Question

我需要通过一个列表数组以编程方式创建一个8位表面。这是我的约束。到目前为止，我可以使用此解决方案：

width = int(4)
height = int(4)
image_data_raw = ['\xf1', '\xf1', '\xf1', '\xf1',
              '\xf1', '\x00', '\x00', '\xf1',
              '\xf1', '\x00', '\x00', '\xf1',
              '\xf1', '\xf1', '\xf1', '\xf1']

def make_surface_from_raw_data():
    global image_data_raw
    global width
    global height

image_data_raw_bytes = [ord(i) for i in image_data_raw]

test_surface = pygame.Surface((width, height))
pxarray = pygame.PixelArray(test_surface)

i = int(0)
for y in range(height):
    for x in range(width):
        pxarray[x, y] = pygame.Color(image_data_raw_bytes[i],
                                     image_data_raw_bytes[i],
                                     image_data_raw_bytes[i])
        i += 1

new_surface = pxarray[:, :].make_surface()
image_data_2d_surface = new_surface
return image_data_2d_surface

但是，我对此不满意，因为它太慢了。

我的“ image_raw_data”将比4x4大得多，并从串行源流式传输。而且它们始终是8位灰度。因此，无需从8位转换为rbga。

我想知道是否有更快的方法来完成此任务。在我的循环中，有几次从char到8位int然后到rgba的转换。

非常感谢您。

Answer 1

如果您已安装NumPy，则使用诸如此类的速度会更快：

def make_surface_from_raw_data():
    global image_data_raw
    global width
    global height

    # we create a numpy array from image_data_raw
    # and do some bit shifting so that the single value is used for each color channel
    # then we have to reshape that array into the right size
    data = np.fromiter((x | x << 8 | x << 16 for x in (ord(i) for i in image_data_raw)), dtype=np.uint32).reshape(width, -width)

    # now let's create a Surface in the right dimension 
    # that's faster than letting pygame figure out the size from the array (with the make_surface function)
    surface = pygame.Surface((width, height))

    # blit_array is the fastet way to get the value of the array into the Surface
    # (faster than creating an array from the Surface and updating it)
    pygame.surfarray.blit_array(surface, data)

    return surface

用pygame以编程方式创建和填充8位曲面的最快方法

1 个答案: