psychopy.org | Reference | Downloads | Github

Colour Transforms Displaying Wrong

Hi Everyone,

This is a weird issue and I can’t make head or tail or it, so if anyone has even an incomplete suggestion as to what on earth is going on I would be very very grateful!

The issue: Once an rgb array has been converted into dkl space and then back again it no longer displays correctly. It displays nine smaller versions of itself in a 3x3 shape.

I’ve checked the conversion code and a whole bunch of other stuff and in the end established that according to numpy.array_equal the array is the same when it goes in as when it comes out. However, it clearly isn’t the same because visual.GratingStim does not display the same thing. I haven’t got the first idea what is different about the two arrays because as I say they are - as far as I can tell identical. To make it even weirder this didn’t used to be an issue in earlier iterations of PsychoPy. As far as I can tell it seems to have cropped up somewhere between 1.77.01 and 1.81.00. I’ve looked at the changelog and I can’t see anything that looks obviously related, but honestly I’m not even sure what I would be looking for. See below for a minimal code example - you can see that testGrating displays correctly whereas rgbTestGrating does not.

So - when are two identical variables not identical?

Thanks so much,
Becky

from psychopy import visual, misc, event
import numpy as np
import copy

myWin = visual.Window((800,600), units = 'pix', monitor = 'testMonitor', fullscr = True)

temp = np.ones([512,512])
temp[:255] *= -1

testGrating = np.empty((512,512,3))
testGrating[:,:,0] = copy.copy(temp)
testGrating[:,:,1] = copy.copy(temp)
testGrating[:,:,2] = copy.copy(temp)

dklTestGrating = rgb2dklCart(testGrating, conversionMatrix = None)

rgbTestGrating = misc.dklCart2rgb(dklTestGrating[:,:,0], dklTestGrating[:,:,1], dklTestGrating[:,:,2], conversionMatrix = None)

print np.array_equal(testGrating, rgbTestGrating)

patch = visual.GratingStim(myWin, tex = rgbTestGrating, size = (512,512))

patch.draw()
myWin.flip()
event.waitKeys()

I think the GratingStim is in units of ‘pix’ and the ‘sf’ is not set appropriately. It’s odd that the matrix is the same but the results are different.

I’ve come across similar issues in the past and have found that using np.ascontiguousarray usually fixes it (e.g. tex=np.ascontiguousarray(rgbTestGrating)).

1 Like

I have no idea why that should work, but it does.

Thanks a million!