How do the Zeiss Batis 18mm f/2.8 and Tokina Firin 20mm f/2.0 AF compare for astrophotography?

Question

I'm trying to choose a lens for astrophotography with Sony A7. Although I choose it for astro, I also want it to function as a normal landscape lens. So I try not to go far wide like Samyang 14mm f/2.8. My choice is narrowed for two lenses:

1. Zeiss Batis 18mm f/2.8 (Used one)
2. Tokina Firin 20mm f/2.0 AF (New one)

Though the first one is far much costly compared to the second one, the used Batis is easy to get but Tokina is not in my country. Considering the noise reduction capacity of Sony A7(M1), is f/2.0 critical to get Tokina? Or just I get Batis with some more bucks?

Answer 1

You probably don't want to use noise reduction in astrophotography as the algorithms may consider stars as noise!

Typically, when choosing a lens for astrophotography, you want short focal length and large aperture area at the same time. You can calculate an "astrophotography index" for the lenses, and the index should be as low as possible. The index is:

I = f * N²

where f is the focal length and N is the aperture number. Example: the index is 141.12 for Zeiss and 80 for Tokina. Thus, you should get the Tokina as it collects 76.5% more light, so you can use 0.57 times the ISO you use for Zeiss. Lower ISO, less noise.

BTW, according to this index, the Samyang 14mm f/2.8 is worse than Tokina 20mm f/2.0.

More information: https://www.lonelyspeck.com/lenses-for-milky-way-photography/ -- do note, however, that their "rating" (which should be high, not low), is calculated using a slightly different formula so the numeric values aren't comparable.

Justification for the index formula: aperture opening in length units squared (proportional to f²/N²) and field of view in angle units squared (proportional to 1/f²) dictates how much you can collect light. However, max exposure time is proportional to 1/f because you don't want star trails, so the light collecting ability is proportional to 1/(f * N²), so you'll want as high as possible 1/(f * N²) or in other words as low as possible f * N².