[schoen]

I don't think so.

The total data on a CD-ROM including the error correcting redundancy exceeds 800 MB, or 6.4 Gb.

If you could imagine getting 1 bit from the optical disc per pixel (which is way too optimistic physically), you would need a 6 gigapixel camera focused super-precisely at the disc surface.

Looking at the problem from a different angle, Wikipedia says the features that store the data on the disc surface are about 800 nm (or about a micrometer) long. So to photograph them, you'd want to have pixels ideally smaller than a micrometer on each side. It's easy to check that an ordinary camera isn't achieving that kind of resolution without adding on external magnifying equipment.

[jackweirdy]

Does it have to be one photo? If you reproduced a spinning drive but with the camera positioned to see half of the spinning disc, I wonder if it could capture the "stream" of pixels in one arc of the spinning disc

[schoen]

You would still need some significant magnification. And there might also be a measurement latency issue if the disc is in motion (the camera CCD might not be fast enough to capture the image before it rotates away).

The optics of a CD-ROM drive are optimized for something pretty different than the optics of a camera. But if you made enough tweaks and adaptations, sure, the data is ultimately there and can be captured by a different kind of sensor than the one it was designed for. It would be a cool project.

I'm mostly just pointing out that adapting your camera to successfully capture billions of sub-micrometer features isn't that trivial.