Color Vision

by Sabina Siddiqi

The idea

Simulate the image that is formed on the retina of the human eye BEFORE it reaches the brain

What does the eye see?

Some facts• "The eyeball is about 1 inch (2.5 cm) in diameter"

• There are 20,000,000 rods and 6,500,000 cones in the eye.

• Out of these 64% detect Red, 32% detect Green, and only 2% detect Blue light.

• Most of the eye has a distribution of both cones and rods except for the fovea centralis, which only contains red and green cones (about 30,000)

• Blue light is detected by cones outside the foveal region and therefore blue objects tend to look more blurry than red or green, creating a dreamy feel to the color blue. Blue cones are also more sensitive than green or red cones.

The fovea centralis

• The fovea covers 0.6% of the retina (fovea ~0.3 mm diameter). i.e. only 0.6% of the image formed on the retina is in focus.

• For reference the image the moon forms on the retina is about 2mm in diameter.

Color Sensitivity

The image on camera

Rethink fovea size on image

• The size being 0.6% of the image works only with 180 degree field of view

• In these circumstances I decided to approximate the diameter of the moon

RGB Channels

Clearly something was wrong…

So I had to fix a bug in my code… (Yes it was a +/- error!

Better results?

Hard to tell!

Try brighter images

Better results?

Decisions

Area of focus: 0.6% vs approximate visual size of the moon

Cones: 3.3% of nerves are red cones – translates to every 30th pixel OR, every pixel has a percentage of it's red value The border between the fovea and retina is not sharp, since light comes from all directions.Have a gradient so the values don’t fall off from perfect focus to suddenly out of focus making it a gradual shift.

Applications

• Possible applications are in perceptual psychology research

• Creating creative and aesthetically pleasing viewpoints in film and photography, and eventually creating realistically "explorable" images which adapt to where your eye is focused.