How animals see: dog, cat and bird vision camera

• Pick an animal. Choose a group (pets, birds, farm, wild, reptiles and fish), then an animal. The camera view changes at once. “You” is always there to compare with.
• Tap a sample at the bottom to see a ready-made scene through the same eyes. The Flower, Budgie and Fruit samples carry real ultraviolet patterns, which a phone camera can’t record. The Night sample shows dim light and body heat.
• Compare (the ‹|› button) splits the picture: your view on the left, the animal’s on the right. Drag the line.
• Full screen turns the page into a camera app; Save downloads the picture.

Cone peaks are the wavelengths each cone type responds to best, measured in each species’ own retina. Sharpness is converted to the clinic eye-chart scale, where 20/20 is normal human vision.

Take the Toys sample: a red ball lying on green grass. Here is how strongly each cone type responds, with white paper counting as 100.

You see the ball at once because of your red and green cones. The ball gives your red cone 3.0 times the signal of your green cone; the grass gives them about the same (26 and 26). That difference is a strong red–green signal.

A dog has only one cone in that part of the spectrum, so the red–green signal does not exist for it. What is left is brightness and a weaker blue–yellow signal: to a dog the ball is about 46% as bright as the grass and a little less yellow, a dull olive-grey on a yellow-olive lawn. The blue ball is different: it gives the dog’s blue cone 66, against 3 for the grass, while its other cone reads 22 against 24. It shows up as clear blue, which is why a blue toy is easier for a dog to spot on grass than a red one.

1. Light. Each pixel is treated as a surface lit by standard daylight (CIE D65). Its three camera values are turned into the smoothest reflectance curve from 380 to 730 nm that gives exactly those values (Burns’ least-slope method).
2. Cones. Each animal’s cones are modelled with the Govardovskii (2000) pigment template at the peak wavelengths measured for that species, filtered by its measured lens and, in birds and lizards, by the coloured oil droplets in its cones. Multiplying light × reflectance × cone sensitivity, wavelength by wavelength, gives each cone’s signal.
3. Two cone types (dogs, cats, horses, cattle, rabbits, rodents, snakes): the picture is redrawn using only blues, yellows and greys, choosing for every pixel the colour that gives the animal’s two cones exactly the same signals as the real scene. The animal could not tell the two pictures apart; you see what it can and can’t distinguish.
4. One cone type (the Syrian hamster): only brightness is left, so the picture is grey.
5. Four cone types (birds, goldfish, turtles, lizards): their three visible cones cover roughly the range yours do, and a camera records only three channels, so visible colours are kept. What they add is ultraviolet. In the samples, where a surface truly reflects UV, magenta is laid over it, the convention used by Vasas and colleagues (2024) for bird-view video. It is false colour, because no screen can emit UV.
6. A check against real animals. From the cone data alone, the model predicts the colour that looks grey to a two-cone animal. For dogs it gives 482 nm; trained dogs confused about 480 nm with grey (Neitz, Geist and Jacobs 1989). For guinea pigs it gives 475 nm; guinea pigs confused about 480 nm with white (Jacobs and Deegan 1994). Nothing in the model was tuned to get those matches.
7. Sharpness. The picture is blurred by the animal’s measured visual acuity, using the AcuityView method (Caves & Johnsen 2018) and the angle the scene covers. On a phone screen a dog’s view looks only slightly soft, which is accurate; the Eye test sample makes the difference obvious.

• What a colour feels like. Nobody can know what yellow is like to a dog. The simulation shows which colours an animal can tell apart, not its experience.
• Ultraviolet through your camera. Phone cameras have a filter that blocks UV, so the live view can’t show it. That is why the samples carry their own UV layer.
• Heat. A snake’s pit organs sense infrared heat, a separate sense from its eyes. Cameras can’t record it, so heat appears only in the Night sample.
• The whole field of view. A horse or rabbit sees almost all the way round. A screen shows one window; the info panel (ⓘ) draws the full field.
• Motion and flicker. Dogs and birds can see faster flicker than you. Your screen can’t show what you can’t see.
• The brain. Animals learn and attend differently. A dog does not mistake its red ball for grass once it smells it.