[Interest] QSharedMemory example has excessive copies?

[Joshua Grauman]

Hello,

I'm looking to use QSharedMemory to grab an image from a widget into 
shared memory (using QWidget::grab() or QWidget::render()), so another 
process/program can process the images.

So I'm looking at the QSharedMemory example here:
http://doc.qt.io/qt-5/qtcore-ipc-sharedmemory-example.html

It looks to me like the process that loads the image from the file into 
shared memory has 3 deep copies. Maybe implicit sharing reduces that, I'm 
not sure. But on the surface, it looks like 1) the file is loaded into a 
QImage using image.load(fileName). Then 2) the QImage is streamed into a 
QBuffer using out << image; And finally, 3) the buffer is memcopied into 
shared memory.

So... isn't there an Qt way to do this with 1 copy? It would be trivial to 
load an image from a file straight into shared memory in straight C. And 
more importantly, is there a way to grab a widget image data (including 
alpha channel) straight into shared memory? I'd prefer a Qt way of doing 
this, but I'd also accept hacks as this doesn't need to be portable. I'm 
just trying to grab a widget 30 frames a second from my Qt app and convert 
it to video.

Josh