Question

I store image’s pixels in a single multiprocessing.Array of integers which can be indexed as

self.data[x*height + y]

as this is convenient to use with IPC.

I’m trying to save the pixels to a file using PIL’s putdata() to make it faster than using putpixel() or the [] indexing. However I cannot figure out in which format does putdata() want the imagedata.

I thought the same type of indexing would work with putdata()

def write(self):
    im = Image.new("RGB", (self.width, self.height))
    imagedata = [self.intToRGB(self.data[i*self.height + j]) 
                     for i in range(self.width) 
                        for j in range(self.height)]
    im.putdata(imagedata)
    im.save(self.filename, "PNG")

self.intToRGB() returns a tuple(r, g, b).

But the code above results to a -90 degrees turned image with a lot of repetition.

If I use putpixel() instead of putdata() like this:

    pic = im.load()
    for i in range(self.width):
        offset = i * self.height
        for j in range(self.height):
            pic[i, j] =  imagedata[i * self.height + j]

it produces

Answer 1

I think I see the problem in your first line:

self.data[x*height + y]

Image data is generally organized the other way around:

self.data[y*width + x]