https://github.com/calcuis/avatar-picture-data-simulator
https://github.com/calcuis/avatar-picture-data-simulator
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/calcuis/avatar-picture-data-simulator
- Owner: calcuis
- Created: 2023-12-14T02:43:15.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-12-23T00:29:24.000Z (over 1 year ago)
- Last Synced: 2025-01-21T13:11:47.138Z (5 months ago)
- Language: Python
- Size: 44.9 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## avatar data simulator
Generate 10,000 samples for ai-picture-model-trainer : P
This Python code defines an image generator using the Pillow library (PIL). The generator creates composite images by sequentially selecting a random image from each layer and combining them.
Here's an explanation of the code:
Layer Class:
- The Layer class represents a layer in the image generation process, associated with a directory containing images.
- `__init__(self, path: str)`: initializes a layer with a given path.
- `get_random_image_path(self)`: returns a randomly selected image path from the layer's directory.
Generator Class:
- The Generator class manages the image generation process.
- `__init__(self, images_path: str)`: initializes the generator with a base path for the layers, sets the background color to None (transparent), defines the output path for generated images as "./data," and creates the output - directory if it doesn't exist. It also loads the layers from the specified images_path during initialization.
- `load_image_layers(self, images_path: str)`: loads and returns a list of Layer objects based on subdirectories in the specified images_path.
- `generate_image_sequence(self)`: generates a sequence of image paths by sequentially selecting a random image from each layer.
- `render_image(self, image_path_sequence: List[str])`: creates a composite image by alpha-compositing the images from the generated sequence onto a blank canvas. The canvas size is 24x24 pixels.
- `save_image(self, image: Image.Image, i: int = 0)`: saves the generated image with a filename based on the iteration index.
- `generate(self, n: int = 1)`: initiates the image generation process. It generates a sequence of images for each iteration, renders the composite image, and saves it. The number of iterations is specified by the parameter n, with a default value of 1.
`generate` Function:
- The generate function is defined outside the classes and serves as a convenient way to create a Generator instance and generate a specified number of images (i.e., 1000 in this case) from the layers located in the "./components" directory.
-
Script Execution:
- The script concludes by calling generate(1000), which triggers the generation of 1k composite images; the number of generation is editable.
The code assumes that each subdirectory in "./components" represents a separate layer, and images within each layer directory are combined to create the final composite image. The generated images are saved in the "./data" directory.
**Reference**
github.com/pixegami/pixel-punk-avatars