0

How do I get a list of all the Indexed Color mode PNGs that are present in a given folder that contains a lot of other files (which are all images, but they all have different Color Modes) using Python?

Previously tried code:

from PIL import Image
import os

path = 'logos/'

for x in os.listdir (path):
    if x.endswith(".png"):
        img = Image.open(path + x)
    cmode = str(img)
    P = 'mode=P' in cmode
    if P == True:
        print (x + " " + str(img))

Using this code, I got a list of images, some of which are Indexed Color mode, and the rest are RGB color mode (checking them through Photoshop) https://www.dropbox.com/s/vlvywqhcfrkk8kq/3978.png?dl=0 This is a link to an image that shows up as P through the script, but it is an RGB image in Photoshop. https://www.dropbox.com/s/x3qiuuhs3gv9bp9/6507.png?dl=0 This is a truly Indexed Color image, like the ones that I need to find.

Francesco Calderone
  • 35
  • 5
  • 1
    Maybe this will help: [Read in an indexed color image in Python](https://stackoverflow.com/questions/33022983/read-in-an-indexed-color-image-in-python) :) – Ori Jul 02 '22 at 08:30
  • Does this answer your question? [Read in an indexed color image in Python](https://stackoverflow.com/questions/33022983/read-in-an-indexed-color-image-in-python) – Shunya Jul 02 '22 at 13:38

1 Answers1

1

You can use this:

#!/usr/bin/env python3

from PIL import Image
from glob import glob

# Start with empty list
paletteImages = []

# Iterate over all PNGs
for f in glob("*.png"):
  # Open with PIL
  im = Image.open(f)
  # Append to list if palette image
  if 'P' in im.mode: paletteImages.append(f)

print(paletteImages)

More discussion on palette/indexed images here.

Note that the above code will also find PaletteAlpha images, i.e. those with mode = 'PA', so change in to == if you don't want PaletteAlpha images.

Or you can do it with ImageMagick in the Terminal more simply:

identify -verbose  *png | egrep "Filename:|Type|png.IHDR.color_type"

Filename: 3978.png
Type: PaletteAlpha
png:IHDR.color_type: 3 (Indexed)
Filename: 6507.png
Type: Palette
png:IHDR.color_type: 3 (Indexed)

Note that you can get exactly the same results from wand which is a ctypes binding to ImageMagick.

Or you can use pngcheck:

pngcheck *png   
                           
OK: 3978.png (64x64, 8-bit palette+trns, non-interlaced, 33.1%).
OK: 6507.png (64x64, 8-bit palette, non-interlaced, 56.3%).

You could call this with a Python subprocess.run().

Or you can use exiftool:

exiftool *png | egrep "Color Type|File Name"   
File Name                       : 3978.png
Color Type                      : Palette
File Name                       : 6507.png
Color Type                      : Palette

You can get the same results with the Python binding to exiftool. Example here.

Or you can search for the PLTE, i.e. palette PNG chunk:

grep PLTE *png     

Binary file 3978.png matches
Binary file 6507.png matches

Or you can slurp the image with Python and search for string in the slurped file to get the same result:

with open('3978.png', 'rb') as fd:
   data = fd.read()
   if b'PLTE' in data:
      print('PLTE was found')
Mark Setchell
  • 191,897
  • 31
  • 273
  • 432
  • I had already tried the first method but mode = P returns a list of images that, when opened in photoshop, are either Indexed Colour mode or RGB mode. Finding those files is a function that I need add to a Python script, which is why I haven't used ImageMagick. – Francesco Calderone Jul 02 '22 at 10:10
  • If you already tried that method, you should have posted your code and explained it didn't work. If you have an image that PIL reports as indexed but you believe is not actually indexed, please share it using some service (other than imgur) that doesn't alter your images - e.g. Google Drive, Dropbox or some such. Thank you. – Mark Setchell Jul 02 '22 at 10:14
  • Please don't put code in comments - it is unformatted and difficult to read. Instead, click `edit` under your original question. Thank you. – Mark Setchell Jul 02 '22 at 10:14
  • Thanks Mark, I've updated the post with both code and link to sample image – Francesco Calderone Jul 02 '22 at 10:27