[open] Pixelsize in Gabor patch generator

Isabel

Hey!
I am using the online Gabor patch generator and I was wondering how to determine the size of the stimulus.
I know that the Gabor patch for my project needs to have a radius of 2° and I also determined the pixels per degree for our setting but I am not sure whether I need to convert the radius into pixels or whether I should use the diameter then (so 4°)? Or does the size asked for in the generator refer to the whole square and not just the Gabor patch within it?
Also, I am trying to use different contrast levels. I am not too familiar with this and I was wondering whether, if I want to have a background with the mean luminance, I would start with the RGB values for both colours at 128 as the lowest possible contrast and then stepwise increase the difference (e.g. colour 1: 129 and colour 2: 127) or whether the lowest possible contrast would be 127 to 128 and then increase/decrease (e.g. next contrast would be at 126 and 129). I used the latter at first but I wasn't too sure about it (as it is only possible to use integers for the generator, but then the average wouldn't be an integer) but then if I use the first approach the background doesn't really seem to look right.
I would be really grateful, if someone could help me out with these questions!

sebastiaan

Hi Isabel,

In theory, the size of a Gabor patch is infinite: The contrast is high in the center and becomes progressively lower with increasing eccentricity--but never exactly zero. So you would typically specify the size of a Gabor patch by the standard deviation of the envelope, and not by a radius. The 'size' option in the Gabor-patch generator is only the size of the generated image, which should be sufficiently large so that there is no visible cut off. Does that clear things up? I'm not sure if this answers your question though.

Regarding the contrast. Using (127,127,127) for color 1 and (128,128,128) will indeed give you the lowest possible contrast. But the background color will be rounded up to 128, because the average (127.5) is not an integer and can therefore not be shown by the monitor. And for the same reason, the luminance modulation will not be a smooth sinusoid, but a square wave (i.e. a single step from one luminance to the next). Basically, with these extremely low contrast levels, you run into the physical limitations of the display. But since you usually work with far higher contrast levels, these problems are mostly theoretical. Is it important for your experiment to have high-quality extremely low contrast patches?

Cheers!
Sebastiaan

Isabel

Hey Sebastiaan,

I think this answers my questions, thank you very much for your help!
For the contrasts, though, would it work to use a brighter background (if I use the average of the two colours), if I want to have smaller contrasts? So, for example using 244 for colour 1 and 246 for colour 2 etc.?

sebastiaan

For the contrasts, though, would it work to use a brighter background (if I use the average of the two colours), if I want to have smaller contrasts? So, for example using 244 for colour 1 and 246 for colour 2 etc.?

In principle, I would say that the contrast is defined as the difference in brightness between the two colors, and is unrelated to the overall brightness. So 10 and 12 would have the same contrast as 100 and 102. However, I suppose that this is open to debate.

But at any rate, you cannot use values over 255, because this is outside the range of the monitor. I just noticed that the Gabor-patch will not warn you about this (except in the description), but choosing values over 255 will cause odd distortions!

Cheers!
Sebastiaan