Vfrdtyky

At high school, we are beginning with Python and we are asked to program a small image processor using the Pillow module. The goal is to open a file, then choose a filter.

The first filter I tried to program is to change the color temperature.
But there's also a minimalist interface with tkinter that shows some buttons and it works without problems.

Here is the function that opens the file

def Ouvrir():

    fichier = askopenfilename(title="Ouvrir une image",filetypes=[('jpg files','.jpg'),('all files','.*')])

    global img

    img =Image.open(fichier)

    l, h = img.size

    img.show()   #visualisation de l'image



    global img2

    img2 = img #img2 est une copie de img

It creates a global img image object that is loaded from the file. Then l and h are loaded with the width and height of the image.

Another global object image img2 is created for the output image, and it is a copy of img.

Then here is the function that processes the image

def filtreTC(): #Filtre permettant de changer la température de couleur

    coef = sliderTC.get() / 100 #On récupère le coefficient à partir de l'échelle. Le coefficient compris entre -1 et 1: -1 = froid (image bleu-vert), 0 = neutre, 1 = chaud (image orangée)

    fenTC.destroy() #On ferme la fenêtre

    if(coef <= 0): #Calcul des coefficients rouges, verte et bleus

        coefR = 1 + coef

        coefV = 1 + (coef / 2)

        coefB = 1

    else:

        coefR = 1

        coefV = 1 - (coef / 2)

        coefB = 1 - coef

    for y in range(0, h, 1):

        for x in range(0, l, 1):

            r, v, b = img.getpixel((x, y))

            r = int(float(r * coefR))

            v = int(float(v * coefV))

            b = int(float(b * coefB))

            img2.putpixel((x, y), (r, v, b))

    img2.show()

It's a loop that scans through the whole image and takes a pixel from img, multiply its rgb values by their corresponding coefficients, then puts that pixel in img2

The problem is that it doesn't work. It doesn't give any errors, but when it shows img2 it's the same as img1, like the putpixel function didn't do anything.

I checked a lot of things so I know the problem is neither the rgb values nor the x/y coordinates.

• I tried replacing img2 = img by img2 = Image.new("RGB", (l, h))and what I get is a black image.


• I then tried replacing img2.putpixel((x, y), (r, v, b)) by img2.putpixel((100, 100), (127, 127, 127)) so that way I should get a grey pixel near the top left corner. But I still got a black image.

I then tried to remove

global img2

img2 = img

from the function that opens a file and put

img2 = Image.new("RGB", (l, h))

right after fenTC.destroy() and I got this:

Exception in Tkinter callback

Traceback (most recent call last):

  File "e:xxxxprogrammesanacondalibtkinter__init__.py", line 1702, in __call__

    return self.func(*args)

  File "H:ISNProgrammesTP-imageprojet.py", line 62, in filtreTC

    img2.show()

  File "e:xxxxprogrammesanacondalibsite-packagesPILImage.py", line 2016, in show

    _show(self, title=title, command=command)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImage.py", line 2876, in _show

    _showxv(image, **options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImage.py", line 2881, in _showxv

    ImageShow.show(image, title, **options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImageShow.py", line 51, in show

    if viewer.show(image, title=title, **options):

  File "e:xxxxprogrammesanacondalibsite-packagesPILImageShow.py", line 75, in show

    return self.show_image(image, **options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImageShow.py", line 95, in show_image

    return self.show_file(self.save_image(image), **options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImageShow.py", line 91, in save_image

    return image._dump(format=self.get_format(image), **self.options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImage.py", line 639, in _dump

    self.save(filename, format, **options)

  File "e:xxxxprogrammesanacondalibsite-packagesPILImage.py", line 1969, in save

    save_handler(self, fp, filename)

  File "e:xxxxprogrammesanacondalibsite-packagesPILBmpImagePlugin.py", line 319, in _save

    (rawmode, stride, -1))])

  File "e:xxxxprogrammesanacondalibsite-packagesPILImageFile.py", line 512, in _save

    e.setimage(im.im, b)

SystemError: tile cannot extend outside image

Basically a lot of stuff I've got no idea what it means except:

SystemError: tile cannot extend outside image

I checked the x and y coordinates and they never went beyond the borders, so I don't understand what's this error.

I spent several hours to try figure out what the problem was and tried lots of different things but none of them worked. So I thought it was better to ask for help

Your problem is that you didn't create a copy. img2 = img creates another reference to the same object, not a new, separate image.

To create a actual copy, use the image.copy() method:

img2 = img.copy()

Next, I'd not use a loop and getpixel() / putpixel() combinations. You are doing double work for RGB values that appear more than once. If you use the image.point() method then you get to use your formula on each unique value in each band in the image, and leave the looping over all pixels to the library (much faster). It also makes the image copy for you!

You need to create a table; for values 0 through 255 for the R, G and B values, calculate the possible result, and put those 3 * 256 results in a long list:

coefG = [i * (1 - (coef / 2)) for i in range(256)]

if coef <= 0:

    coefR = [i * (1 + coef) for i in range(256)]

    coefB = list(range(256))

else:

    coefR = list(range(256))

    coefB = [i * (1 - coef) for i in range(256)]

img2 = img1.point(coefR + coefG + coefB)

The resulting table is used for each unique color value in the image.

You could also split the RGB image into separate bands, apply your different coefficient formulas to each separate band as a function then re-combine the bands into a new image:

r, g, b = img.split()

g = g.point(lambda i: i * (1 - (coef / 2)))

if coef <= 0:

    r = r.point(lambda i: i * (1 + coef))

else:

    b = b.point(lambda i: i * (1 - coef))

img = Image.merge('RGB', (r, g, b))

When creating an image filter, you'd really want to pass in the input image as a function argument rather than use globals. Pass in the coefficient as well:

def filtreTC(source_image, coef):

    coefG = [i * (1 - (coef / 2)) for i in range(256)]

    if coef <= 0:

        coefR = [i * (1 + coef) for i in range(256)]

        coefB = list(range(256))

    else:

        coefR = list(range(256))

        coefB = [i * (1 - coef) for i in range(256)]

    return source_image.point(coefR + coefG + coefB)

You can then store the result of the call in a global if you must, but the function can now stand on its own, and can be re-used anywhere that has a RGB PIL image object. You'd call the function with the slider value:

coef = sliderTC.get() / 100

fenTC.destroy()

img2 = filtreTC(img, coef)

Using the above on your profile image

with a coefficient of 0.75, gives us:

while -0.75 results in: