Well, I'm not an expert on this stuff. But what I've read indicates that both the source and image have to be embedded in adjacent substrates, one with a positive index of refraction, the other a negative index. This would seem to prevent sub-diffraction imaging of distant celestial objects. Furthermore, this experiment was done with a UV laser line at 365 nm wavelength, rather than white light. I don't know if it is required, but if the light from the source has to be coherent for this experiment to work, then it will never work to image distant objects because even the best lasers only have a coherence length of ~100 meters. This constraint might even set an upper limit on the size of interferometer telescopes. If coherent light is not required, but monochromatic light is, then the image will suffer from poor exposure, since very little light from celestial objects is concentrated at a narrow band of the spectrum. To allow white light would mean that this 'perfect lens' would still suffer from chromatic aberration, since not all wavelengths are refracted at the same angle. That last paragraph in the article still makes the claim however.

- I still don't get it.