funkycam 0.0.6

A small package to make funky video, images, and real-time video

FunkyCam

This is a simple package with one class, Funk, that allows you to recolor images by "luminance" of each pixel and also extract and thicken edges. The final product is a cartoon-like version of the original image or video. Like this:

and this:

To install this package just use:

pip install funkycam

Examples are given in the "examples" folder to demonstrate how to use this package with a webcam, image files, and video files. But, to use this class all you need to do is:

import cv2
from funkycam.funkycam import Funk

img = cv2.imread('example.jpg')

funk = Funk()
funky_img = funk.funkify(img)

Note: The code expects images that are read using cv2 which means they are in BGR format. If you read with other packages, images are often in RGB format. It may still work fine, but the calculations will be incorrect. Thus, the channels will need to be flipped. This can be accomplished on an np array with array slicing:

img_flipped = img[:,:,::-1] # flips third dim

Customizing the Output

To select colors, there are 3 options:

1. Use the seaborn color_palettes or matplotlib colormaps (https://matplotlib.org/stable/tutorials/colors/colormaps.html).
2. Manually input RGB colors in a list of lists. Resolution is 8 bits: 0-255 each color.
3. Specify "random_colors=True".

If you can't find colors you like, I recommend randomly generating colors and recording the colors that look good. The colors are printed every time it runs, but you can also access them from the Funk object with "funk.colors".

You can also specify the block size and blur value for finding edges, and blur value when finding colors. Increasing the block size seems to increase the thickness of the edges. Increasing blur values will reduce detail in the edges and colors, i.e. fewer edges and larger splotches of colors. These values depend a lot on the resolution of the image or video. NOTE: The block size and blur values must be odd numbers.

Other Notes

The luminance calculation is not real luminance, it is simplified so that it can recolor video in "real-time" (depends on image quality and number of colors chosen).

Turns out this method is a little bit like "cel-shading", a commonly used shader algorithm.