Coder - show images

Sometimes we want to use PsychoPy to show images — either read from an image file (such as .png) or from a Numpy array. We can use either the psychopy.visual.ImageStim or psychopy.visual.GratingStim to achieve this. However, some of the nuances of actually getting the correct image to screen can be difficult to figure out.

This recipe demonstrates (1) a way to use psychopy.visual.ImageStim to read an image from disc and show it, and (2) using psychopy.visual.ImageStim to show a numpy array as an image.

When showing and converting images, you need to be careful about data types and channels. The scikit-image docs on this are quite good.

from pathlib import Path

import numpy as np
from PIL import Image
from psychopy import core, event, visual

# ---------------------
# Setup window
# ---------------------
win = visual.Window(
    (900, 900),
    screen=0,
    units="pix",
    allowGUI=True,
    fullscr=False,
)
# ---------------------
# Example 1: load a stimulus from disk
# ---------------------

# assume we're running from the root psychopy repo.
# you could replace with path to any image:
path_to_image_file = Path() / "PsychoPy2_screenshot.png"

# simply pass the image path to ImageStim to load and display:
image_stim = visual.ImageStim(win, image=path_to_image_file)
text_stim = visual.TextStim(
    win,
    text="Showing image from file",
    pos=(0.0, 0.8),
    units="norm",
    height=0.05,
    wrapWidth=0.8,
)

image_stim.draw()
text_stim.draw()
win.flip()
event.waitKeys()  # press space to continue

# ---------------------
# Example 2: convert image to numpy array
#
# Perhaps you want to convert an image to numpy, do some things to it,
# and then display. Here I use the Python Imaging Library for image loading,
# and a conversion function from skimage. PsychoPy has an internal
# "image2array" function but this only handles single-layer (i.e. intensity) images.
#
# ---------------------

pil_image = Image.open(path_to_image_file)
image_np = np.array(
    pil_image, order="C"
)  # convert to numpy array with shape width, height, channels
image_np = (
    image_np.astype(np.float) / 255.0
)  # convert to float in 0--1 range, assuming image is 8-bit uint.

# Note this float conversion is "quick and dirty" and will not
# fix potential out-of-range problems if you're going
# straight from a numpy array. See the img_as_float
# function of scikit image for a more careful conversion.

# flip image (row-axis upside down so we need to reverse it):
image_np = np.flip(image_np, axis=0)
image_stim = visual.ImageStim(
    win,
    image=image_np,
    units="pix",
    size=(
        image_np.shape[1],
        image_np.shape[0],
    ),  # here's a gotcha: need to pass the size (x, y) explicitly.
    colorSpace="rgb1",  # img_as_float converts to 0:1 range, whereas PsychoPy defaults to -1:1.
)
text_stim.text = "Showing image from numpy array"
image_stim.draw()
text_stim.draw()
win.flip()
event.waitKeys()  # press space to continue

win.close()
core.quit()

Back to top