Originally posted by barondla I've always thought star tests were one of the ultimate tests of lenses. Recently I read something here that has me confused. The earliest Pentax 6x7 lens is only a 4 element design and only decently sharp. It is rated much lower than the newer 200. However, 2 different reviews talk about how amazing the 4 element is for shooting the stars. One even discusses how the lens doesn't clip a color channel like most do. So here it seems a so so lens does well for astrophotography. I wouldn't have thought that possible.
Have you used this lens? Have you encountered the color channel clipping with other lenses?
Here's the link:
S-M-C Takumar 6x7 / Super Takumar 6x7 200mm F4 Reviews - 67 Telephoto Primes - Pentax Lens Reviews & Lens Database.
Thanks,
barondla
This issue shows the subtlety of what makes a great lens, especially great one for astrophotography.
When we look at "good" astro photographs, the stars look to be of different sizes. Nearby bright stars appear to be larger balls and only the distant faint stars appear as tiny points. That's very intuitive to us because our own human eyes tend to see bright stars as "bigger" then faint stars.
In theory, however, the ultimate ultra-sharp lens would render all stars as almost infinitesimal points of light. With the perfect lens, even a physically large and nearby red giant star such as Betelgeuse (which is 764X the diameter of the sun and only 548 light years away) would be a pinprick only 0.2 microradians (1/25th arcseconds) across which is a very tiny fraction of a pixel on a 645Z with a 200mm lens. A perfectly sharp lens would concentrate all the light from a bright star such as Arcturus into a single pixel. For most color cameras, that would either be either a red, green, or blue pixel of a Bayer filter. A bright star would saturate that one color pixel and render the star as a single RGB dot regardless of the star's original color. If the camera has a low-pass filter designed to prevent these kinds of Bayer pattern artifacts, the star be spread over about 2x2 pixels and be white.
In real lenses, diffraction blurs the stars (as does atmospheric turbulence) but a diffraction-limited f/4 lens would still be too sharp to properly render the true colors of bright, nearby stars. With a perfect 200 mm f/4 lens, bright stars such as Betelgeuse would still render as white points a couple of pixels across.
What makes the best astro lenses are lenses that are pretty sharp but not too sharp. More importantly, they have perfectly circular point spread functions (PSF) that have a nice central peak (essential for rendering faint stars as a rich field of points rather than invisible blobs) and broad enough tails on the PSF to smear the bright stars into color-preserving balls. Most importantly, the best lenses have a uniform PSF across the field so that all stars have the identical shapes and sharpnesses.
The importance of a circular and uniform PSF carries over into other types of photography. Non-circular aberrations and non-uniform aberrations create noticeable center-to-corner variations in the appearance of the image. With non-circular aberrations and non-uniform aberrations, textures might be sharp in the center, fuzzy on some edges, and have strange smearing that differs across the image.
TL;DNR: Great lenses don't need to be perfectly sharp but they do need to be very uniform and have blur that are very circular.