatkowski"
<email@hidden> wrote:
Maybe I missed something, but it would seem you need to call
glPixelStorei(
GL_UNPACK_ROW_LENGTH,...) before each call to glTexImage2D() for each
of the
textures. I am not 100% certain, but I think this parameter is bound
with the
rest of the texture state in glBindTexture(). Even if not, just
making sure
to call glPixelStorei() after each glBindTexture() should garantee a
fix.
- Dennis
=================================== iChat: DrakkenWulf1 (AIM)
http://DrakkenWulfEnterprises.com
On Apr 20, 2005, at 4:23 PM, email@hidden wrote:
I'm looking for a way to display both a YUV image extracted from
QuickTime
(using either a sequence grabber or a movie controller) and an RGB
image
generated by my processing code. I can display the YUV data
correctly using
GL_YCBCR_422_APPLE inside a GL_TEXTURE_RECTANGLE_EXT texture. The
problem
currently appears with the use of glPixelStorei(
GL_UNPACK_ROW_LENGTH, width
* 2 ). With 'width*2', I get the correct height for the output YUV
image, but
the RGB texture data is only then half the height. Going for just
'width'
makes the RGB image work fine, but only provides half the YUV image.
Each
texture is rendered onto a single quad with varying transparency so
that I
can see the image through the marks made by the processing code.
I've seen a bit about YUV -> RGB shaders, and I'm curious whether I
need to
go down this path, but I can't see how that solves the texture
unpacking
problem. I don't really want to generate the RGB image as YUV as
well.
Does anybody have any suggestions? I'm stumped.
- Simon Wilson
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/mac-opengl/
email@hidden
This email sent to email@hidden
