Add new features

Author: OmarSamirz

README.md

@@ -119,6 +119,14 @@ IFTG offers a wide variety of noise effects that you can apply to your images to
   </tr>
 </table>
 
+<table>
+  <tr>
+    <th>Pixelate</th>
+  </tr>
+  <tr>
+    <td><img src="https://drive.google.com/uc?export=view&id=1r3_sA2A4HM2ILnoxkw1TFrQYYeuIjsk-" alt="Pixel Dropout" width="100%"></td>
+</table>
+
 ## Installation
 To get started with IFTG, you'll need to install the package. You can do this using pip.
 ```bash

docs/assets/pixelate_img.png

@@ -83,6 +83,15 @@ enhancing the robustness of your models.
   </tr>
 </table>
 
+<table>
+  <tr>
+    <th>Pixelate</th>
+  </tr>
+  <tr>
+    <td><img src="/ImageFromTextGenerator/assets/pixelate_img.png" alt="Pixel Dropout" width="100%"></td>
+  </tr>
+</table>
+
 ## **Quick Start**
 To get started with IFTG, follow these simple steps:
 

docs/introduction.md

@@ -17,6 +17,7 @@ def _create_base_image(cls,
                            text: str,
                            font: ImageFont,
                            font_color: tuple[int, int, int],
+                           font_opacity: float,
                            background_color: str,
                            margins: tuple[int, int, int, int],
                            background_img: Image

iftg/creators/creator.py

@@ -20,12 +20,13 @@ def _create_base_image(cls,
                            text: str,
                            font: ImageFont,
                            font_color: tuple[int, int, int],
+                           font_opacity: float,
                            background_color: str,
                            margins: tuple[int, int, int, int],
                            background_img: Image
                            ) -> Image.Image:
         """
-        Creates a base image with the specified background color and dimensions, 
+        Creates a base image with the specified text, background color and dimensions, 
         and optionally adds a background image.
 
         Parameters:
@@ -51,13 +52,18 @@ def _create_base_image(cls,
         image_width, image_height = cls.get_image_dimensions(
             margins, text_dimensions)
 
-        image = Image.new('RGB',
-                          (image_width, image_height+text_dimensions[1]),
-                          color=background_color
-                          )
+        base_img = Image.new('RGBA',
+                             (image_width, image_height+text_dimensions[1]),
+                             color=background_color
+                             )
+        text_layer = Image.new('RGBA', 
+                               (image_width, image_height+text_dimensions[1]),
+                               color=(255, 255, 255, 0)
+                               )
 
         # add a background image to the text
         if background_img != None:
+            background_img = background_img.convert("RGBA")
             bg_width, bg_height = background_img.size
 
             x1 = np.random.randint(0, bg_width - image_width)
@@ -67,19 +73,23 @@ def _create_base_image(cls,
 
             random_bg_part = background_img.crop((x1, y1, x2, y2))
 
-            image.paste(random_bg_part)
+            base_img.paste(random_bg_part)
 
         # Draw the text on the image
-        draw = ImageDraw.Draw(image)
+        opacity = int(font_opacity * 255)
+        draw = ImageDraw.Draw(text_layer)
         draw.text((margins[0], -text_dimensions[1]+margins[1]),
-                  text, font=font, fill=font_color)
-
-        return image
+                  text, font=font, 
+                  fill=(*font_color, opacity)
+                  )
+        final_img = Image.alpha_composite(base_img, text_layer)
+        
+        return final_img.convert('RGB')
 
     @classmethod
     def _apply_noise(cls, noises: list[Noise], image: Image) -> Image:
         """
-        Applies text, and noise effects to the base image.
+        Applies noise effects to the base image.
 
         Parameters:
             noises (list[Noise]):
@@ -95,29 +105,6 @@ def _apply_noise(cls, noises: list[Noise], image: Image) -> Image:
 
         return image
 
-    @classmethod
-    def _blend_colors(cls, bg_color: str, text_color: str, font_opacity: float) -> tuple[int, int, int]:
-        """
-        Blends the text color with the background color to simulate transparency.
-
-        Parameters:
-            bg_color (str): The background color in any valid PIL color format.
-            text_color (str): The text color in any valid PIL color format.
-            alpha (float): The transparency level (0.0 to 1.0).
-
-        Returns:
-            tuple: The blended color as an (R, G, B) tuple.
-        """
-
-        bg_r, bg_g, bg_b = ImageColor.getrgb(bg_color)
-        text_r, text_g, text_b = ImageColor.getrgb(text_color)
-
-        r = int((1 - font_opacity) * bg_r + font_opacity * text_r)
-        g = int((1 - font_opacity) * bg_g + font_opacity * text_g)
-        b = int((1 - font_opacity) * bg_b + font_opacity * text_b)
-
-        return r, g, b
-
     @classmethod
     def create_image(cls,
                      text: str,
@@ -165,9 +152,9 @@ def create_image(cls,
         """
         font = ImageFontManager.get_font(font_path, font_size)
 
-        r, g, b = cls._blend_colors(background_color, font_color, font_opacity)
+        font_rgb = ImageColor.getrgb(font_color)
         image = cls._create_base_image(
-            text, font, (r, g, b), background_color, margins, background_img)
+            text, font, font_rgb, font_opacity, background_color, margins, background_img)
 
         image = cls._apply_noise(noises, image)
         image.info['dpi'] = dpi

iftg/creators/image_creator.py

@@ -20,6 +20,7 @@
 def main0():
     image = ImageCreator.create_image(
         'Hello, world', './fonts/Arial.ttf', font_opacity=0.3)
+    print(image.mode)
     image.save('opacity_img.png', **image.info)
 
 
@@ -96,8 +97,9 @@ def main4():
 
     texts = ['Hello, World!', 'how are you', 'what are you doing'] * 10
     results = ImagesGenerator(
-        texts=texts, font_path='fonts/Arial.ttf', font_opacity=0.7,
-        noises=[RandomPixelateNoise()]
+        texts=texts, font_path='fonts/Arial.ttf', font_opacity=1.0,
+        noises=[RandomPixelateNoise()],
+        img_format='.png'
     )
     results.generate_images_with_text()
 
@@ -112,4 +114,4 @@ def main5():
 
 
 if __name__ == '__main__':
-    main4()
+    main0()

main_test.py

@@ -4,7 +4,7 @@
 
 setuptools.setup(
     name='iftg',
-    version='1.2.7',
+    version='1.2.8',
     description='IFTG (ImageFromTextGenerator) is a Python package that simplifies creating robust datasets for OCR models. Generate images from text, apply over 10 built-in noise effects, and customize fonts and layouts. IFTG supports all languages and offers endless noise combinations, including custom noise creation.',
     long_description=pathlib.Path('README.md').read_text(),
     long_description_content_type='text/markdown',

setup.py

@@ -12,42 +12,30 @@ def mock_font():
     font = ImageFontManager.get_font('tests/Arial.ttf', 12)
     return font
 
-
 @pytest.fixture
 def mock_image():
-    image = MagicMock(spec=Image.Image)
-    # Add required attributes and methods for ImageDraw
-    image.size = (500, 500)
-    image.readonly = False
-    image.getdraw = MagicMock()
-    # Mock the drawing context
-    draw_context = MagicMock()
-    image.getdraw.return_value = draw_context
-    return image
-
+    # Create a real PIL Image with RGBA mode to support more operations
+    return Image.new("RGBA", (500, 500), color=(255, 255, 255))
 
 @pytest.fixture
 def noise_list():
     return [BlurNoise(blur_radius=2.0), BlurNoise(blur_radius=5.0)]
 
-
 @pytest.mark.parametrize(
     "text, margins, bg_color, font_color, expected_size",
     [
-        ("Sample Text", (5, 5, 5, 5), "white", (255, 255, 255), (500, 500)),
-        ("Another Text", (10, 10, 10, 10), "black", (255, 255, 255), (500, 500)),
+        ("Sample Text", (5, 5, 5, 5), (255, 255, 255), (0, 0, 0), (500, 500)),
+        ("Another Text", (10, 10, 10, 10), (0, 0, 0), (255, 255, 255), (500, 500)),
     ]
 )
-def test_create_base_image(mock_font, mock_image, text, margins, bg_color, font_color, expected_size):
-    with patch('PIL.Image.new', return_value=mock_image):
+def test_create_base_image(mock_font, text, margins, bg_color, font_color, expected_size):
+    # Create a patch that returns a real RGBA image
+    with patch('PIL.Image.new', return_value=Image.new("RGBA", expected_size, color=bg_color)):
         image = ImageCreator._create_base_image(
-            text, mock_font, font_color, bg_color, margins, None)
-
-        assert image == mock_image
+            text, mock_font, font_color, 1.0, bg_color, margins, None)
+        
         assert image.size == expected_size
-        # Changed from Image.Image since we're using a mock
-        assert isinstance(image, MagicMock)
-
+        assert isinstance(image, Image.Image)
 
 def test_invalid_font_path():
     with patch('iftg.image_font_manager.ImageFontManager.get_font', side_effect=FileNotFoundError):