Mailing Lists: Apple Mailing Lists
Image of Mac OS face in stamp
Re: Screen snapshot example code posted
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Screen snapshot example code posted




On Aug 13, 2005, at 9:10 AM, Ben Haller wrote:

On the flip side, it's probably slow compared to what the core video system could do itself; reading pixels one by one is not as fast as getting down-and-dirty with cacheline reads and such, and certainly isn't as fast as copying from one buffer on the video card to another without involving the CPU at all, which is what the system perhaps does (or should do, or someday will do) on at least some accelerated video cards.

Absolutely! The best way to handle the abstraction of the hardware, and get good performance, is probably to go with the OpenGL software. Now, using OpenGL isn't exactly obvious, though, so some sample code is in order.


Finally, it's uncomfortably dependent upon the layout of video memory; Josh's code only works for 16-bit and 32-bit color, and only with interleaved (non-planar) data, which is fine for now AFAIK, but Apple has made no guarantees about sticking with only those memory layouts in the future that I am aware of. It's possible right now to switch the screen to 8-bit mode, and many games do, but Josh's code wouldn't be able to take a snapshot of the screen in that state, while my CopyBits-base code probably would (haven't tried). And so forth.

Again, OpenGL can help here, as it includes the ability to convert pixel formats, and can even deal with hardware-specific pixel reading problems such as microtiling, or getting the correct framebuffer when full screen buffer swapping (the GL version of page flipping) is in effect. OpenGL will also use a DMA transfer for reading the framebuffer content, where available.


OpenGL provides both a synchronous read path suitable for simple screen captures using glReadPixels, and a fast asynchronous path using the texture read mechanism, which is a bit more complex, but works well for repeated capture operations, such as screencasting or recording a movie from the screen.

Here's a simple mechanism that can be used to read a rectangle from a display to a CGImageRef. It's more comment than code:

/*
* glGrab.c
*/
/*
IMPORTANT: This Apple software is supplied to you by Apple Computer, Inc. ("Apple") in
consideration of your agreement to the following terms, and your use, installation or modification
of this Apple software constitutes acceptance of these terms. If you do not agree with these terms,
please do not use, install or modify this Apple software.


In consideration of your agreement to abide by the following terms, and subject to these
terms, Apple grants you a personal, non-exclusive license, under Apple\xd5s copyrights in
this original Apple software (the "Apple Software"), to use and modify the Apple Software,
with or without modifications, in source and/or binary forms; provided that if you redistribute
the Apple Software in its entirety and without modifications, you must retain this notice and
the following text and disclaimers in all such redistributions of the Apple Software.
Neither the name, trademarks, service marks or logos of Apple Computer, Inc. may be used
to endorse or promote products derived from the Apple Software without specific prior
written permission from Apple. Except as expressly tated in this notice, no other rights or
licenses, express or implied, are granted by Apple herein, including but not limited
to any patent rights that may be infringed by your derivative works or by other works
in which the Apple Software may be incorporated.


The Apple Software is provided by Apple on an "AS IS" basis. APPLE MAKES NO WARRANTIES,
EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, REGARDING THE APPLE SOFTWARE OR ITS
USE AND OPERATION ALONE OR IN COMBINATION WITH YOUR PRODUCTS.


IN NO EVENT SHALL APPLE BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) ARISING IN ANY WAY OUT OF THE USE,
REPRODUCTION, MODIFICATION AND/OR DISTRIBUTION OF THE APPLE SOFTWARE, HOWEVER CAUSED AND
WHETHER UNDER THEORY OF CONTRACT, TORT (INCLUDING NEGLIGENCE), STRICT LIABILITY OR
OTHERWISE, EVEN IF APPLE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/


#include <CoreFoundation/CoreFoundation.h>
#include <ApplicationServices/ApplicationServices.h>
#include <OpenGL/OpenGL.h>
#include <OpenGL/gl.h>


/* * perform an in-place swap from Quadrant 1 to Quadrant III format * (upside-down PostScript/GL to right side up QD/CG raster format) * We do this in-place, which requires more copying, but will touch * only half the pages. (Display grabs are BIG!) * * Pixel reformatting may optionally be done here if needed. */ static void swizzleBitmap(void * data, int rowBytes, int height) { int top, bottom; void * buffer; void * topP; void * bottomP; void * base;

    top = 0;
    bottom = height - 1;
    base = data;
    buffer = malloc(rowBytes);

    while ( top < bottom )
    {
        topP = (void *)((top * rowBytes) + (intptr_t)base);
        bottomP = (void *)((bottom * rowBytes) + (intptr_t)base);

/*
* Save and swap scanlines.
*
* This code does a simple in-place exchange with a temp buffer.
* If you need to reformat the pixels, replace the first two bcopy()
* calls with your own custom pixel reformatter.
*/
bcopy( topP, buffer, rowBytes );
bcopy( bottomP, topP, rowBytes );
bcopy( buffer, bottomP, rowBytes );


        ++top;
        --bottom;
    }
    free( buffer );
}

/*
* Given a display ID and a rectangle on that display, generate a CGImageRef
* containing the display contents.
*
* srcRect is display-origin relative.
*
* This function uses a full screen OpenGL read-only context.
* By using OpenGL, we can read the screen using a DMA transfer
* when it's in millions of colors mode, and we can correctly read
* a microtiled full screen OpenGL context, such as a game or full
* screen video display.
*
* Returns a CGImageRef. When you are done with the CGImageRef, release it
* using CFRelease().
* Returns NULL on an error.
*/
CGImageRef grabViaOpenGL(CGDirectDisplayID display, CGRect srcRect)
{
CGContextRef bitmap;
CGImageRef image;
void * data;
long bytewidth;
GLint width, height;
long bytes;
CGColorSpaceRef cSpace = CGColorSpaceCreateWithName (kCGColorSpaceGenericRGB);


    CGLContextObj    glContextObj;
    CGLPixelFormatObj pixelFormatObj ;
    long numPixelFormats ;
    CGLPixelFormatAttribute attribs[] =
    {
        kCGLPFAFullScreen,
        kCGLPFADisplayMask,
        0,    /* Display mask bit goes here */
        0
    } ;

    if ( display == kCGNullDirectDisplay )
        display = CGMainDisplayID();
    attribs[2] = CGDisplayIDToOpenGLDisplayMask(display);

    /* Build a full-screen GL context */
    CGLChoosePixelFormat( attribs, &pixelFormatObj, &numPixelFormats );
    if ( pixelFormatObj == NULL )    // No full screen context support
        return NULL;
    CGLCreateContext( pixelFormatObj, NULL, &glContextObj ) ;
    CGLDestroyPixelFormat( pixelFormatObj ) ;
    if ( glContextObj == NULL )
        return NULL;

    CGLSetCurrentContext( glContextObj ) ;
    CGLSetFullScreen( glContextObj ) ;

    glReadBuffer(GL_FRONT);

    width = srcRect.size.width;
    height = srcRect.size.height;

bytewidth = width * 4; // Assume 4 bytes/pixel for now
bytewidth = (bytewidth + 3) & ~3; // Align to 4 bytes
bytes = bytewidth * height; // width * height


/* Build bitmap context */
data = malloc(height * bytewidth);
if ( data == NULL )
{
CGLSetCurrentContext( NULL );
CGLClearDrawable( glContextObj ); // disassociate from full screen
CGLDestroyContext( glContextObj ); // and destroy the context
return NULL;
}
bitmap = CGBitmapContextCreate(data, width, height, 8, bytewidth,
cSpace, kCGImageAlphaNoneSkipFirst /* XRGB */);
CFRelease(cSpace);



/* Read framebuffer into our bitmap */
glFinish(); /* Finish all OpenGL commands */
glPixelStorei(GL_PACK_ALIGNMENT, 4); /* Force 4-byte alignment */
glPixelStorei(GL_PACK_ROW_LENGTH, 0);
glPixelStorei(GL_PACK_SKIP_ROWS, 0);
glPixelStorei(GL_PACK_SKIP_PIXELS, 0);


/*
* Fetch the data in XRGB format, matching the bitmap context.
*/
glReadPixels((GLint)srcRect.origin.x, (GLint)srcRect.origin.y, width, height,
GL_BGRA,
GL_UNSIGNED_INT_8_8_8_8_REV,
data);
/*
* glReadPixels generates a quadrant I raster, with origin in the lower left
* This isn't a problem for signal processing routines such as compressors,
* as they can simply use a negative 'advance' to move between scanlines.
* CGImageRef and CGBitmapContext assume a quadrant III raster, though, so we need to
* invert it. Pixel reformatting can also be done here.
*/
swizzleBitmap(data, bytewidth, height);



/* Make an image out of our bitmap; does a cheap vm_copy of the bitmap */
image = CGBitmapContextCreateImage(bitmap);


    /* Get rid of bitmap */
    CFRelease(bitmap);
    free(data);

/* Get rid of GL context */
CGLSetCurrentContext( NULL );
CGLClearDrawable( glContextObj ); // disassociate from full screen
CGLDestroyContext( glContextObj ); // and destroy the context


    /* Returned image has a reference count of 1 */
    return image;
}


_______________________________________________ Do not post admin requests to the list. They will be ignored. Cocoa-dev mailing list (email@hidden) Help/Unsubscribe/Update your Subscription: This email sent to email@hidden
References: 
 >Screen snapshot example code posted (From: Ben Haller <email@hidden>)
 >Re: Screen snapshot example code posted (From: Andreas Mayer <email@hidden>)
 >Re: Screen snapshot example code posted (From: Ben Haller <email@hidden>)



Visit the Apple Store online or at retail locations.
1-800-MY-APPLE

Contact Apple | Terms of Use | Privacy Policy

Copyright © 2011 Apple Inc. All rights reserved.