[PyOpenGL-Users] funny return format of glGetTexImage

[Patrick D. <pyo...@pd...>]

Hi all,

using glGetTexImageI have noticed two bugs:

1.
according to the documentation the return value should always be an 
array if
OpenGL.UNSIGNED_BYTE_IMAGES_AS_STRING == False
which is not the case: OpenGL.UNSIGNED_BYTE_IMAGES_AS_STRING seems to be 
ignored completely and
with type=GL_UNSIGNED_BYTE and the default for outputType a string is 
returned. Setting outputType=None returns an array.

2.
image = glGetTexImage( GL_TEXTURE_RECTANGLE, level=0, format=GL_RGBA, 
format, type=GL_UNSIGNED_BYTE, outputType=None )
print "\timage:", type(image), image.shape, image.dtype

prints
     image: <type 'numpy.ndarray'> (472L, 575L, 1L, 4L) uint8

(for format=GL_RGB the shape is (472L, 575L, 1L, 3L) )

In addition to the unused dimension 2 of the array, width and height are 
mixed up.
My workaround is:

image.shape = (image.shape[1], image.shape[0], image.shape[3])

which creates an image that looks fine apart from being mirrored along 
the horizontal axis.
So to save it to disk, I do:
im = PIL.Image.fromarray(image_data)
im = im.transpose(PIL.Image.FLIP_TOP_BOTTOM)
im.save('test.png', format="PNG")

Regards,
Patrick