Re: [gtkmm] pixmap to pixbuf conversion

Tobias Roth <roth iam unibe ch> writes:


> Now, if I want to edit the buffer, e.g. apply some filters, I read
> the buffer in an array using
> guint8* get_pixels () const
> to get a pointer to the actual pixel data. This data is in the
> format 'first pixel r value, first pixel g value, first pixel
> b value, second pixel r value, second pixel g value and so on'.
> So you need an array width*height*3 to hold all the rgb values
> or three width*height arrays to hold the three color 'planes'.
> If you have an alpha channel, the format is rgbargba and the
> array size is width*height*4.
> Then fiddle with the array, and finally write it back to the buffer
> with Gdk::Pixbuf::create_from_data(...).

Maybe I've misunderstood you, but I think this is wrong. The pointer
you get from get_pixels is a pointer into the pixbuf datastructure. So
you don't have to write anything back, you're changing the actual data.

The interface isn't very nice, though. I have a _much_ better one
which uses iterators and can be used to access pixels directly - if
you are interested, you can have a look here:

There is test example here:

I want to do some timings with GCC 3.2 before I write documentation
and announce it. If you have GCC 3.2, I would be very grateful if you
would run

  time ./test 1
  time ./test 2
  time ./test 3
  time ./test 4
  time ./test 5
  time ./test 6

and send the results. A compilation command is at the top of the
test.cpp file.

Ole Laursen

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]