pypng and graphics.Image

[uhm57]

Hi everyone,

I have recently found a pypng module ( http://packages.python.org/pypng/index.html ) which is an excellent tool to work with .png images. However, I can't figure out if this module can be used to create a standard graphics.Image objects directly. The reason why I want to use pypng and not the graphics.Image.open function is that pypng can extract alpha channel from the file. My goal is to create two graphics.Image objects, one which contains RGB data from file, and the other one which contains an alpha channel from file, to be used as a mask in blit operations. Could somebody help me with that?

[jge93]

You don't need that module for make a mask, see this:

def automask(im):
    width, height = im.size
    mask = Image.new(im.size, '1') # black and white
    tran = im.getpixel((0,0))[0]   # transparent top-left
    for y in range(height):
        line = im.getpixel([(x, y) for x in range(width)])
        for x in range(width):
            if line[x] == tran:
                mask.point((x,y), 0)  # mask on the point
    return mask 

example of usage:

from graphics import Image

ship = Image.open('E:\\Images\\ship.gif')
mask = automask(ship)

canvas.blit(ship, mask=mask)  # don't forget to create canvas first 

Regards.

[uhm57]

I've seen that article, thanks, but that's not the thing I want. PNG files have an alpha channel, not just one transparent color, so this easy way doesn't really work.

[pavan.pareta]

Hi Uhm57,

For masking and transparency look this article.

Unleash the power of PNG transparency with masks

Hope it helps.

[uhm57]

Sorry, guys, but it doesn't help (the previous answer was a part of that article, too).

To clarify what I need, let's change the question a bit. Suppose that I have an array of numbers (or RGB triplets with R=G=B). Is there any easy way to convert it to a grayscale Image object (as in graphics.Image)? Alternatively, is there some documentation or code which explains how graphics.Image objects are created?

[marcelobarrosalmeida]

Sorry, guys, but it doesn't help (the previous answer was a part of that article, too).

To clarify what I need, let's change the question a bit. Suppose that I have an array of numbers (or RGB triplets with R=G=B). Is there any easy way to convert it to a grayscale Image object (as in graphics.Image)? Alternatively, is there some documentation or code which explains how graphics.Image objects are created?

I think you can create the image using the desired size and color scheme. After, you can set all points using the expression Y = 0.3*R + 0.59*G + 0.11*B (see here for details). If you have created the image using colors, just use same Y value for R, G and B.

Marcelo

[uhm57]

If by "set points" you mean using point method, yes, that should work, but I think copying an image from an array pixel-by-pixel would be terribly slow (correct me if I'm wrong, though). Isn't there a faster way, like working with Image as an array too?

[marcelobarrosalmeida]

If by "set points" you mean using point method, yes, that should work, but I think copying an image from an array pixel-by-pixel would be terribly slow (correct me if I'm wrong, though). Isn't there a faster way, like working with Image as an array too?

Yes, it will be slow but I can not realize any other method to accomplish this task
Let's wait more answers ...

Marcelo