ansi-image 0.1.0

A library and tool for displaying images in the terminal using ANSI color codes

ansi-image

A library and tool for displaying images in the terminal, using ANSI color codes.

The pixel-to-ansi algorithm is a straight port from the tiv (TerminalImageViewer) implementation, so this library should produce exactly the same images.

The main difference compared to most other terminal image viewers is that this project is mainly designed as a library, instead of a command-line application.

For displaying the images, your terminal must support ANSI, 24 bit true color and unicode rendering. Unless you're a retro computing enthusiast, your current terminal does support these.

Usage

Basic Usage

from ansi_image import AnsiImage

image = AnsiImage.from_file("tests/test_image.png")
print(image)

Size Control

The output width and height paramers can be used to specify the maximum width and height of the rendered image. The image will be rendered with the largest possible size that fits within the given bounding box and keeps the original image's aspect ratio.

To ensure an image with exact dimension, use the 'fill' keyword option to add add a background fill extending the image to the specified size.

Note that the width and height is measured in termianl characters, which are not square but rectangular.

from ansi_image import AnsiImage

image = AnsiImage.from_file("tests/test_image.png")

# Use the current terminal size as default width and height
print(image)
input("press enter...")

# Set an explicit max width and height
render = image.render(max_width=80, max_height=24)
print(render)
input("press enter...")

# Using format strings
print(f"{image:w=40,h=20}")
print(f"{image:width=60}")
input("press enter...")

# Add background color to fill the entire bounding box
rendered = image.render(fill="#ffffff")
print(rendered)
print(f"rendered image with dimensions f{rendered.width}x{rendered.height}")

Image Stretching

For precise control over image dimensions without maintaining aspect ratio, you can manipulate the image using PIL before rendering:

from ansi_image import AnsiImage
from PIL import Image

# Load and stretch the image to exact dimensions
img = Image.open("tests/test_image.png")
stretched_img = img.resize((160, 48))  # Stretch to exact dimensions
ansi_stretched = AnsiImage.from_image(stretched_img)
print(ansi_stretched)

Command Line Tool

The package also includes a command-line tool:

uvx ansi-image tests/test_image.png
uvx ansi-image --width 80 --height 24 tests/test_image.png
# From the source repository
uv run ansi-image tests/test_image.png

API Reference

AnsiImage

Main class that stores the original PIL Image and provides rendering methods.

• img = render(max_width=None, max_height=None, flags=0, fill=None) - Render to RenderedAnsiImage
• AnsiImage.from_image(img, ...) - Static method to create directly from PIL Image
• AnsiImage.from_file(path, ...) - Static method to load and create from file
• str(img) - Convert to printable string

RenderedAnsiImage

Contains the pre-rendered text representation that can be printed.

• str(rendered) - Convert to printable string
• rendered.width - Width in terminal columns
• rendered.height - Height in terminal rows
• rendered.data - List of strings with ANSI codes, one per row.

Memory Usage

The AnsiImage object keeps the full PIL Image in memory, allowing multiple renders with different parameters.

The RenderedAnsiImage objects only contain the text representation. Use the render() method to obtain the latter and discard the former if memory usage is a concern.

Algorithm

The rendering algorithm is a direct port from the C++ implementation in TerminalImageViewer, providing the same high-quality terminal image display in pure Python.

On a high-level, it works by splitting the image into 4x8 pixel blocks and selecting the most appropriate unicode block character for each, with the closest matching foreground and background colors.

Installation

Install via pypi:

pip install ansi-image

Or use as a standalone tool:

uvx ansi-image tests/test_image.png