The size and scale of all five million cones of the human eye are hard vizulize. To help some, below is a link to the file "retinalCones.jpg", an 8Kx8K jpeg image of the central 3.4 degrees of one of my synthetic retinas. Because the image is 8K by 8K pixels in area, and 11.9 megabytes in size, it will open slowly, and on your system may be more easily viewed in a different application than your browser's default image viewer. While both Microsoft Explorer and Netscape browsers will directly display the image, you may prefer to use an image editor (such as Adobe PhotoShop) to have more control over the detail of viewing.
If you open the image with Microsoft Explorer, it may display by default with the "Windows Picture and Fax Viewer". Here the magnifying glass buttons will zoom in and out, and once zoomed in the image can be panned by using the scroll bars on the edges of the window. Because the image is large you may have to wait a few seconds for some of the zoom or pan actions to take effect. The "actual size button" (Ctrl+A) will zoom all the way into the actual pixels of the image.
If you open the image with Netscape, it may display by default directly inside the browser. Here a left mouse click will zoom all the way into the actual pixels of the image, which can then be panned about using the scroll bars on the edges of the window. Additional mouse clicks will just toggle between no zoom and full zoom.
The 5.1 million cone data set is really big, and it is hard to get across the size and variations in the paper itself (the 10 page continued paper Figure 1 was our best try). We rendered the first 3.4 degrees of visual eccentricity (~120 thousand cones, ~2% of our data) into a 8K by 8K jpeg image; the cones are about the same pixel size as in paper Figure 1. Zooming and panning around this "retina.jpg" image can give a good feel for the synthesized retina data set.
The center most cone is covered by a red dot. The dashed black circles mark off unit degrees of visual eccentricity; the length of each dash is set to 10 microns for scale. The cones were colored using the same coloring scheme as Figures 1 and 6 in the paper: L cones are white, M cones are slightly darker and very slightly greenish, and S cones are even darker and slightly bluish.
The dark gaps between the cones at low eccentricities are just the cone boarders, but after the first 0.5 degrees the black areas also start to represent where the rods would be if we were placing them too. The change in the density and the regularity of the packing order can be easily seen by panning through this image, but subtle detail (such as the slightly higher cone density about the horizontal meridian) are harder to verify visually. The way the pattern breaks up the regular hexagonal pattern can be examined; just like real retinas this is usually caused by four cones rather than three meeting at a single point. Breaks are usually cause by several such four-way meetings lining up.
The retina is spherical, but this image doesn't go out very far, so we just flatten the coordinates (orthographic projection).
The same synthesized retina was used to generate all the figures in the paper as well as this jpg file, and all the video images except for the growth sequence in the video.
retinalCones.jpg (12 MB)